Contributions to High Energy Γ-Ray Astronomy Jean-Philippe Lenain

Contributions to high energy γ-ray astronomy Jean-Philippe Lenain

To cite this version:

Jean-Philippe Lenain. Contributions to high energy γ-ray astronomy: Active galactic nuclei and leptonic cosmic rays. From H.E.S.S. to CTA. High Energy Astrophysical Phenomena [astro-ph.HE]. Sorbonne Université UPMC, 2018. tel-01740556

HAL Id: tel-01740556 https://tel.archives-ouvertes.fr/tel-01740556 Submitted on 22 Mar 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Contributions to high energy γ-ray astronomy Active galactic nuclei and leptonic cosmic rays From H.E.S.S. to CTA

Jean-Philippe LENAIN Laboratoire de Physique Nucléaire et de Hautes Énergies Sorbonne Université, CNRS/IN2P3

Mémoire présenté en vue de l’obtention de l’

Habilitation à diriger des recherches de Sorbonne Université

Spécialité: Physique

Soutenue le 19 Février 2018 devant le jury composé de:

Wystan BENBOW — Rapporteur Frédéric DAIGNE — Examinateur Kumiko KOTERA — Examinatrice Benoît LOTT — Rapporteur Tanguy PIEROG — Examinateur Jérôme RODRIGUEZ — Rapporteur January 30, 2018 À Maud, Lysandre & Abigail

c Jorge Cham

Contents

Contents vii

List of Figures ix

List of Acronyms xi

Acknowledgments xiii

Preamble xv

1 High energy emission in AGN1 1.1 Introduction1 1.1.1 The AGN phenomenology and zoology1 1.1.2 The link between AGN and ultra high energy cosmic rays6 1.2 The H.E.S.S. experiment and the future CTA observatory7 1.2.1 The H.E.S.S. experiment8 1.2.2 The CTA observatory9 1.3 Some studies on high energy emission in active galactic nuclei 11 1.3.1 High energy γ-ray emission from radio-quiet systems 11 1.3.2 First H.E.S.S. II results on AGN: the case of PKS 2155 304 − and PG 1553+113 in monoscopic mode 16 1.4 Towards time-domain high-energy astrophysics 20 1.4.1 Flaring AGN at (very) high energies 21 1.4.2 FLaapLUC: a pipeline for the generation of prompt alerts on transient Fermi-LAT γ-ray sources 25

2 From the sky to the ground: characterising the instrument performances 31 2.1 High Energy Stereoscopic System 32 2.1.1 The H.E.S.S. simulations & analysis frameworks 32 2.1.2 H.E.S.S. II performances 33 2.1.3 Run-wise simulations 37

vii Contents

2.1.4 H.E.S.S. I upgraded cameras 41 2.2 Two speciﬁc studies on the response of the future CTA: observations with Moon light, and site related studies 49 2.2.1 High altitude site 51 2.2.2 Performances under Moon light 53 2.3 Conclusion 55

3 Cosmic-ray electron-positron spectrum 57 3.1 Context & motivation 57 3.2 Updated cosmic-ray e± spectrum with H.E.S.S. 58

4 Prospects 63

Bibliography 67

Appendices 83

A Selected publications 85 A.1 FLaapLUC: A pipeline for the generation of prompt alerts on transient Fermi-LAT γ-ray sources 86 A.2 Cloud ablation by a relativistic jet and the extended ﬂare in CTA 102 in 2016 and 2017 93 A.3 Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 102 − A.4 The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT 115 A.5 Seyfert 2 galaxies in the GeV band: jets and starburst 129

B Curriculum Vitæ 137

C Publication list 141

Abstract 156

viii List of Figures

1.1 Composite image of the radio galaxy 3C 3482 1.2 Schematic representation of an AGN SED3 1.3 Schematics of the AGN classiﬁcation model4 1.4 SED of a sample of blazars5 1.5 All particle cosmic ray spectrum7 1.6 Hillas diagram8 1.7 A view of the H.E.S.S. array9 1.8 Artist rendering of CTA 10 1.9 TS maps of NGC 1068 and NGC 4945 from Fermi-LAT data 13 1.10 Relationship between SN rate, total gas mass and γ-ray luminosity for a few systems 14 1.11 SED of NGC 1068 15 1.12 Excess maps of events in the directions of PKS 2155 304 and PG 1553+113 − with H.E.S.S. II mono 16 1.13 Observed high energy SED of PKS 2155 304 and PG 1553+113 observed − with Fermi-LAT and H.E.S.S. II mono 17 1.14 Intrinsic high energy SED of PKS 2155 304 and PG 1553+113 derived − from Fermi-LAT and H.E.S.S. II mono observations 18 1.15 Sensitivities of CTA and Fermi-LAT with respect to integration time 20 1.16 Activity subsequent from cloud ablation by the relativistic jet in CTA 102 23 1.17 Light curves of PKS 1510 089 in 2015 24 − 1.18 Comparison of FLaapLUC and a likelihood analysis for PKS 2155 304 27 − 1.19 FLaapLUC false alarm trigger rate 28 1.20 Example light curve from FLaapLUC on PKS 0736+01 28 1.21 Results of the follow-up likelihood analysis automatically launched by FLaapLUC on PKS 0736+01 29

2.1 Illustration of the classical massive simulation framework 32 2.2 Sketch of the different H.E.S.S. II reconstruction and analysis modes 34 2.3 Example of a H.E.S.S. II effective area after analysis cuts 34 2.4 Differential sensitivity of H.E.S.S. II in the combined mode 35

ix List of Figures

2.5 Effet of the shower propagation discretisation step in the Čerenkov light distribution 36 2.6 Influence of the shower propagation discretisation on the effective areas 37 2.7 Illustration of the run-wise simulation framework 38 2.8 Measured NSB rate for an observation of the Galactic centre with CT5 39 2.9 Energy distribution comparisons between run-wise simulations, classical Monte Carlo simulations and data 39 2.10 Comparison between run-wise simulations and data of the squared angular distance (Θ2) to the source position, for PKS 2155 304 40 − 2.11 Crab extension as measured with H.E.S.S., overlaid on a Chandra image in X-rays 40 2.12 Layout of the trigger sectors in H.E.S.S. cameras 42 2.13 Sector trigger patterns 43 2.14 Next-neighbour trigger patterns 44 2.15 Pixel pattern that would fire a 2-NN trigger, but not a 4-NN one 44 2.16 3-NN trigger pattern overlapping between two drawers 44 2.17 Effect of the next-neighbour trigger on γ rays 46 2.18 Same as Fig. 2.17 for simulated protons 47 2.19 Effect of the pixel amplitude threshold on γ rays, for the N-majority trigger, under different NSB rates 48 2.20 Schematic view of the 13 equivalent cost arrays considered for the Prod1 Monte Carlo simulations in CTA 50 2.21 Comparison of the sensitivity for the different CTA array candidates for an altitude of 3700 m 52 2.22 Comparison of the sensitivity for the different CTA array candidates under Moon light 54

3.1 H.E.S.S. observations used for the derivation of the electron spectrum 59 3.2 Electron/proton discrimination with the model reconstruction 59 3.3 High energy electron and positron spectrum 61

x List of Acronyms

AGN active galactic nucleus BLR broad line region CAT Čerenkov Array at Themis CMB cosmic microwave background CTA Čerenkov Telescope Array EBL extragalactic background light EIC external inverse Compton EGI European Grid Infrastructure FSRQ ﬂat spectrum radio quasar GRMHD general-relativistic magnetohydrodynamics GZK Greisen-Zatsepin-Kuzmin H.E.S.S. High Energy Stereoscopic System HBL high-frequency-peaked BL Lac object HE high energy HEGRA High Energy Gamma-Ray Astronomy IACT imaging atmospheric Čerenkov telescope IBL intermediate-frequency-peaked BL Lac object LBL low-frequency-peaked BL Lac object LST large-sized telescope MAGIC Major Atmospheric Gamma-Ray Imaging Čerenkov MST medium-sized telescope NAG noyau actif de galaxie NSB night sky background PMT photomultiplier tube PSF point spread function SED spectral energy distribution SSC synchrotron self-Compton SST small-sized telescope ToO target of opportunity VERITAS Very Energetic Radiation Imaging Telescope Array System VHE very high energy

Acknowledgements

First and foremost, my thanks goes to the member of the jury, Wystan Benbow, Frédéric Daigne, Kumiko Kotera, Benoît Lott, Tanguy Pierog and Jérôme Rodriguez, for accepting this charge. I would like to warmly thank my colleagues from the H.E.S.S. collaboration, for all these years of fruitful exchanges, discussions, collaborations, hard work, and friendships. Many thanks goes to Catherine Boisson, Hélène Sol and Andreas Zech, for bringing me in this field, for their trust, support and friendship, since the time of my PhD studies. I am indebted to lots of people from the H.E.S.S. collaboration and the CTA consortium. I would not venture to provide an exhaustive list, but I reserve special mentions for (in no special order): Mathieu de Naurois, Santiago Pita, Bruno Khélifi, Arache Djannati-Ataï, Michael Punch, Markus Holler, Fabian Schüssler, Jonathan Biteau, Martin Raue, Wystan Benbow, Berrie Giebels, Gilles Henri, Jean-François Glicenstein, Yvonne Becherini, Heike Prokoph, Stefan Wagner, Michael Zacharias, Pol Bordas, Vincent Marandon, Lucie Gérard, Gabriele Cologna, Emma de Oña Wilhelmi, Andrew Taylor, David Sanchez, Gianluca Giavitto, Carlo Romoli, Manuel Meyer, Nukri Komin, François Brun, Pierre Brun, Frank Rieger, Justine Devin, Léa Jouvin, Arnim Balzer, Michael Gajdus, Joachim Hahn, Yves Gallant, Steve Fegan, Aldée Charbonnier, Stefan Ohm, Matthieu Renaud, Stefan Klepser, Karl Kosack, Susumu Inoue, Fabio Acero, Christian Farnier, Thierry Stolarczyk, Clementina Medina, Armand Fiasson. Thanks to my colleagues at LPNHE for their welcome. Many thanks to Pascal Vincent, Agnieszka Jacholkowska, Jean-Paul Tavernet, Julien Bolmont, François Toussenel, Matteo Cerruti, Sonia Karkar, Vincent Voisin, Jean-Luc Meunier, Jean- François Huppert, Patrick Nayman, Olivier Martineau-Huynh, Sophie Trincaz-Duvoid, Jacques Dumarchez, Mathieu Chrétien, Raphaël Chalmes-Calvet. Special thanks to Daniel Kerszberg for putting up with me during three years. Many thanks to all the colleagues I mixed with, and from whom I learned a lot, during my stays in different laboratories, or at conferences, workshops, symposia, meetings or just around a beer or a coffee. I am grateful to Pascal Vincent and Matteo Cerruti for their thorough reading of this manuscript.

xiii Acknowledgements

Un grand merci à mes proches, mes amis et ma famille. Merci à mes parents, pour m’avoir toujours fait conﬁance et m’avoir soutenu. Enﬁn, et surtout, un énorme merci à Maud, pour son amour, sa patience et son soutien. Merci de nous avoir donné deux petits anges. Merci à Lysandre et Abigail pour la joie que vous nous apportez, je vous aime, de tout mon cœur.

xiv Preamble

lmost 10 years ago, I started working on active galactic nuclei (AGN) within A the H.E.S.S. collaboration for my PhD studies. Since then, I have always been working in a way or another on high energy extragalactic astrophysics. However, starting from phenomenology with the elaboration and study of emission models for very high energy (VHE) emitting AGN during my PhD, I moved to high energy data analysis with Fermi-LAT and Monte Carlo simulations for CTA during my first post-doctoral contract, and focused on data analysis in optical for ATOM, Fermi-LAT, H.E.S.S. and governance matters for CTA during my second postdoctoral contract. Once hired at CNRS in 2012, I kept on working on data analysis for Fermi-LAT and H.E.S.S., as well as massive Monte Carlo simulations, especially dedicated to H.E.S.S. II. Over the last few years, in the framework of Daniel Kerszberg’s PhD thesis (2014–2017) dedicated to the study of diffuse Galactic emission from electrons and positrons and that I co-supervised with Pascal Vincent, I have also been working on the discrimination between photon- and electron/positron-induced air showers, and on the astrophysical e± spectrum. All in all, even though sticking with high energy extragalactic astrophysics, I think I developed different skills in phenomenology, analysis and instrument characterisation. In this manuscript, I will summarise some aspects of the work accomplished throughout these years on the two last items. In addition to scientific activities, I have also been engaged in more technical and managerial work, partly addressed in this manuscript, within the H.E.S.S. collaboration and CTA consortium. Since 2012, I have been in charge of the production of Monte Carlo simulations for one of the two simulation chains in use in H.E.S.S. I have also opened up the collaboration to the use of the European Grid Infrastructure (EGI) and have modified our softwares accordingly. From 2012 to 2016, I had served as deputy and then principal convener of the H.E.S.S. Extragalactic Working Group. During my post-doctoral fellowship at Landessternwarte in 2011 and 2012, I also worked on the government scheme and definitions of the CTA consortium and the CTA observatory, as well as on the interfaces between the two. Since July 2016, I have been serving as the scientific CTA group leader at LPNHE. The main contribution of the LPNHE in the CTA construction is the development and delivery of the front-end electronic boards for NectarCAM, a proposed camera design to equip the medium-sized telescopes of

xv Preamble

CTA. This electronic board is in charge of the signal acquisition, processing and trigger in the camera. Those last aspects will not be covered here. I have also endorsed several responsibilities of collective interest. For instance, I contributed to the organisation of the seminars held at LPNHE in 2013–2014, I have been serving as elected member of the laboratory council at LPNHE since 2014, or have been acting as academic mentor of three diﬀerent PhD students at LPNHE since 2016. A list of such involvements is provided in my curriculum vitæ in AppendixB. Chapter1 will introduce the AGN concept and explore some of my contributions in their study and the detection of their ﬂares with Fermi-LAT and H.E.S.S. Chapter2 will be about my involvement into instrument simulations and performance characterisation for H.E.S.S. and CTA. Even though this manuscript is mainly devoted to AGN researches and the characterisation of instrument responses for H.E.S.S. and CTA,I dedicate a short chapter (Chapter3) summarising a subset of Daniel Kerszberg’s PhD thesis, focusing on the updated e± spectrum measured with H.E.S.S. phase I. Finally, I will present some prospects in Chapter4.

xvi 1

High energy emission in active galactic nuclei

The universe is full of magical things patiently waiting for our wits to grow sharper. — Eden Phillpotts (1862–1960)

Contents 1.1 Introduction1 1.1.1 The AGN phenomenology and zoology1 1.1.2 The link between AGN and ultra high energy cosmic rays6 1.2 The H.E.S.S. experiment and the future CTA observatory7 1.2.1 The H.E.S.S. experiment8 1.2.2 The CTA observatory9 1.3 Some studies on high energy emission in active galactic nuclei 11 1.3.1 High energy γ-ray emission from radio-quiet systems 11 1.3.2 First H.E.S.S. II results on AGN: the case of PKS 2155 304 − and PG 1553+113 in monoscopic mode 16 1.4 Towards time-domain high-energy astrophysics 20 1.4.1 Flaring AGN at (very) high energies 21 1.4.2 FLaapLUC: a pipeline for the generation of prompt alerts on transient Fermi-LAT γ-ray sources 25

1.1 Introduction

1.1.1 The AGN phenomenology and zoology ctive galactic nuclei (AGN) count amongst the most energetic sources in the A Universe. They harbour a supermassive black hole in the centre of their host

1 1. High energy emission in AGN

Figure 1.1: Composite image of the radio galaxy 3C 348. Visible light was obtained with Hubble, and the radio image in purple was taken with the VLA. Credits: NASA, ESA, S. Baum and C. O’Dea (RIT), R. Perley and W. Cotton (NRAO/AUI/NSF), and the Hubble Heritage Team (STScI/AURA) galaxy, which is powered by an accretion disk emitting UV and X-ray thermal radiation. In some cases, relativistic jets are present on a much larger scale than the central engine. Active galactic nuclei come in a whole different set of flavours, which were basically given different names before it was realised that they shared some common properties. For instance, optical violently variables, flat spectrum radio quasars, blazars, BL Lac objects, Seyfert galaxies, Fanaroff-Riley radio galaxies (see e.g. Fig. 1.1), quasi stellar objects all denote some kind of AGN, categorised differently depending on some properties in a given wave band. An overview of the AGN “zoo” is given in Table 1.1, taken from Table 1 of the recent excellent review on AGN by Padovani et al. (2017). Accumulating information from different energy ranges thus enabled the recognition of common features among them. Antonucci (1993) and Urry & Padovani (1995) proposed the first scheme to unify AGN, with only the viewing angle of the system with respect to our light line of sight as key ingredient, the now so-called “strict” unified model. However, AGN were historically still split in two main categories: radio-loud and radio-quiet AGN, depending on their power at radio wavelengths (see Fig. 1.3). A more complete picture emerging nowadays also includes intrinsic physical parameters, such as the ratio between the AGN luminosity over the Eddington luminosity, the jet power, the host galaxy properties as well as the orientation. It is apparent now that a more correct categorization would be between jetted and non-jetted AGN, as advocated by Padovani (2017). Blazars are a specific class of AGN, dominated by non-thermal radiation from

2 1.1. Introduction

cm/mm MIR-NIR X-ray Gamma HE VHE Radio Sub-mm/FIR Optical-UV

Accretion disc Non-jetted AGN Hot corona Jet (HSP) Reflection Jet (LSP) "Soft excess" Dusty torus

Figure 1.2: Schematic representation of an AGN SED, for a non-jetted quasar, and jetted LBL and HBL. From Padovani et al. (2017). Credits: C. M. Harrison.

Table 1.1: The AGN zoo: list of AGN classes. Credits: Excerpt from Table 1 in Padovani et al. (2017).

Class/Acronym Meaning Main properties/reference Quasar Quasi-stellar radio source (originally) Radio detection no longer required −1 Sey 1 Seyfert 1 FWHM & 1, 000 km s −1 Sey 2 Seyfert 2 FWHM . 1, 000 km s QSO Quasi-stellar object Quasar-like, non-radio source QSO 2 Quasi-stellar object 2 High power Sey 2 RQ AGN Radio-quiet AGN see Padovani (2017) RL AGN Radio-loud AGN see Padovani (2017) Jetted AGN with strong relativistic jets; see Padovani (2017) Non-jetted AGN without strong relativistic jets; see Padovani (2017) Type 1 Sey 1 and quasars Type 2 Sey 2 and QSO 2 FR I Fanaroff-Riley class I radio source radio core-brightened (Fanaroff & Riley 1974) FR II Fanaroff-Riley class II radio source radio edge-brightened (Fanaroff & Riley 1974) BL Lac BL Lacertae object see Giommi et al. (2012) Blazar BL Lac and quasar BL Lacs and FSRQs

3 1. High energy emission in AGN

Figure 1.3: Schematics of the AGN classification model, presenting both radio loud and quiet classes. The observer sees different multi-wavelength features depending on the orientation of the system with the line of sight. Credits: Adapted from NASA. radio to γ-rays, exhibiting rapid variability and high degree of polarization. These properties are explained as non-thermal emission from a relativistic jet embedded in the source pointing close to the line of sight (Blandford & Rees 1978). Blazars themselves are further divided in subclasses: flat spectrum radio quasars (FSRQs), low-frequency-peaked BL Lac object (LBL), intermediate-frequency-peaked BL Lac object (IBL), high-frequency-peaked BL Lac object (HBL), depending on the peak frequency of their synchrotron component, but which also display different luminosities. FSRQs exhibit emission lines in the optical/UV range from e.g. the accretion disk, while BL Lac objects are devoid of it, and thus dominated by non-thermal radiation across the whole spectrum (see Fig. 1.2). The spectral energy distribution (SED) of blazars shows two main components, peaking in the infrared to X-ray band for the former, and at high energies for the latter. The low energy component is commonly attributed to synchrotron emission from relativistic electrons1 present in their jet. The nature of the high energy component is more controversial and subject to each source properties. Electrons from the jet can interact with their own synchrotron emission by inverse Compton scattering, the so-called synchrotron self-Compton (SSC) mechanism (Ginzburg & Syrovatskii 1965; Konigl 1981). In some sources such as FSRQs, the thermal emission from the

1The term “electron” in fact indiscriminately denotes electrons and positrons.

4 1.1. Introduction

Figure 1.4: SED of a sample of blazars, picturing the blazar sequence. From Fossati et al. (1998) accretion disk, the broad line region (BLR) or the dusty torus is important enough not to be negligible in the jet frame, and can give rise to external inverse Compton (EIC) radiation (Begelman & Sikora 1987) with such external radiation fields as target. In a hadronic framework, proton synchrotron and p–γ interactions can dominate the high energy emission (see e.g. Mannheim et al. 1991; Mücke & Protheroe 2001). Fossati et al. (1998) proposed the existence of a blazar sequence, characterised by an anti-correlation between peak frequency and luminosity (see Fig. 1.4). Since then, several studies have been conducted to assess whether this sequence is real or due to selection effects (see e.g. Padovani 2007; Nieppola et al. 2008; Ghisellini & Tavecchio 2008; Maraschi et al. 2008; Meyer et al. 2011; Giommi et al. 2012). In the meantime, several outliers from this sequence have been detected, with for instance objects exhibiting high peak frequency and bolometric luminosity (e.g. Padovani et al. 2012; Cerruti et al. 2017a). The Doppler boosting in blazar jets shifts their intrinsic SED to higher frequencies. Such an observational feature, as well as the observation of the blazar sequence, led Costamante & Ghisellini (2002) to propose a list of promising blazar candidates as very high energy (VHE, E >100 GeV) emitters, and especially composed of HBL objects, since the emission of their high energy component peaks the closest to the VHE energy window. Observations of such HBL candidates with imaging atmospheric Čerenkov telescope (IACT) led to the first extragalactic campaigns, thus naturally biasing the emerging population of VHE-emitting AGN towards HBL. However, more and more FSRQs have recently been detected at VHE, the first one being 3C 279 with MAGIC during a flaring event (MAGIC Collaboration et al. 2008), followed by PKS 1510 089 with H.E.S.S. (H.E.S.S. Collaboration et al. 2013a), as well as IBL − 5 1. High energy emission in AGN such as 3C 66A (Acciari et al. 2009), and LBL such as AP Lib (H.E.S.S. Collaboration et al. 2015). Because of the very nature of BL Lac objects, which exhibit only weak emission lines in their optical spectrum, if any, redshift measurement in these objects is challenging. The future CTA is expected to detect many such objects (Sol et al. 2013; Dubus et al. 2013; CTA Consortium et al. 2017), and population studies which will follow these detections will at the very least require basic knowledge such as their redshift so as to derive involved energetic budgets. It is thus very important to pave the path and prepare for large observational campaigns dedicated to redshift follow-up measurements of new AGN detected with CTA. I am currently involved in such preparatory campaigns, one of the first led to observations taken with the X-Shooter spectrograph on the VLT (Vernet et al. 2011), the results of which were published in Pita et al. (2014). For a given intrinsic luminosity, the detectability of blazars at VHE is not only a question of distance. VHE γ rays interact preferably with infrared and optical photons. Actually, the second most intense diffuse radiation field in the Universe after the cosmic microwave background (CMB) is the extragalactic background light (EBL) (see e.g. Hauser & Dwek 2001). The SED of this radiation field is composed of two main components, the first one peaking in the optical/near infrared, typically from 0.1 µm to 10 µm comprises all the light emitted by stars and galaxies since the recombination era. The second feature emits at longer wavelengths, from 10 µm to 500 µm, and is due to the reprocessing by dust of the first component. Along their propagation from astrophysical sources to the Earth, VHE γ rays will interact with the EBL by produing e± pairs, yielding an effective absorption of their VHE spectra, which increases with the energy of the VHE photons and the source distance. Inversely, the VHE observations of blazars can help putting upper limits on the flux level of the EBL, as was done e.g. in Aharonian et al. (2006b); MAGIC Collaboration et al. (2008). An example of such constraints is given in H.E.S.S. Collaboration et al. (2013)2 following the detection at VHE of the blazar PKS 0301 243. By combining γ-ray data from − different blazars, it is even possible to measure the flux and spectral shape of the EBL (Ackermann et al. 2012b; H.E.S.S. Collaboration et al. 2013b; Pueschel 2017; Moralejo et al. 2017).

1.1.2 The link between AGN and ultra high energy cosmic rays Cosmic rays are detected up to energies of 1021 eV, far exceeding human-made ∼ accelerators (see Fig. 1.5). The cosmic rays at the highest energies, beyond 1018 eV, ∼ are thought to be of extragalactic origin. This is corroborated by the detection of a flux suppression with the Pierre Auger Observatory above 3 1019 eV interpreted ∼ × either as the Greisen-Zatsepin-Kuzmin (GZK) effect, or an energy exhaust at the sources (Abraham et al. 2010). The GZK effect is due to the interaction of cosmic

2for which I am one of the corresponding authors.

6 1.2. The H.E.S.S. experiment and the future CTA observatory

Figure 1.5: All particle cosmic ray spectrum. The LHC energy reach, in the proton frame, is shown for comparison. From Kotera (2014). rays with photons of the CMB and which was independently predicted by Greisen (1966) and Zatsepin & Kuz’min (1966). In order to reach such high energies, particles need to sit close to zones of accelerations long enough to accumulate energy before escaping without important energy losses. In order to stay conﬁned, the Larmor radius of a particle should not exceed the characteristic size of its accelerator. This so-called Hillas3 criterion (Hillas 1984) is helpful to assess which type of sources could be accelerators of cosmic rays (see Fig. 1.6). Among the potential candidates, in the framework of hadronic emission processes dominating their high energy emission, AGN can thus be viewed as particle accelerators, even more eﬃcient than the LHC, but without control knobs.

1.2 The H.E.S.S. experiment and the future CTA observatory

Čerenkov light is produced when a high energy particle (cosmic hadron, electron or positron, γ-ray) hits the atmosphere, generating the development of an air shower

3A special thought for Michael Hillas, a monument in the ﬁeld, who passed away during the writing of this manuscript.

7 1. High energy emission in AGN

Figure 1.6: Updated Hillas (1984) diagram. Above the blue (respectively red) line, protons (respectively iron nuclei) can be confined up to an energy of 1020 eV. The most powerful source classes are reported. From Kotera (2014). of secondary particles which travel faster than light in the medium. This dim, fast flash of light can then be collected by imaging atmospheric Čerenkov telescope (IACT) equipped with fast electronics and large collective areas. Major currently operating IACT are MAGIC, VERITAS and H.E.S.S. The community at large has also accreted to work together on the next generation of instrument, CTA. H.E.S.S. and CTA are briefly introduced in the following.

1.2.1 The H.E.S.S. experiment I have been working with the H.E.S.S. collaboration from 2006 to 2009, and then since 2011 onward. A short description of the experiment is given here. The High Energy Stereoscopic System experiment is an IACT system located in the Khomas Highlands in Namibia, near the Gamsberg. The phase I of the experiment, run from 2003 to 2012, consisted in four 12 m telescopes, laid out in a square-like shape. The second phase of the experiment began with the addition in 2012 of a ﬁfth, larger 28 m IACT (dubbed CT5 in the following), in order to both increase the achievable energy range at low energies, and enhance the system sensitivity in its core

8 1.2. The H.E.S.S. experiment and the future CTA observatory

Figure 1.7: A view of the H.E.S.S. array. The four H.E.S.S. I telescopes sit on corners of a square of 120 m of side length. The H.E.S.S. II telescope, in operation since 2012, is located at the centre of the array. Credits: H.E.S.S. collaboration. energy range. With an increased collective area, thus enabling the detection of lower energy events with respect to H.E.S.S. I, CT5 makes of H.E.S.S. II the first hybrid IACT system currenlty in operation, consisting in telescopes of different sizes (see Fig. 1.7). H.E.S.S. is equipped with fast cameras, with photomultiplier tube (PMT) as photosensors in order to catch the faint Čerenkov light. The calibration of the instrument is described in Aharonian et al. (2004a) and Rolland(2005), the digitization of the signal is documented in Guy(2003) and Rolland(2005), and details on the trigger system can be found in Funk et al. (2004). More details on the H.E.S.S. I response and performances can be found e.g. in Aharonian et al. (2006a). Apart from observations of extragalactic sources, some of which presented below, one of the prime targets of H.E.S.S., thanks to its location in the Southern hemisphere, have been deep investigations of the Galactic centre (H.E.S.S. Collaboration et al. 2016), as well as an extensive scan of the Galactic plane (Aharonian et al. 2006c; Chaves & for the H.E.S.S. Collaboration 2009; Gast et al. 2011; Carrigan et al. 2013) which revealed a richness of sources (H.E.S.S. Collaboration et al. 2017b) and diffuse emission (Aharonian et al. 2006d; Abramowski et al. 2014; H.E.S.S. Collaboration et al. 2017c) not anticipated before. Chapter2 will discuss the instrument response of H.E.S.S., in particular H.E.S.S. II and of the recently upgraded H.E.S.S. I cameras. Section 1.3.2 in this chapter will also address first results obtained with H.E.S.S. II on two active galactic nuclei.

1.2.2 The CTA observatory The Čerenkov Telescope Array is the next generation, world leading class of IACT instrument. The CTA consortium gathers together more than 1400 scientists and engineers around the world, from more than 200 institutes in 32 countries. Its aim is to be build a γ-ray observatory at two sites, one in the Northern hemisphere, and another in the Southern hemisphere, so as to provide a complete VHE γ-ray sky coverage (see

9 1. High energy emission in AGN

Figure 1.8: Artist rendering of the Northern (top) and Southern hemisphere (bottom) sites of CTA. Credits: Gabriel Pérez Diaz, IAC, SMM.

Fig. 1.8). The chosen sites are near Paranal in Chile for the Southern site, and on the island of La Palma in the Canary Islands for the Northern one. Each array of telescopes will be made of three diﬀerent types of IACT:

a few large-sized telescopes (LSTs), specialised in the detection of low energy γ • rays, from 20 GeV to 150 GeV; ∼ from ﬁfteen medium-sized telescopes (MSTs) in the Northern site to twenty ﬁve • in the South, operating in the core energy range of 150 GeV–10 TeV. They will be at the heart of CTA, delivering most of CTA sensitivity4; a few tens of small-sized telescopes (SSTs), in the Southern site only, spread on • a large area to cover the highest energy range accessible with CTA, above a few TeV.

Each telescope type is optimised for the detection of γ rays in a particular energy range, thus supplying CTA with the largest possible energy coverage and sensitivity. For the MSTs and SSTs, diﬀerent versions of telescopes are proposed, with a single mirror structure or equipped with dual mirrors. The single-mirror telescopes are similar to the current H.E.S.S. I or VERITAS ones. The latter type would use, for the ﬁrst time applied in astronomy, a Schwarzschild-Couder optical design, notably cancelling

4I would like to note here that since July 2016, I have been the CTA group leader at LPNHE. The team is responsible for the delivery of front-end electronic boards for the NectarCAM project, a proposed camera design for the mid-sized telescopes of CTA. This hardware component embeds the logic to trigger and acquire Čerenkov data.

10 1.3. Some studies on high energy emission in active galactic nuclei aberrations. The proposed Schwarzschild-Couder telescopes, both for MSTs and SSTs, will be equipped with cameras embarking silicon photomultipliers, providing ﬁnely pixelated, compact focal planes. The scientiﬁc questions CTA will address are general to VHE astrophysics and are threefold:

understanding the origin and role of relativistic cosmic particles: • – what are the sites of high-energy particle acceleration in the Universe ? – what are the mechanisms responsible for cosmic particle acceleration ? – what is the interplay between accelerated particles and star formation and galaxy evolution ? probing extreme environments: • – what processes are at work close to neutron stars and black holes ? – what are the characteristics of relativistic jets, winds and explosions ? – how intense are radiation and magnetic ﬁelds in cosmic voids, and how do they evolve with time ? exploring frontiers in physics: • – what is the nature of dark matter and how is it distributed ? – are there quantum gravitational eﬀects on photon propagation ? – do axion-like particles exist ?

With many telescopes, CTA will be operated under versatile configurations (Oya et al. 2015). It will be possible to observe several targets at the same time by means of sub-array operations, which, for instance, could be used to monitor AGN, before repointing the whole array in case of major flares. Divergent pointing of the telescopes (Gerard & for the CTA Consortium 2015) could be an observing mode to perform large surveys (Dubus et al. 2013). Also, CTA will not be operated as an experiment, contrary to the existing MAGIC, H.E.S.S. and VERITAS instruments, but as an observatory, in the astronomical sense of the word. This means that a significant part of the observation time will be open, with data and analysis pipelines as deliverables of the CTA consortium to the community. More details on the CTA science cases can be found in CTA Consortium et al. (2017), as well as in the special issue of the Astroparticle Physics journal dedicated to CTA (see e.g. CTA Consortium et al. 2013).

1.3 Some studies on high energy emission in active galactic nuclei

1.3.1 High energy γ-ray emission from radio-quiet systems This section summarises a study which led to the ﬁrst reported detection of the prototypical Seyfert 2 galaxy NGC 1068 at high energies using Fermi-LAT public data.

11 1. High energy emission in AGN

The detection of another Seyfert 2 galaxy, NGC 4945, is also addressed. This work was followed by the publication Lenain et al. (2010), also presented in Appendix A.5. NGC 1068 and NGC 4945 display starburst activity, thus potentially enabling studies on the interplay between cosmic rays and the interstellar medium within such systems. With this topic in mind, and motivated by the then recent reports of high energy (HE, 100 MeV < E < 100 GeV) emission from M 82 and NGC 253 with Fermi-LAT ∼ (Abdo et al. 2010b), and respectively with VERITAS (VERITAS Collaboration et al. 2009) and H.E.S.S. (Acero et al. 2009) at VHE, as well as from the Andromeda galaxy (Abdo et al. 2010a), we enlarged such a research to other non-jetted galaxies close to our Milky Way, published in Lenain & Walter (2011).

High energy cosmic rays (E < 1018 eV) detected on Earth are commonly believed to originate from our Galaxy. Indeed, the magnetic field amplitude and coherence length is such that the Larmor radius of protons is smaller than the typical size of the Milky Way at these energies, thus trapping those cosmic rays therein. The study of cosmic ray production, propagation and diffusion within other extragalactic sources could help shedding light on our understanding of these phenomena in our own Galaxy. The core of NGC 1068 harbours both an AGN and starburst activity (see e.g. Lester et al. 1987; Jaffe et al. 2004). The former is mainly revealed through X-ray observation of Compton-thick emission (Matt et al. 2004), while the later is mainly detected at infrared wavelengths from the emission of the starburst activity (Thronson et al. 1989). The Fermi-LAT data from NGC 1068 were analysed using the standard Science Tools with gtlike, and revealed a HE emission at the 8.3σ level, later confirmed by the Fermi-LAT collaboration (Ackermann et al. 2012a), unveiling HE emission for the first time in a Seyfert 2 galaxy. The HE emission, depicted in Fig. 1.9, found when analysing the Fermi-LAT data is not significantly variable. Any variation on a yearly time scale or quicker would have denoted a signature from the AGN itself, which could have helped disentangling between the two possibilities – AGN or starburst dominated emission. However, the energetics involved in the HE radiation process can help. In pure starburst systems such as NGC 253 and M 82, Abdo et al. (2010b) reported HE 40 1 emission from these objects with a γ-ray luminosity of the order of 10 erg s− in the ≈ 100 MeV–5 GeV band. For comparison, in the same energy band, the γ-ray luminosity 40 1 41 1 of NGC 4945 is 2.0 10 erg s− , while the one of NGC 1068 is 1.7 10 erg s− , × × about one order of magnitude higher. Following the model by Pavlidou & Fields (2001), Abdo et al. (2010b) suggested to compare the γ-ray luminosity Lγ to the product of the total gas mass Mgas with the supernova rate RSN of the host galaxy. Indeed, if the γ-ray emission stems from starburst activity, it is expected that the HE emission is due to p–p processes, with high energy hadrons ejected by supernovæ interacting with the cold gas in the host system. Figure 1.10 shows a comparison of this relationship for a few systems, where NGC 1068 is already seen as peculiar. In this figure, other sources from the Local Group are also reported, in some of which we searched forHE γ-ray emission as well (Lenain & Walter 2011).

12 1.3. Some studies on high energy emission in active galactic nuclei

Figure 1.9: Top: TS map of NGC 1068 between 100 MeV and 100 GeV. The green ellipses show the 68% and the 95% position errors from the 1FGL catalogue, the cyan and magenta circles show respectively the position error (at 68% and 95% conﬁdence level) for the full data set with all the events accounted for, and for front events only. The white contours are taken from an optical image from the Digital Sky Survey, showing the extent of the Seyfert galaxy. The red boxed points denote the position of the two quasars nearby NGC 1068. Bottom: Same as the top panel, for NGC 4945. For clarity, we only present here the position error circle for all events. From Lenain et al. (2010).

13 1. High energy emission in AGN

45 ) 10 •1 NGC1068

(ph s 44 IC342 γ 10 L M94 NGC4945 M82 M81 M83 43 10 NGC253

MilkyWay 42 10 M31 M33

1041 LMC

SMC 1040

106 107 108 109 1010 R × M (M yr•1) SN gas sun

Figure 1.10: γ-ray luminosity against RSN Mgas for NGC 1068, NGC 4945 as well as local galaxies and starburst galaxies, as measured× or with reported upper limits using Fermi-LAT in black. The red points show the expected γ-ray luminosity from the model of Pavlidou & Fields (2001), accounting for uncertainties on RSN and Mgas. From Lenain & Walter (2011).

We thus presented in Lenain et al. (2010) an alternative scenario, in which the HE emission from NGC 1068 is proposed to originate from the AGN activity and dominated by leptonic processes. The approach of Lenain et al. (2008) is adopted, where the jet is misaligned with respect to the line of sight. The details of the radiative modelling are reported in Katarzyński et al. (2001); Lenain et al. (2008); Lenain (2009). At a few tenths of parsecs from the core of NGC 1068 lies a mildly relativistic, wind-like outflow (Gallimore et al. 2004, 2006), in which HE emission could be produced through EIC with infrared photons from the central starburst zone (Lester et al. 1987; Jaffe et al. 2004) as target field. All available public data in hard X-rays from INTEGRAL IBIS/ISGRI as of May 2010 were also analysed, and attributed to EIC emission from another population of leptons located in the vicinity of the accretion disc, which is typical to interpret X-ray emission from Seyfert galaxies. The overall SED of NGC 1068 is presented in Fig. 1.11. Since then, a few other radio-quiet systems have been detected at high energy with Fermi-LAT, such as the nearby M 31 galaxy (Abdo et al. 2010a), Circinus (Hayashida et al. 2013), or the ultra-luminous infrared galaxy Arp 220 (Peng et al. 2016). No firm conclusion has arisen yet on the origin of the HE emission from NGC 1068, several authors have argued in favour of the starburst scenario (e.g. Lacki et al. 2014; Persic & Rephaeli 2014; Massaro et al. 2015; Peng et al. 2016), while others have put arguments

14 1.3. Some studies on high energy emission in active galactic nuclei

Figure 1.11: SED of NGC 1068, including the Fermi-LAT spectrum. The black and red points are taken from the NED archive, the red ones denote data taken from the central part of NGC 1068. In blue, INTEGRAL IBIS/ISGRI data are reported in hard X-rays, which analysis is also described in Lenain et al. (2010). The EIC model for the outflow is shown in blue line, the corresponding SSC emission is shown in thin red and magenta lines for the first and second order SSC, respectively. The thick red line shows the different radiative components from the outflow. The green lines show the EIC component from the accretion disc. From Lenain et al. (2010).

forward for the AGN case (Yoast-Hull et al. 2014; Wojaczyński et al. 2015; Lamastra et al. 2016). It should be stressed however, that such interpretations are subject to the chosen estimator for the star formation histories in galaxies (see e.g. Ackermann et al. 2012a, for a discussion on the subject), as well as better measurements of the inner galaxy properties, such as the supernova rate, in order to better disentangle the starburst and AGN activities. Any hint of variability at HE would also betray an AGN origin of this emission.

On another point: Concerning M 31, it is interesting to note that the last ﬁndings by the Fermi-LAT collaboration (Ackermann et al. 2017b) are reminiscent from the excess reported in the Galactic centre (Ackermann et al. 2017a), but a detailed discussion on the potential implications for dark matter is beyond the scope of this manuscript.

15 1. High energy emission in AGN

13° H.E.S.S. H.E.S.S. -29° 12°

-30° PKS 2155-304 PG 1553+113

Declination (J2000) 4000 Declination (J2000) ° 2500 3500 11 2000 3000 ° 2500 1500 -31 PSF 2000 PSF 1000 1500 10° 1000 500 500 0 -32° 0

22h05m00s 22h00m00s 21h55m00s 21h50m00s 16h00m00s 15h55m00s 15h50m00s Right Ascension (J2000) Right Ascension (J2000)

Figure 1.12: Left: Excess map of events in the direction of PKS 2155 304 with H.E.S.S. II mono. The inset shows the PSF of the instrument. The energy threshold− for this analysis is 80 GeV. Right: Same as the left panel, for PG 1553+113. Here, the energy threshold is ≈100 GeV. From H.E.S.S. Collaboration et al. (2017a). ≈

1.3.2 First H.E.S.S. II results on AGN: the case of PKS 2155 304 and PG 1553+113 in monoscopic mode − This section is about the ﬁrst H.E.S.S. II paper devoted to AGN, focusing on the case study of PKS 2155 304 and PG 1553+113 to demonstrate the H.E.S.S. II ca- − pabilities in monoscopic mode. This work was realised in collaboration with Dmitry Zaborov, Andrew Taylor, Carlo Romoli and David Sanchez, and led to the publication H.E.S.S. Collaboration et al. (2017a), also presented in Appendix A.3. The famous VHE-emitting AGN PKS 2155 304 and PG 1553+113 were observed − during the commissioning phase of H.E.S.S. II. The aim of this study was to demonstrate the monoscopic capabilities of H.E.S.S. II, and to probe these AGN near their SED peaks at energies 100 GeV. The analysis of monoscopic events allows to lower the ≈ energy threshold of a factor 4 with respect to CT1–5 stereo analyses, at the expense ∼ of a larger background, and a worse angular resolution leading to a slightly poorer sensitivity to point-like sources. However the lower energy threshold of H.E.S.S. II mono analyses is well suited to studies of bright variable sources and/or soft spectra sources such as AGN. The observations of PKS 2155 304 with H.E.S.S. II used in this study took place − from April to November 2013, and in May–June 2014. PG 1553+113 was observed from May to August 2013. After data quality checks, as detailed in H.E.S.S. Collaboration et al. (2017a), the total live time on PKS 2155 304 is 56 h, with 43.7 h taken in − 2013 and 12.3 h in 2014. For PG 1553+113, the available data set amounts to 16.8 h of live time. The data are analysed using the model reconstruction (de Naurois & Rolland 2009, which is brieﬂy described in Chapter3, page 59), which was adapted to work with monoscopic events (Holler et al. 2015a). Figure 1.12 shows the resulting excess maps for both PKS 2155 304 and PG 1553+113, which are clearly detected − 16 1.3. Some studies on high energy emission in active galactic nuclei

ν [Hz] 1023 1024 1025 1026 1027 ] -1 −10 10 PKS 2155-304 [ erg s 1045 ν ] L -1 ν

s −11 -2 10

1044

− −10 10 12 CT5 mono 10 − dN/dE [ erg cm Fermi-LAT E>100MeV 11

2 10

E Fermi-LAT E>10GeV 10−12 43 Fermi-LAT E>50GeV 10 −13 H.E.S.S. I (CT1-4) 10 −13 − 10 10 1 1 10 102 103

− 10 1 1 10 102 103 E [GeV] ν [Hz] 1023 1024 1025 1026 1027 ] -1

−10 47 10 10 [ erg s

PG 1553+113 ν ] L -1 ν s -2 46 10−11 10

−10 CT5 mono 10 −12 45 10 −11 10 dN/dE [ erg cm Fermi-LAT E>100MeV 10 2

E Fermi-LAT E>10GeV 10−12 Fermi-LAT E>50GeV −13 H.E.S.S. I (CT1-4) 10 − 3 44 10−13 10 1 1 10 102 10 10

− 10 1 1 10 102 103 E [GeV] Figure 1.13: Top: HE SED of PKS 2155 304 obtained from the H.E.S.S. II mono analysis (2013 data only, blue circles with confidence− band) in comparison with the contemporaneous Fermi-LAT data with an energy threshold of 0.1 GeV (red triangles and confidence band), 10 GeV (green band), and 50 GeV (purple band) and contemporaneous CT1–4 data (grey squares). Bottom: Same as top panel, for PG 1553+113, assuming a redshift of z = 0.49 (Abramowski et al. 2015). The insets compare the H.E.S.S. confidence band with the Fermi-LAT catalogue data (3FGL, 1FHL and 2FHL). From H.E.S.S. Collaboration et al. (2017a).

17 1. High energy emission in AGN

ν [Hz] ν [Hz] 1023 1024 1025 1026 1027 1023 1024 1025 1026 1027 ] ]

-1 48 -1 10−9 10−9 10 PKS 2155-304 PG 1553+113 46

10 [ erg s [ erg s ν ν

] −10 L ] −10 47 L -1 -1 10 ν 10 10 ν s s -2 -2 1045 46 10−11 10−11 10

1044 45 dN/dE [ erg cm −12 dN/dE [ erg cm −12 2 10 2 10 10 E E CT5 mono, corrected for EBL CT5 mono, corrected for EBL 1043 Fermi-LAT E>100MeV Fermi-LAT E>100MeV 10−13 10−13 1044

− − 10 1 1 10 102 103 10 1 1 10 102 103 E [GeV] E [GeV] Figure 1.14: Left: Intrinsic HE SED of PKS 2155 304, with 2013 data of H.E.S.S. II mono corrected for EBL absorption. The black line is the− best-fit log-parabola model, and the cyan envelope denotes the 1σ errors on the combined Fermi-LAT and H.E.S.S. II analysis, statistical only. Right: Same as left panel, for PG 1553+113. From H.E.S.S. Collaboration et al. (2017a). with a significance of 36σ and 21σ for PKS 2155 304 and PG 1553+113, respectively. − PKS 2155 304 is still detected at 7.3σ when restricting the data set to energies below − 100 GeV, with an energy threshold of 80 GeV. This demonstrates the capabilities of ≈ CT5 at low energies. One of the validations of the analysis of CT5 data consisted in performing another analysis a la H.E.S.S. I, using only the CT1–4 telescopes with data collected simultaneously with the H.E.S.S. II mono data. These CT1–4 data were analysed using the H.E.S.S. I version of the model analysis method with loose cuts for a better overlap at low energies. This comparison yields an excellent agreement and is presented along with the HE and VHE SED obtained with Fermi-LAT and H.E.S.S. II mono data in Fig. 1.13. The Fermi-LAT data presented in this SED were analysed on a data set acquired contemporaneously with the H.E.S.S. ones, as well as the catalogue data from the 3FGL (Acero et al. 2015), 1FHL (Ackermann et al. 2013) and 2FHL (Ackermann et al. 2016a). The H.E.S.S. II mono data are best described assuming a log-parabolic spectral shape, for both PKS 2155 304 and PG 1553+113, with a sharp − peak observed around 100 GeV for both sources between the Fermi-LAT and H.E.S.S. ranges. When correcting the observed spectra from EBL absorption, using the model by Franceschini et al. (2008), the Fermi-LAT and H.E.S.S. II mono spectrum for the intrinsic source emission from PKS 2155 304 is found to be significantly curved as − well, with the intrinsic SED peaking around 10 GeV (see Fig. 1.14), presumably probing the peak of the intrinsic SSC emission in this HBL. The same exercice applied to PG 1553+113 revealed no significant curvature in the intrinsic SED, due to larger statistical and systematic uncertainties, and suggests that the peak of the intrinsic SED lies at 500 GeV (see Fig. 1.14). ∼

Systematic uncertainties A large part of this study was devoted to the estimation of the systematic uncertainties associated to the H.E.S.S. II mono reconstruction and analysis. Table 1.2 gives a summary of the diﬀerent contributions inﬂuencing the spectral parameters derived for

18 1.3. Some studies on high energy emission in active galactic nuclei

Table 1.2: Estimated contributions to the systematic uncertainties in the spectral measurements using H.E.S.S. II mono for the analyses presented here. Numbers separated by “/” correspond to PKS 2155 304 and PG 1553+113, respectively. From H.E.S.S. Collaboration et al. (2017a). −

Source of Uncertainty Energy Scale Flux Index Curvature MC shower interactions – 1% – – MC atmosphere simulation 7% – – Instrument simulation / calibration 10% 10% – – Broken pixels – 5% – – Live Time – <5% – – Reconstruction and selection cuts 15% 15% 0.1 / 0.46 0.01 / 0.8 Background subtraction – 6%/10% 0.14 / 0.46 0.12 / 0.6 Total 19% 20%/22% 0.17 / 0.65 0.12 / 1.0

PKS 2155 304 and PG 1553+113. In particular, the systematic error on the energy − scale has an impact on the flux normalisation, as well as on the spectral slope and curvature. Apart from background subtraction, all the listed uncertainties relate to the conversion of measured event counts into flux. This conversion is performed using the instrument response functions, which are derived from Monte Carlo simulations (see also Chapter2). The first group of systematic uncertainties reported in Table 1.2 is related to the particle interactions and the absorption of the induced Čerenkov light in the atmosphere. From a comparison of the γ-ray showers simulated with the CORSIKA and KASKADE generators (see Section 2.1.1 for more details), the uncertainty due to the shower interaction model does not exceed 1%. The atmospheric uncertainty corresponds to the effects of the atmospheric density profile, which influences the height of shower maximum and the production of Čerenkov light, and the atmospheric transparency, related to light diffusion by Mie and Rayleigh scattering. This contribution is dominated by the transparency, which directly affects the detected Čerenkov light yield, and thus the energy reconstruction. The remaining instrumental effects, such as the optical efficiency of the system and the electronics response, are included in the instrument simulation and calibration uncertainty. These latter effects are controlled through calibration devices (Aharonian et al. 2004b), as well as the measurement of Čerenkov light from atmospheric muons (Leroy 2004). The non-operational pixels in the CT5 camera, which fraction is kept below 5% in the data quality selection, and the electronics dead time contribute only marginally in the overall systematic uncertainty. The third group of uncertainties are due to analysis matters. The uncertainty on event reconstruction and analysis selection cuts is evaluated from a comparison of the measured spectra obtained with an alternative analysis chain. The night sky background (NSB) as well as irregularities, non-axial symmetry in the camera acceptance both affects the background subtraction (Berge et al. 2007). The uncertainties are summed in quadrature and given in the last row of Table 1.2. The reconstruction, event selection and background subtraction dominate the spectral index and curvature uncertainties, while the description of the atmosphere and

19 1. High energy emission in AGN

10•3

) •4 •1 10

s Fermi•LAT

•2 E = 25 GeV 10•5 E = 40 GeV 10•6 E = 75 GeV 10•7

dN/dE (erg cm •8 2 10 10 years 10•9 CTA

10•10

10•11

•12 Differential Flux E 10

10•13 10 102 103 104 105 106 107 108 109 1010 Time (s)

Figure 1.15: Differential sensitivities of CTA and Fermi-LAT with respect to integration time, at selected energies. From Funk et al. (2013) the instrument calibration have the largest influence on the energy scale and flux normalisation. These results demonstrated the successful use of monoscopic data from H.E.S.S. II for AGN studies, which are particularly well suited given the access to low energies and the generally soft spectrum of these objects. Efforts to achieve a better accuracy of the measurements are now under way within the H.E.S.S. collaboration, in particular by combining information from monoscopic and stereoscopic events, as well as by the development of Monte Carlo simulations matching closer to the real observational conditions (see Chapter2).

1.4 Towards time-domain high-energy astrophysics

AGN are highly variable objects, thus calling for coordinated multi-wavelength campaigns, with observations performed with different facilities as simultaneous as possible. Other transient events also require such campaigns and follow-up observations, such as γ-ray bursts, or for the search of electromagnetic counterparts to the recently discovered gravitational wave events or neutrino events. One of the biggest challenges in astrophysics for the coming years is the online treatment of an ever growing amount of data, especially with the aim of generating and reacting quickly to automatic alerts about transient events. This is exemplified by alerts for high-energy starting events from IceCube now integrated through GCN alerts and the VOEvent system (Cowen et al. 2016). Also, the recent direct discovery of gravitational waves (Abbott et al. 2016), especially for the neutron star merger event GW 170817 (Abbott et al. 2017b), and the associated observational campaigns brilliantly illustrate the connection of different multi-messenger collaborations and the need for quick follow-up observations. In H.E.S.S., efforts have been deployed to adapt and plug the DAQ system to the

20 1.4. Towards time-domain high-energy astrophysics

VOEvent system. Although not yet fully enabled for all types of alerts, automatic alerts from e.g. the IceCube or the LIGO/Virgo collaborations could then be caught directly and followed up with H.E.S.S. (Schüssler et al. 2015, 2017). The fifth, large central telescope in the H.E.S.S. array is indeed a transient machine. As shown by Funk et al. (2013) for CTA, any large size Čerenkov telescope system has an effective area so large, with respect to space based γ-ray telescope such as e.g. Fermi-LAT, that one of its scientific sweet spot is the search for transient objects (see Fig. 1.15). In the following, a few aspects related to the study of AGN flare and the search for transient events inHE data will be developed.

1.4.1 Flaring AGN at (very) high energies During the last years, I have been involved in several activities related to AGN flares with H.E.S.S. and/or Fermi-LAT data, a glimpse of which is given here. Blazars are notably variable sources, at all wavelength, from radio to VHE ranges, and on different timescales, from years down to minutes. The study of these flares provide information to constrain emission models. If the underlying emission mechanism involves the same process in quiescent and flaring states, flares could be due to a sudden increase in particle injection. They could also stem from intrinsically different processes, and be related to other emitting regions, other radiation processes on top of the one responsible for quiescent state, or other particle populations. For instance, so-called orphan γ-ray flares, that is, flares detected in γ-ray without counterpart at larger wavelengths, as observed e.g. in 1ES 1959+650 (Krawczynski et al. 2004), are very difficult to reconcile with the standard SSC model. They could arise from multiple emitting zones in a leptonic dominated jet (see e.g. Kusunose & Takahara 2006, or Section 4.2.3.4 in Lenain 2009) or from hadronic emission (see e.g. Böttcher 2005; Sahu et al. 2013). Interactions between the relativistic jet and orbiting stars could also give rise to flares (see e.g. Barkov et al. 2010, 2012). Propagation effects can modulate the observed electromagnetic emission and apparent source extension, such as diffuse halos or spectral spillover induced by cascades initiated by γ rays (see e.g. Aharonian et al. 1994) or by ultra-high energy cosmic rays (Oikonomou et al. 2014; CTA Consortium et al. 2017). The study of such effects in HE and VHE blazars can in turn constrain the amplitude and structure of the intergalactic magnetic field (Neronov & Vovk 2010; H.E.S.S. Collaboration et al. 2014) In an ideal world, observations of AGN flares would thus be conducted simultaneously and with facilities operating at all wavelengths. However, such exhaustive observational campaigns are prohibitive. Nonetheless, monitoring programs with regular observations of AGN, performed with instruments with small or large field of view (such as Fermi-LAT), can help to catch active states (see e.g. Section 1.4.2), the detection of which can then be used to trigger follow-up target of opportunity (ToO) observations involving a larger set of instruments. Among recent and/or major outbursts observed with γ-ray instruments, we can cite the cases of PKS 2155 304 (Aharonian et al. 2007; H.E.S.S. Collaboration et al. 2012) − with a 3-minute variability trend, IC 310 (Aleksić et al. 2014b,a) with less than 5 min

21 1. High energy emission in AGN characteristic timescale, 1ES 1215+303 which is variable on a daily scale (Abeysekara et al. 2017d), PKS 1441+25 (Abeysekara et al. 2015; Ahnen et al. 2015) with variability on a week scale, the prototypical BL Lac itself (Arlen et al. 2013; Tsujimoto et al. 2017; Feng et al. 2017) with hour- to minute-scale evolutions, 1ES 1959+650 (Kaur et al. 2017; Santander & for the VERITAS Collaboration 2017), or the very bright and fast outburst (on the minute scale, which is exceptional for Fermi-LAT) seen with Fermi-LAT in 3C 279 (Ackermann et al. 2016b) in June 2015 (see also below, page 24).

Extragalactic transients and search for Lorentz invariance violation Additionally, AGN flares can be used to address fundamental questions. For instance, constraints can be obtained on the minimum energy scale at which quantum gravitational effects could arise, in the framework of classes of quantum gravitation models implying a violation of Lorentz invariance (see Liberati 2013, for a review), and resulting in a modification of the speed of light with energy via an emerging dispersion relation (see e.g. Amelino-Camelia et al. 1998; Ellis et al. 2008; Bolmont 2016):

∆t 1 + n τn = s κn (1.1) n ± n ∆ (E ) ' 2H0EQG at order n, with EQG the energy at which quantum gravity eﬀects are expected to occur, H0 the Hubble constant, s = 1 (respectively +1) in the so-called superluminal ± − (respectively, subluminal) case, and κn a normalised distance to the source:

z n Z (1 + z 0) dz 0 κn = q (1.2) 0 3 Ωm(1 + z 0) + ΩΛ with z the redshift of the source, and Ωm and ΩΛ the standard ΛCDM cosmology parameters. Using the 2012 ﬂare from PG 1553+113 as observed with H.E.S.S. (Sanchez et al. 2015; Abramowski et al. 2015), which refereed article is reproduced in Appendix A.4, following an alert from the MAGIC collaboration (Cortina 2012), we5 extended the method developed by Martínez & Errando (2009) to non-background-free data sets. Limits on the energy scale at which quantum gravity eﬀects causing Lorentz invariance violation may arise are found to be, for the subluminal case, E > 4.11 1017 GeV QG,1 × and E > 2.10 1010 GeV for the linear and quadratic cases, respectively (see QG,2 × Abramowski et al. 2015, and Appendix A.4 for more details).

The swan song of a gas cloud in CTA 102 This work was pursued in collaboration with Michael Zacharias, Markus Böttcher, Stefan Wagner, Felix Jankowsky and Alicja Wierzcholska. The corresponding article (Zacharias et al. 2017a) , reproduced in Appendix A.2, has recently been accepted for publication in ApJ and is currently in press.

5This work was conducted in the framework of Camille Couturier’s PhD thesis (Couturier 2014), partially under my supervision for this particular study on PG 1553+113.

22 1.4. Towards time-domain high-energy astrophysics

Nov16 Dec16 Jan17 Feb17 Mar17 57670 Fermi (>100MeV) Day (a) 57745

] Accretion Disk s / 1e-05 BLR 2 1e-09

m Synchrotron c / SSC h p

[ IC/BLR

F 1e-06

1e-10

XRT (2-10keV) (b) ] s / ] 2 s / m 2 c / m c g / r g e r [

e ν [

F 1e-11 F 1e-11 ν

ATOM/R 1e-10 (c) ] s /

2 1e-12 m c / g r

e 1e-11 [

1e-13 57670 57700 57730 57760 57790 57820 1e+10 1e+13 1e+16 1e+19 1e+22 1e+25 MJD ν [Hz] Figure 1.16: Left: Light curves of Fermi-LAT, Swift/XRT and ATOM R-band data on CTA 102. The thick red lines are the proposed modelling result for a cloud being ablated in the jet of CTA 102. Note the logarithmic ﬂux scale. Right: Corresponding SED of CTA 102 with our proposed model, for two particular dates, MJD 57670 and MJD 57745. The thick black and red lines show modelled spectra for the beginning and the peak of the ﬂare, the coloured thin solid lines show the evolution of the model in 10-day steps towards the maximum. From Zacharias et al. (2017a).

Another case worth to note is the one of the quasar CTA 1026. This object has remained quiescent for a long time, before displaying an impressively active state late 2016, in optical, X-ray and HE bands. On top of a rising trend on a few weeks time scale, shorter, bright flares can be observed at these wavelengths. In a recent paper (Zacharias et al. 2017a), we propose that the 2 months-long activity (see Fig. 1.16) can be due to a gas cloud being ablated by the relativistic jet (Araudo et al. 2010), gradually feeding the jet with freshly injected material, giving rise to emission through leptonic processes, with inverse Compton scattering off the photon field from the BLR, dominating the SED in X rays andHE.

Location of the γ-ray emitting zone in FSRQ Recent studies on FSRQ with H.E.S.S., in which I contributed, include an extensive monitoring campaign on PKS 1510 089 (Zacharias et al. 2017b), initiated following − its discovery at VHE in 2009 with H.E.S.S. (H.E.S.S. Collaboration et al. 2013a). With this campaign, several flares were identified in different bands, showing different behaviours from one energy range to the other, and from a flare to the other (see Fig. 1.17 and also e.g. Barnacka et al. 2014; Saito et al. 2015). VHE flares appear to be rarer with respect to the other probed energy ranges, which could well be due to the lower duty cycle of IACT with respect to e.g. Fermi-LAT, and activity at lower energies is not necessarily accompanied with VHE activity, and vice versa. Indeed, in May 2016, a VHE flare of PKS 1510 089 was observed both by MAGIC − and H.E.S.S. (Zacharias et al. 2017c), with a variation of the VHE flux by more than a factor 10, while the source barely varied, at the 30% level at most, in the optical

6A quasar so powerful that its radio signal detected in the 1960’s as a then-unidentiﬁed source was once thought of being the work of a technologically advanced extraterrestrial civilization, which inspired The Byrds for a song, “C.T.A. 102” (cf. Wikipedia: CTA 102). A lot more could be told about inﬂuences of astrophysics in music, and vice versa, but that is another story...

23 1. High energy emission in AGN

Figure 1.17: Light curves of PKS 1510 089 in 2015, with nightly-averaged H.E.S.S. and − ATOM data, daily-binned Fermi-LAT data, and exposures with Swift/XRT. Dashed lines show the 2015 average in the respective bands. and HE bands. The VHE variation set on a time scale of less than 1 h, revealing intra-night variability for the first time in this source at VHE. The observed spectral break between HE seen with Fermi-LAT and VHE observed with H.E.S.S. and MAGIC suggests an absorption of γ rays by soft photons from the BLR. Along with the VHE variability, this constraints the emission zone at the inner edge of the BLR, or further away from the central engine. Similarly, the FSRQ 3C 279 exhibited a flare in June 2015 at VHE as observed with H.E.S.S. (Romoli et al. 2017), following the aforementioned major outburst detected by Fermi-LAT (Ackermann et al. 2016b). The VHE data were also used to set limits on Lorentz invariance violation effects. Once the VHE spectrum corrected from EBL absorption, and given the FSRQ nature of 3C 279, the data constrained the VHE emission zone to lie at the edge of the BLR in this source as well. Finally worth noting here, VHE emission was discovered in the FSRQ PKS 0736+017 with H.E.S.S. during an active state (Cerruti et al. 2017b,c), following an alert from a flare detected in Fermi-LAT data using a custom-made pipeline (see Section 1.4.2). The emerging picture concerning FSRQs, which also applies to this source, is that, in order to explain effective EIC emission on BLR photons while keeping absorption of γ rays low enough to account for VHE observations, the emission region should be at the edge of the BLR, at the closest, and not at the very base of the jet as is

24 1.4. Towards time-domain high-energy astrophysics sometimes assessed to explain VHE emission from misaligned AGN (see e.g. Lenain et al. 2008).

1.4.2 FLaapLUC: a pipeline for the generation of prompt alerts on transient Fermi-LAT γ-ray sources This section deals with a technical work I started in 2012, with the aim of providing near real time alerts within H.E.S.S. on active states in high energy sources using Fermi-LAT data. A paper has been submitted to the journal Astronomy & Computing, and was recently accepted, on November 21, 2017 (Lenain 2017a) and is reproduced in Appendix A.1. The corresponding code is available on GitHub. I developed these last years an automated analysis pipeline for Fermi-LAT data, written in Python, in production in the H.E.S.S. collaboration since early 2013, with the aim of processing a few hundred sources every morning, to be able to quickly react to any interesting activity and to help assessing whether to launch ToO follow-up observations with H.E.S.S. or other facilities. This pipeline (Lenain 2017b), called FLaapLUC for Fermi-LAT automatic aperture photometry Light C Urve, is described ↔ in the following.

Aperture photometry analysis The main aim of FLaapLUC being to quickly provide results for a bunch of sources on a daily basis, a full likelihood procedure is just prohibitive in terms of computation time, even when splitting the computation across many CPUs. In contrast with the standard likelihood analysis approach usually applied to Fermi-LAT data, an alternative, much quicker approach lies in the aperture photometry method. The aperture photometry is a simple method consisting in summing up photon counts in a region of interest, and weighing it with the instrumental exposure to derive an approximate flux measurement. The aperture photometry method uses the photon counts from a small region around the source of interest to be analysed, typically within 1◦, and assumes that the extracted signal is background-free. This is of course not true, especially for sources close to the Galactic plane, where the Galactic γ-ray diffuse emission can account for up to 90% of the signal, depending on the chosen region of interest. We will present in Section 1.4.2 a comparison with the usual full likelihood approach. The main difficulty resides in defining what a flare should look like, while the event is actually ongoing. Even once the entire data set is available, the definition of what a flare is is somewhat subject to debate. For instance, Nalewajko (2013) proposes that a flare would consist in finding the peak in a light curve and to consider the contemporaneous temporal window when the flux is at least half of the peak flux. Here, one can not use such a definition, which requires to have the full observation data set at hand. Indeed, the aim of FLaapLUC is to alert for an ongoing event, without knowing whether the last flux measurement still corresponds to a trend of rising flux, or whether the flare is already on its decay. Instead, the following approach is proposed. For a given source, a long-term weekly-binned light curve is pre-computed, thus currently using more than 9 years of

25 1. High energy emission in AGN

Fermi-LAT data. To assess whether the source is experiencing an active state, another light curve is computed every day, using bins of N1 days of duration, and the last flux bin measurement is compared to the long-term flux mean. If the last flux bin is significantly higher than the average flux, FLaapLUC flags the current flux state as active, and yet another, but finer, light curve is generated with bins of N2 days (with N2 < N1). If the last shortly binned flux is also significantly above the long-term flux mean, FLaapLUC will issue an alert. More quantitatively, and accounting for flux errors, the averaged flux (FLT ) on long-term data is computed. Let’s denote FN1,2 the last N1,2-days binned flux value, and δFN1,2 its error. A two-level criterion on the flux is set, based on the N1-days binned and N2-day binned light curves. The trigger threshold on the flux is such that:

FN > FLT + αN RMS(FLT ) + δFN 1 1 · 1 and FN > FLT + αN RMS(FLT ) + δFN (1.3) 2 2 · 2

Comparison between a likelihood and aperture photometry analysis As described above, the aperture photometry method can provide quick results, but the absolute scale of the output fluxes is biased since the background is accounted as a signal in this analysis. To compare the resulting light curves with respect to a proper likelihood analysis, I used the standard gtlike tool from the Fermi-LAT ScienceTools (version v10r0p5) to produce time-binned fluxes for some example sources (see Lenain 2017a). Figure 1.18 presents such a comparison, for the bright BL Lac object PKS 2155 304. − With respect to fluxes obtained with gtlike, the error distributions show that the aperture photometry overestimates the fluxes by 30–50% on average, and could be ∼ off up to a factor 2 in case of low activity. However, again, the point of FLaapLUC ∼ is not to give an absolute flux level, but rather to quickly alert when a source is experiencing a significant activity relative to its long-term state. The activity patterns in the light curves are indeed matching well between FLaapLUC and gtlike. This is strengthened in the bottom panels, which show a comparison of FLaapLUC and gtlike, once unbiased from their respective mean.

False alarm rate The false alarm trigger rate of FLaapLUC is evaluated by means of simulations, scanning over the αN1 parameter. For each monitored sources which never triggered FLaapLUC so far, 1000 mock light curves were simulated preserving the underlying probability density function and power spectral density from real data using the method described in Emmanoulopoulos et al. (2013), and adapted to Python by Connolly (2015). FLaapLUC was run using the simulated light curves as inputs, and the false alarm probability on αN1 is taken from the rate at which N2-day binned light curve are generated. Figure 1.19 presents the false alarm probability when varying the threshold parameter αN1 . The operation point used in the H.E.S.S. extragalactic working group is shown in red, corresponding to wrongly generating the ﬁner light curve in 0.3%

26 1.4. Towards time-domain high-energy astrophysics

Likelihood 0.80 FLaapLUC ) 1 − s 2

− 0.60 cm 6 − 10 ( 0.40

0.20 100 MeV – 500 GeV F

0.00 10 20 30

like 2 2 F −

like 1 1 F Relative error

FLaapLUC 0 0 F

9 9 5 5 FLaapLUC like F F − − 1 1

like 3 3 F − −

FLaapLUC 7 7

F − − 500 1000 1500 2000 2500 3000 3500 20 40 MJD 54000 −

Figure 1.18: Top panel: Comparison of the ﬂuxes for PKS 2155 304 obtained using − FLaapLUC (in blue) and gtlike with a photon index frozen to the 3FGL value (in red). Middle left: Relative error between FLaapLUC and gtlike. Middle right: Distribution of the relative errors. Bottom left: Ratio between the aperture photometry and the likelihood results, once unbiased from their respective average. Bottom right: Corresponding distribution. of the cases. The false alarm probability of the whole double-pass procedure is not evaluated, because too much resource consuming to be properly computed, but it is safe to state that with such running settings on αN1 = 2 and αN2 = 3, the false alarm rate is well below 0.1% for the AGN monitored within the H.E.S.S. collaboration. ∼

Fermi-LAT as a proxy for the H.E.S.S. observation program of transient sources As an example, Fig. 1.20 shows the light curve produced with FLaapLUC on February 18, 2015 in the morning on a high-energy ﬂare of PKS 0736+01. Following this event, H.E.S.S. observations were triggered, and the source was observed on February 18, 19 and 21. During the peak of the high-energy activity, the source was detected during the night of Feb. 19, 2015 at the 11σ level in a monoscopic analysis, and at 5.5σ in a stereoscopic analysis (Cerruti et al. 2017c, presented at the ICRC 2017). In the production instance of FLaapLUC within H.E.S.S., once an alert is issued on a potential ﬂaring event and if the source is well visible at the H.E.S.S. site the next night, a follow-up likelihood analysis is automatically launched, in order to quickly get more robust results before issuing any ToO with H.E.S.S. or other facilities. This

27 1. High energy emission in AGN

10 1

10 2 False alarm trigger rate

10 3

1.0 1.5 2.0 2.5 3.0

Figure 1.19: FLaapLUC false alarm trigger probability on the ﬁrst iterative light curve generation step as a function of αN1 . The setting used within H.E.S.S. for extragalactic sources is shown with the dotted gray lines and red point.

PKS0736+01, 3FGL J0739.4+0137 (z=0.189)

2.50 ) 1 s

m 2.00 c

h p

0 1.50 1

× (

) V e

G 1.00

0 0 5 - V e

M 0.50

0 0 1 (

F 0.00

3040 3045 3050 3055 3060 3065 3070 MJD-54000.0

Figure 1.20: FLaapLUC light curve on PKS 0736+01 issued on Feb. 18, 2015. The blue points show the 3-day binned light curve, while the red points denote the daily-binned one. The horizontal solid blue line is the long-term flux average FLT , while the horizontal dotted blue lines are 2 RMS(FLT ) away from FLT (αN = 2). The horizontal bold red line shows · 1 the flux threshold for αN2 = 3 above which FLaapLUC issues an alert on this specific source.

28 1.4. Towards time-domain high-energy astrophysics

Figure 1.21: Results of the follow-up likelihood analysis automatically launched by FLaapLUC, after issuing an alert on PKS 0736+01. The Fermi-LAT spectrum, shown in blue, is extrapolated into the VHE regime in grey, assuming an ad-hoc intrinsic spectral break of ∆Γ = 1 at 100 GeV, and then absorbed by the EBL in red. As a qualitative assessment on the chance of detection of such an event with H.E.S.S., a H.E.S.S. II sensitivity curve for 20 h of observations is shown as a solid black line. procedure avoids the generation of likelihood results for all monitored sources, but preselects interesting active targets with FLaapLUC. In the follow-up likelihood analysis, the source model is taken from the 3FGL catalogue, thus also allowing the spectrum of the source of interest to be curved if this source is described by a log-parabola in the 3FGL catalogue. The photon index is also let free to vary, so as to probe any spectral hardening feature with respect to the 3FGL values. The Fermi-LAT spectrum is then extrapolated to very high energies, imposing an ad-hoc spectral break of ∆Γ = 1 at 100 GeV to account for intrinsic spectral curvature or shortage of high-energy particles, and absorbed by the EBL using the model by Franceschini et al. (2008). The corresponding SED is then automatically transmitted to the H.E.S.S. AGN ToO core group, who is in charge of evaluating such alerts, and of triggering follow-up ToO observation with H.E.S.S. and facilities at other wavelengths, such as Swift. An example SED is given in Fig. 1.21 for the February 2015 event observed in PKS 0736+01. FLaapLUC has been successfully used within H.E.S.S. since early 2013, regularly delivering alerts on Fermi-LAT transients potentially interesting to follow with H.E.S.S. If adapted to IACT data, such a quick analysis pipeline will also be very useful in the CTA era (see e.g. Bulgarelli et al. 2013, for ongoing eﬀorts in this direction). One of the observing modes of CTA will consist in using sub-arrays to monitor diﬀerent parts of the sky simultaneously. With such a quick analysis framework at hand, it will be possible to promptly provide self triggers when a transient event occur in one of these patches, which could lead to repointing a target with the full array, and/or to alert

29 1. High energy emission in AGN external partners.

30 2

From the sky to the ground: characterising the instrument performances

Coming out of space and incident on the high atmosphere, there is a thin rain of charged particles known as primary cosmic rays. — Cecil Frank Powell (1903–1969)

Contents 2.1 High Energy Stereoscopic System 32 2.1.1 The H.E.S.S. simulations & analysis frameworks 32 2.1.2 H.E.S.S. II performances 33 2.1.3 Run-wise simulations 37 2.1.4 H.E.S.S. I upgraded cameras 41 2.2 Two speciﬁc studies on the response of the future CTA: observations with Moon light, and site related studies 49 2.2.1 High altitude site 51 2.2.2 Performances under Moon light 53 2.3 Conclusion 55

This Chapter will touch more technical works I have been involved in, related to the characterisation of the instrument response functions of H.E.S.S. II and CTA. Such characterisations are key in understanding the instrument and in analysing their data to derive proper scientiﬁc results.

31 2. From the sky to the ground: characterising the instrument performances

Figure 2.1: Illustration of the classical massive simulation framework. The optical eﬃciencies and NSB rate are the same for all telescopes. Credits: Jill Chevalier.

2.1 High Energy Stereoscopic System

I have been in charge since 2012 of the massive Monte Carlo productions for the characterisation of the H.E.S.S. II performances. I will brieﬂy cover some of the related aspects.

2.1.1 The H.E.S.S. simulations & analysis frameworks wo main simulation frameworks are in use within the H.E.S.S. collaboration, T with the CORSIKA (Heck et al. 1998) and KASKADE-C++ packages for the shower generation, and sim_telarray (Bernlöhr 2008) and Smash (Guy 2003) for the detector simulation. CORSIKA is a publicly available, open-source code and is used widely in the astroparticle community. KASKADE has been developed by the ARTEMIS-Whipple, CAT and H.E.S.S. collaborations, based on the original KASKADE code by Kertzman & Sembroski (1994). For the generation of γ-ray showers, Bernlöhr et al. (2013) have demonstrated an agreement of the Čerenkov light yield to within 5% between CORSIKA and KASKADE (see their Fig. 1). sim_telarray is based ∼ on software developed for HEGRA and is in use both in H.E.S.S. and CTA, while Smash was developed speciﬁcally for H.E.S.S. sim_telarray and Smash have also been shown to be in good agreement with each other (see Fig. 3 in Bernlöhr et al. 2013). The advantage of using diﬀerent chains for the air shower generation, and/or the detector simulation, is to internally check the consistency of the Monte Carlo simulatons.

32 2.1. High Energy Stereoscopic System

Similarly, two main analysis frameworks are used: HAP (H.E.S.S. Analysis Package) and parisanalysis. In both frameworks, diﬀerent reconstruction algorithms are implemented. For instance, the Hillas-based reconstruction (Hillas 1985) is available in both chains. More sophisticated methods are also implemented, such as, among others, template-based reconstructions: the ImPACT reconstruction chain (Parsons & Hinton 2014) in HAP, and the semi-analytical model one (de Naurois & Rolland 2009) in parisanalysis. In the following, the parisanalysis framework is used, along with KASKADE and Smash for the simulation part.

In the classical simulation framework, massive Monte Carlo simulations are performed over a multi-dimensional phase space, gridified on particular values of the different input parameters. These parameters typically are the simulated (thrown) energy of the initial particle or the spectral index of the input particle energy distribution, the zenith and azimuth angles, the optical efficiency of the detection system, the off-axis angle in the camera frame, and the NSB rate, which is assumed to be homogeneous across the whole filed of view in the cameras. Figure 2.1 illustrates this framework. The different instrument response functions (effective area, energy and angular resolutions) are then interpolated in between those fixed parameters when analysing actual data. Such a methodology is widely used for the analyses of IACT data, including the simulations performed for the CTA design studies and preparation. In Section 2.1.3, we will discuss about an alternative framework using the actual observation configuration during data taking, to provide a closer match between Monte Carlo simulations and data.

2.1.2 H.E.S.S. II performances In the last years, a lot of work has been accomplished within the H.E.S.S. collaboration to adapt the diﬀerent reconstruction and analysis chains to include CT5 in the frameworks. With a larger collection area, and thus lower energy threshold, CT5 can be used either to reconstruct events in monoscopic mode, or combined with the H.E.S.S. I telescopes in stereoscopic mode (see Fig. 2.2). The central array trigger can indeed deal with single-telescope triggers from CT5 and stereoscopic triggers from CT1–5 in the same data taking run. The analysis of sources can then be performed with three diﬀerent modes:

monoscopic: only the information of CT5 is used, neglecting the other four • H.E.S.S. I telescopes; stereoscopic: similar to the one of the H.E.S.S. I phase, but adapted to a • ﬁve-telescope array; combined: uses all the available information. When an event in reconstructed • both monoscopically and stereoscopically, the uncertainty on the reconstructed direction is used as a decision criterion on which result to use (Holler et al. 2015b).

33 2. From the sky to the ground: characterising the instrument performances

Figure 2.2: Sketch of the diﬀerent H.E.S.S. II reconstruction and analysis modes. From van Eldik et al. (2015).

Figure 2.3: Example of a H.E.S.S. II effective area after analysis cuts, for low-zenith angle observations. The solid blue line corresponds to the standard cut configuration as defined in Holler et al. (2015b), and the dashed curve to a looser configuration. From van Eldik et al. (2015).

The combined analysis thus uses both the monoscopic and stereoscopic information, providing the best energy coverage. Such an analysis was developed and checked both in the parisanalysis and HAP frameworks, and we presented first results applied on observations of the Crab nebula at the ICRC 2015 conference (van Eldik et al. 2015). In this work, I was in charge of the production and qualification of Monte Carlo simulations to derive the H.E.S.S. II instrument response functions. Figure 2.3 represents the corresponding analysis effective area for the combined analysis configuration, using standard cuts defined in Holler et al. (2015b). Around 300 GeV, the H.E.S.S. II effective area connects with the H.E.S.S. I one, showing, as expected, that the addition of CT5 in the H.E.S.S. array mainly influences the lowest energies. It is possible to reconstruct spectra down to 70–80 GeV, as was shown at the same ICRC 2015 conference for ≈ observations of AGN (Zaborov et al. 2015), which is detailed in Section 1.3.2, as well

34 2.1. High Energy Stereoscopic System

Figure 2.4: Differential sensitivity of H.E.S.S. II in the combined mode, with standard (blue solid line) and looser (dashed blue line) cut configurations, compared with H.E.S.S. I (red line). From van Eldik et al. (2015). as in the article H.E.S.S. Collaboration et al. (2017a), reproduced in Appendix A.3. To assess the overall H.E.S.S. II performance, Monte Carlo simulations of γ rays were generated, using the standard H.E.S.S. simulation framework (see Fig. 2.1 and Section 2.1.3) for different azimuth and zenith angles, off-axis angles in the camera, and optical efficiencies1. As an example, the differential sensitivity is calculated, for low-zenith angle observations ( 20◦), in the combined mode. Background events are ∼ taken from actual data with observations matching closely to the simulated zenith angle. Fig. 2.4 shows the corresponding sensitivity, that is the required flux to detect a source at 5σ confidence level in each energy bin, for an observation time of 50 h. Five bins in energy are used, with a minimum of 10 signal events and a signal-to-noise ratio of S/B > 5% in each bin. It can be seen that CT5 makes the array sensitive to energies much lower than previously achievable with H.E.S.S. I, as expected. However, the current combined analysis is not yet as sensitive as the stereo analysis around 200 GeV–2 TeV, since selection cuts still have to be optimised, which is a work in progress in the H.E.S.S. collaboration. In case systematics uncertainties can be reduced at the lowest energies, like for pulsar observations, it is even possible to detect γ rays below 20 GeV, as was shown for instance with the detection of pulsed emission from the Vela pulsar (Gajdus et al. 2015).

Validation of air shower simulations During the H.E.S.S. I phase, the simulations performed with KASKADE and Smash were extensively tested. KASKADE is known not to reproduce correctly the hadronic

1for which, actually, CT5 can be treated diﬀerently than CT1–4.

35 2. From the sky to the ground: characterising the instrument performances

Figure 2.5: Effect of the shower propagation discretisation step in the average Čerenkov light yield, for γ-ray showers thrown at 80 GeV and 0◦ of zenith angle. Credits: Heike Prokoph. interaction processes, but is suitable for simulations of γ rays. A comparison between CORSIKA and KASKADE was performed within CTA and was shown to give comparable results within 5–10% (Medina & Zech 2010; Bernlöhr et al. 2013). ∼ In the H.E.S.S. II era however, new checks were performed on the shower simulations and compared to available H.E.S.S. II data, as a validation test at the lowest energies. As an example, it was shown by Heike Prokoph that the discretisation in radiation length of the propagation of particles within the air showers was too coarse at low energies, which led to an overestimation of the Čerenkov light density on the ground (see Fig. 2.5), and thus a bias on the energy reconstruction. Reducing the step length for the computation of the particles path corrects for this effect, at the cost of increased computing time, but which is still fairly manageable using facilities such as the CC-IN2P3 or the European Grid Infrastructure (EGI) grid. Figure 2.6 shows this effect on a resulting effective area, where it can be seen that this numeric step length only has an influence at the lowest energies, below the H.E.S.S. I energy range.

Technical details: Grid computing A part of my recent work since 2012 has been dedicated to managing the massive Monte Carlo simulations for H.E.S.S. II, which are used to derive the instrument response functions to be used in an end-user analysis of H.E.S.S. data, in the framework of parisanalysis. Having worked earlier with the EGI during my PhD thesis (see Appendix C in Lenain 2009), I implemented the support of job submission and data handling through the EGI within the parisanalysis framework, which then allows the end users to seamlessly run jobs either at in a local computing cluster or on EGI, or even use both at the same time for input/output handling and/or CPU processing. This technical eﬀort has been presented at the SUCCES workshop2 in

2succes2013.sciencesconf.org

36 2.1. High Energy Stereoscopic System

Figure 2.6: Inﬂuence of the shower propagation discretisation on the resulting eﬀective areas.

2013 and documented in the corresponding proceedings (Lenain & de Naurois 2013, in French). The job submission on the H.E.S.S. virtual organization took advantage of the workload management system gLite3, and of DIRAC (Tsaregorodtsev et al. 2008) as of mid-2017, the last one being used within CTA as well since 2011 (Arrabito et al. 2012). As of writing this manuscript, the H.E.S.S. virtual organization on EGI has access to 11 sites, where more than 6000 concurrent jobs can be run, and to more ∼ than 500 TB of storage dedicated to H.E.S.S.

2.1.3 Run-wise simulations This work is currently ongoing, and is pursued in collaboration with Markus Holler, David Sanchez, Jill Chevalier and Mathieu de Naurois. Preliminary results were presented at the ICRC 2017 conference (Holler et al. 2017b). With the completion of the CT5 telescope, H.E.S.S. has become the first hybrid IACT system, with two different types of telescopes. As such, it can be viewed as a precursor for CTA. This hybrid system thus also increases the complexity for handling the simulations needed to estimate the instrument response functions, since the parameter space for massive Monte Carlo simulation literally explodes with the new available possibilities. For instance, the optical efficiency of the system depends on the telescope, and is especially different between CT1–4 and CT5, and evolves with time while not being in sync between the telescopes. The classic Monte Carlo framework in which massive simulations are run across the whole phase space, and

3http://grid-deployment.web.cern.ch/grid-deployment/glite-web

37 2. From the sky to the ground: characterising the instrument performances

Figure 2.7: Illustration of the run-wise simulation framework. Credits: Jill Chevalier. then interpolated to assess the instrument response functions (see Section 2.1.1) tends to become more and more expensive in terms of CPU time and storage needed to properly derive the instrument response functions then used at the analysis level. Instead, simulating each observation run with parameters as close to reality as possible becomes a viable alternative, and has the huge advantage of better matching to real data taking conditions, such as inhomogeneous NSB in the field of view, het- erogeneous optical efficiencies across telescopes, telescope tracking, or non-operational pixels (see Fig. 2.7). Each observation run, which lasts typically up to 28 min, can thus be accurately simulated. As an example, some fields of view exhibit highly inhomogeneous NSB rate (see Fig. 2.8) which are usually assumed as homogeneous in the classical Monte Carlo simulations, but are adequately accounted for in the run-wise simulations by using the real pixel per pixel calibration information. The reconstruction of the events are thus also more accurate, and comparisons of the simulated energy distributions with actual data and with the classical Monte Carlo simulations are presented in Fig. 2.9 as an example. As for the angular resolution of the experiment, the PSF is usually evaluated using lookup tables at similar observation configurations, and is typically around 30. Within the run-wise framework, the PSF of a given source can be directly simulated, thus improving morphological studies. This is illustrated in Fig. 2.10 where the squared angular distances to the position of PKS 2155 304 of individual events are compared − 38 2.1. High Energy Stereoscopic System

Figure 2.8: Measured NSB rate, in MHz per pixel, for an observation of the Galactic centre with CT5. From Holler et al. (2017b).

Figure 2.9: Energy distribution comparisons between run-wise simulations (left panel), classical Monte Carlo simulations ( right panel) and data ( both panels). In the right panel, the green histogram represents real data, while the black one is from the classical, massive Monte Carlo simulations. The latter clearly shows a systematic energy bias with respect to the run-wise simulations and data. Credits: Markus Holler.

39 2. From the sky to the ground: characterising the instrument performances

Figure 2.10: Comparison between run-wise simulations and data of the squared angular distance (Θ2) to the source position, for PKS 2155 304. The blue histogram corresponds to the ON events, while the green one is for OFF− events, scaled to the size of the ON region. The red histogram shows the simulated PSF obtained with the run-wise scheme, and re-weighted to the actual spectral shape of the source. From Holler et al. (2017b).

Figure 2.11: Chandra image in the X-rays of the Crab nebula. The H.E.S.S. extension is depicted as a solid white circle. The shaded annuli indicate the statistical and systematic uncertainties of the measurement. From Holler et al. (2017a).

40 2.1. High Energy Stereoscopic System with the corresponding run-wise simulation. The agreement between Monte Carlo simulation and real data is very good, and upper limit on a potential Gaussian extension width is evaluated at 2300 (at 3σ conﬁdence level), bringing extension measurements to a whole new level for IACT data analyses. Using such a simulation framework, and with correspondingly more precise instrument response functions such as the PSF description, it has been possible to reveal an extended emission from the Crab nebula of 5200 assuming a Gaussian source shape, as reported at the ICRC 2017 conference (see Holler et al. 2017a, and Fig. 2.11). The corresponding extension in the VHE range is larger than the synchrotron emission seen with Chandra, which is naturally explained by the radiative cooling of electrons. The size of the Crab nebula decreases with the electron energy, and the electrons producing the VHE emission have an energy well below the ones directly radiating synchrotron emission in X-rays (e.g. Kennel & Coroniti 1984).

2.1.4 H.E.S.S. I upgraded cameras In early 2017, with Michael Punch, we explored the value of a topological trigger for the upgraded H.E.S.S. I cameras (in the following, H.E.S.S. 1U), which is implemented but not activated for standard operations. This work is summarised here, which conclusions are briefly discussed in a paper on the H.E.S.S. 1U upgraded cameras led by Gianluca Giavitto, submitted to Astroparticle Physics and currently under review. The standard trigger scheme in the H.E.S.S. I cameras is based on the so-called sector threshold, or N-majority trigger, where 38 overlapping sectors are logically defined within a camera (see Fig. 2.12). An event is triggering the camera readout if at least N1 pixels within a sector are lit with an amplitude above N2 photoelectrons. The typical operations in the H.E.S.S. I phase are performed with a camera setting of at least 4 photoelectrons in at least 3 pixels (see Fig. 2.13). More details can be found in Rolland (2005). The hardware trigger operates before any data is recorded on disk, so it is important to ensure it is efficient in filtering interesting events, but also that it can limit the noise such as the night sky background. From Monte Carlo simulations of the full chain of air shower generation and detector simulations, the overall efficiency4 for γ rays at trigger level is about 15%, while it drops to about 1% for protons for a realistic optical efficiency of the system, and an NSB of 100 MHz. After almost 15 years of operations, the cameras of the H.E.S.S. I phase have been upgraded so as to refurbish the system with new electronics, as well as to decrease the dead time of the H.E.S.S. I cameras to trigger more often on stereoscopic events along with CT5. Indeed, the original CT1–4 cameras had a dead time of about 460 µs, during which potentially new coincident events with CT5 could not be recorded, ending up as monoscopic events in CT5 which has a much shorter dead time of 15µs (Bolmont ∼ et al. 2014). The upgrade of the H.E.S.S. 1U cameras helps recovering those events in stereoscopy, with a dead time of about 7 µs.

4The eﬃciency is energy dependent, the numbers given here are integrated over the whole energy range.

41 2. From the sky to the ground: characterising the instrument performances h rwr.Rgt aoto h etr.Teaaou ad tteeg fasco r omnt w sectors. two to common are sector a of edge the at cards analogue The sectors. the of Layout Right: drawers. the 2.12: Figure aoto h etr o h rge oi fthe of logic trigger the for sectors the of Layout H.E.S.S. aea.Lf:Lyu fteaaou ad,wt h ae ceeof scheme label the with cards, analogue the of Layout Left: cameras. Rlad(2005) . Rolland From Credits:

42 2.1. High Energy Stereoscopic System

Figure 2.13: Example of a pattern of pixels which could trigger a H.E.S.S. camera under the sector trigger scheme.

With H.E.S.S. 1U (see e.g. Klepser et al. 2017), along with the standard N-majority trigger, alternative trigger schemes have been proposed, such as the next-neighbour (NN) trigger which uses the topological information of triggered pixels in the camera. The main goal is to push the instrument performances at low energies. In order to decrease the energy threshold, the idea is to identify clusters of triggered pixels within the camera, which would then have higher chances to be due to actual air showers than to NSB fluctuations. Such a trigger logic has been studied for HEGRA (Bulian et al. 1998), as well as for MAGIC (Bastieri et al. 2001). In H.E.S.S., the NN-trigger is not implemented for the first H.E.S.S. I cameras, but it is, yet not activated, for both CT5 and H.E.S.S. 1U cameras. We studied such a topological trigger scheme for the upgraded H.E.S.S. 1U cameras, to assess whether the stereoscopic energy threshold could be lowered. The correspondingly increased statistics on sources with soft spectra would then help in having a better response, and thus also a better instantaneous sensitivity to extragalactic transients such as AGN flares. We focused here on the 3-NN trigger patterns (see Fig. 2.14), which are more directly comparable to the standard sector trigger in use with H.E.S.S., where 3 pixels in a sector above a typical pixel amplitude threshold of 4 photoelectrons will trigger the camera acquisition. The implementation in H.E.S.S. 1U is such that the NN trigger pattern should be fully contained within a given drawer. This means that a pattern looking like the one presented in Fig. 2.15 could fire a 2-NN trigger in two different drawers independently, although it could not fire a 4-NN one, and a pattern like the one in Fig. 2.16 would not trigger the camera for a 3-NN setting. As a solution, we proposed a slight modification of the logic, which could circum- vent these pathological behaviours. Introducing a score for each NN-trigger pattern depending on how the NN clusters are sharing pixels amongst drawers within a sector, the aforementioned patterns could be recovered or filtered out. This hybrid proposal is the following:

43 2. From the sky to the ground: characterising the instrument performances

Figure 2.14: Examples of patterns of pixels which could trigger a H.E.S.S. 1U camera with the next-neighbour topology. Top: 3-NN patterns. Bottom: 4-NN patterns.

Figure 2.15: Pixel pattern that would ﬁre a 2-NN trigger, but not a 4-NN one.

Figure 2.16: 3-NN trigger pattern overlapping between two drawers. Such a pattern does not ﬁre a trigger in the implemented H.E.S.S. 1U logic.

44 2.1. High Energy Stereoscopic System

For a 3-NN setting: • 1. Grant a score sd of 2 for 3-NN contained within the drawer d; 2. Grant a score sd of 1 for 2-NN at the edge of the drawer d; P 3. Trigger the camera if d sd > 2. For a 4-NN setting: • 1. Grant a score sd of 4 for 4-NN contained within the drawer d; 2. Grant a score sd of 3 for 3-NN contained within the drawer d; 3. Grant a score sd of 1 for 2-NN at the edge of the drawer d; P 4. Trigger the camera if d sd > 4.

An ideal case would be that each pixel “knows” about the exact topological distribution of its neighbouring ﬁring pixels, whatever the drawer they belong to. Even though this ideal NN-trigger scheme can not be implemented in H.E.S.S. 1U due to hardware limitations, it is anticipated that the hybrid NN scheme could be implemented quite easily on the FPGA of the drawer front-end boards. Such an implementation was tested on simulations with Smash, and we then compare the eﬃciency and purity of simulated air showers following:

the classical sector trigger; • the NN-trigger as implemented in H.E.S.S. 1U, that is, only focusing on the • topological information within one drawer; the hybrid NN-trigger, summing up the scores of different drawers with clusters; • the ideal NN-trigger scheme, ignoring drawer boundaries. • Figure 2.17 presents the effective areas obtained for the three different schemes of topological triggers presented above, for N = 3, and are compared with the N-majority logic. Interestingly, the overall effective areas are very similar between the different schemes. Thus, the topological trigger does not improve the γ-ray efficiency. However, Fig. 2.18 shows the equivalent for simulated protons. One should note that the lowest energies are dominated by events triggering also CT5, which trigger scheme remains the N-majority one. However, around 100 GeV–1 TeV, one notices that protons tend to trigger less, by 10%, with respect to the N-majority trigger, ∼ thus influencing the purity in γ rays by triggering less on hadronic showers. The NN trigger thus yields some potential to deliver purer data sets, thus decreasing the overall stereoscopic system dead time, even with the currently implemented NN algorithm. However, discussions within the H.E.S.S. collaboration led to a more direct improvement. An alternative approach, directly usable in H.E.S.S. 1U, proposed by our colleague Gianluca Giavitto, is to stick to the N-majority trigger as already implemented and in use for regular data tacking, but to tune the pixel amplitude trigger threshold depending on the mean night sky background rate of the observed field of view. For typical extragalactic fields, previous observations with H.E.S.S. have shown a mean NSB rate of about 75 MHz, while in Galactic regions, the NSB rate can be much higher, up to 300 MHz in very bright regions such as η Carinæ. For this study, we ∼ 45 2. From the sky to the ground: characterising the instrument performances h ahdvria iesos o h yrdN ae h nrytrsodatraayi cuts. analysis after threshold energy the are case, cuts trigger. NN analysis N-majority hybrid after the the areas to for effective respect shows, the with line comparison, panel vertical For left dashed here. top The discussed the schemes in trigger case. presented 3-NN areas NN different hybrid the the as for well shown as trigger, sector N-majority 2.17: Figure ffc fteN rge oooyo simulated on topology trigger NN the of Effect o right: Top rge ffiiny(ubro rgee vrsmltdevents). simulated over triggered of (number efficiency Trigger γ rays. o left: Top otmright: Bottom ffcieaesa rge ee for level trigger at areas Effective itiuin ftesmltdadtigrdevents. triggered and simulated the of Distributions otmleft: Bottom γ asudrtestandard the under rays ai fteeffective the of Ratio

46 2.1. High Energy Stereoscopic System Same as Fig. 2.17 for simulated protons. Figure 2.18:

47 2. From the sky to the ground: characterising the instrument performances cs,teeeg hehl fe nlsscuts. analysis after threshold energy the case, NSB extragalactic the in shows, line vertical dashed The events. triggered and simulated the of extragalactic the for shown left: are cuts analysis after for level trigger at 2.19: Figure ai fteeetv ra rsne ntetplf ae ihrsett tnadpxlapiuethreshold. amplitude pixel standard to respect with panel left top the in presented areas effective the of Ratio ffc ftepxlapiuetrsodon threshold amplitude pixel the of Effect γ asudrteNmjrt etrtigrshm,frdffrn ie mltd hehls o oprsn h ffcieareas effective the comparison, For thresholds. amplitude pixel different for scheme, trigger sector N-majority the under rays NSB case. γ as o h -aoiytigr ne different under trigger, N-majority the for rays, o right: Top rge ffiiny(ubro rgee vrsmltdevents). simulated over triggered of (number efficiency Trigger NSB rates. otmright: Bottom o left: Top ffcieareas Effective Distributions Bottom

48 2.2. Two specific studies on the response of the future CTA: observations with Moon light, and site related studies took a typical value of 75 MHz extragalactic fields, 100 MHz for Galactic fields, and 300 MHz for extremely bright fields. Figure 2.19 shows the resulting effective areas, with respect to the standard settings of 3 pixels above 4 photoelectrons in any sector used since the H.E.S.S. phase 1. Here, the reference (“standard”) simulations have been generated with an NSB rate of 100 MHz. One can clearly see an improvement of the effective areas for extragalactic fields, with a pixel amplitude threshold of 3.5 p.e., from the lowest energies up to 1 TeV. The proposal has thus been made in the ∼ H.E.S.S. collaboration to operate under these new configurations, and is currently under discussion.

2.2 Two speciﬁc studies on the response of the future CTA: observations with Moon light, and site related studies

In 2010 and 2011, I worked with Christian Farnier on preliminary characterisation of CTA performances with two aims: obtaining sensitivities and resolutions for high altitude sites, which was purposeful at that time when site searches for CTA was ongoing. The second aim was to evaluate the CTA response under mildly bright Moon light. The results were summarised in a CTA internal note (Farnier & Lenain 2011), which conclusions were reported in the refereed article Bernlöhr et al. (2013). Using the so-called Prod1 Monte Carlo simulation available at that time for CTA (Bernlöhr et al. 2013), we simulated the response of different proposed CTA array configurations. Using the sim_telarray package provided by Konrad Bernlöhr (Bernlöhr 2008), air showers generated with CORSIKA (Heck et al. 1998) were processed through the IACT detector simulation, and then analysed using a Hillas-type based reconstruction (Hillas 1985). The tail cuts image cleaning were adapted depending on the considered NSB level (see below). The CTA candidate array layouts considered in this study corresponds to those depicted in Fig. 2.20. Since then, the array layouts for the proposed sites and the corresponding Monte Carlo simulations have been refined (Hassan et al. 2015; Cumani et al. 2017), and updated performances for the overall CTA project have been presented in Maier et al. (2015, 2017). The aim of these Monte Carlo simulation studies were two-fold:

Assess the performances of the different array configurations for a high-altitude • site; Assess the performances of the different array configurations under moderate • Moon light.

In the ﬁrst case, this study focused on a site at an altitude of 3700 m, which was being considered as a potential site candidate at that time (Bulik et al. 2011). In the second case, the study focused on the performance degradations under a moderately increased NSB due to Moon light. For the present study, the considered primary particles are γ rays, protons and electrons. The zenith angle is set to 20◦ and the azimuth angle to 90◦ for all simulations.

49 2. From the sky to the ground: characterising the instrument performances

Figure 2.20: Schematic view of the 13 equivalent cost arrays considered for the Prod1 Monte Carlo simulations in CTA.

50 2.2. Two specific studies on the response of the future CTA: observations with Moon light, and site related studies Protons were simulated in the 5 GeV–500 TeV energy range, while electrons and γ rays were simulated between 3 GeV and 300 TeV. The simulations were performed using a photon index of Γ = 2 for the different particle energy distributions, and then − re-weighed using the Crab photon index for γ rays, and using the known spectra for cosmic rays otherwise. The simulations are then analysed using a standard Hillas-based reconstruction, using a pre-cleaning of the telescope images, named tail-cuts, so as to select only those for which the shower signal exceeds the NSB noise. Only images holding pixels with a content at least n times higher than the mean amplitude A of the images, 1 h i surrounded by pixels with a content at least n (n < n ) times higher than A , are 2 2 1 h i kept for the determination of the so-called Hillas moments (Hillas 1985). In the following, the presented differential sensitivities have been computed for 50 h of observations. We consider 5 bins per energy decade, with at least 10 events per bin, and with a significance above 5σ (see eq. 17 in Li & Ma 1983). We also require a number of 10 gammas above 5% of the background. These are the standard assumptions agreed upon within the CTA consortium (Bernlöhr et al. 2013). The results presented below have been optimised with respect to the minimum number of telescopes triggering the events (from 2 to 8 telescopes, in steps of 2) and to the different cuts (on the minimum amplitude of the images and the pixel cleaning) used. For each energy bin, we loop over all the configurations and keep the one achieving the best differential flux sensitivity. All the other instrument responses (angular resolution, energy resolution, background rate, effective area, etc...) are obtained for this particular configuration in this energy bin.

2.2.1 High altitude site A consequence of a high altitude site is to reduce the typical distance of the telescopes to the shower maximum, and hence to increase the density of received Čerenkov light at the observation level. The energy threshold of the array should therefore be lowered and the sensitivity in the low energy range increased. The aim of the exercice was to assess whether low energy performances could be signiﬁcantly increased and balance the corresponding higher costs due to deployment and operations at higher altitude.

51 2. From the sky to the ground: characterising the instrument performances

Figure 2.21: Top: Differential sensitivity for the different CTA array candidates for an altitude of 3700 m, optimised in telescope multiplicity. Middle:Ranking of the arrays for each energy bin for an altitude site of 3700 m. The lowest value (1, in deep blue) corresponds to the array with the best sensitivity in the considered bin, the highest value (13, pale orange) corresponds to the worst sensitive array. A value of 20 means that the sensitivity of the array is not defined for this energy bin. Bottom: Ratio of the differential sensitivities obtained at 3700 m and at 2000 m. A ratio of 0.1 means that at this energy the standard 2000 m performance is not defined, and a ratio of 0.05 that at this energy the 3700 m performance is not defined. 2.2. Two specific studies on the response of the future CTA: observations with Moon light, and site related studies Figure 2.21 presents the differential sensitivities at 3700 m of altitude for the thirteen studied arrays, as well as their respective ranking for each energy bin and ratios with respect to the standard 2000 m altitude. From this, it can be seen that the arrays E, I and K provides the overall best sensitivity over the largest energy range. The array with the best differential sensitivity for energies below 1 TeV is array B, but its performances degrade rapidly after such energies. Actually, this array is almost a factor 2 worse than the best array above 1 TeV, and the discrepancy is going up to a factor 4 for energies above 30 TeV. The gain in energy threshold is overall quite limited since it is at most lowered by one bin in energy and with a limited gain in differential sensitivity as the standard altitude sensitivity already catches up the high altitude one at 100 GeV.

2.2.2 Performances under Moon light Very high energy (VHE) observations under high night sky background (NSB) conditions, such as with Moon light, has the potential to increase the duty cycle of the observatory, and thus to allow more observations to be conducted in a given amount of time. Indeed, MAGIC and VERITAS have special trigger threshold conditions and filters at the camera entrances which allow to observe under moderate Moon light5, that is, when the telescopes does not point too close to the Moon, and when the lunar phase is not too important, typically less than 60–80% (Archambault et al. ∼ 2017; Ahnen et al. 2017). The increased duty cycle allowed by observations under moonlight conditions would be greatly beneficial for CTA, with an expected gain of the duty cycle increasing by 30% compared to dark nights, corresponding to a total ∼ observing time of 1300 h per year. Such potential increase of the observation time, ∼ and duty cycle, will be critical in the context of searches for transient events (AGN flares, gravitational waves or neutrino counterparts, ...) in the CTA era. However, the analysis energy threshold is expected to increase, which might have an impact on the observation strategies, such as privileging observations of bright sources and/or sources with hard spectra during these periods. Under a moderate Moon light, with a Moon phase of 60% and with telescopes ∼ pointing at about 90◦ away from the Moon, the NSB rate is increased by a factor 4.5 compared to standard dark sky conditions. For the analysis cuts optimization, ∼ the mean amplitude of the images is proportional with the square root of the NSB, s NSB A tel h i ∝ 100 MHz so that, for instance, a cleaning of n2 = 5, n1 = 10 for dark time observation (NSBtel = 100 MHz p.e./channel) will translate to n2 = 10.6, n1 = 21.2 for a night sky background of NSBtel = 450 MHz p.e./channel. The tail cuts have been adapted accordingly.

5On the contrary, H.E.S.S. does not observe when the Moon is above the horizon.

53 2. From the sky to the ground: characterising the instrument performances

Figure 2.22: Top: Differential sensitivity for the different array candidates for an altitude of 2000 m with Moon light, optimised in telescope multiplicity. Middle: Ranking of the arrays for each energy bin for observations under Moon light conditions. The lowest value (1, in deep blue) corresponds to the array with the best sensitivity in the considered bin, the highest value (13, pale orange) corresponds to the worst sensitive array. A value of 20 means that the the sensitivity of the array is not defined for this energy bin. Bottom: Ratio of the differential sensitivities obtained under Moon light conditions and for dark time observations. A ratio of 0.1 means that at this energy the standard 2000 m sensitivity is not defined, and 0.05 that at this energy the 2000 m sensitivity with moonlight is not defined. 2.3. Conclusion

Similarly to Fig. 2.21, Fig. 2.22 presents the differential sensitivities under Moon light for the thirteen studied arrays, as well as their respective ranking for each energy bin and ratios with respect to dark conditions6. In this case again, arrays E, I and K shows the best balance over the whole energy range. The energy threshold for observations under Moon light conditions is higher than what was found for dark time conditions, as expected. However, already at 160 GeV, the differential sensitivity ∼ performances are compatible within a factor 2 for all the 13 arrays under these two very different NSB configurations. The results obtained under Moon light conditions reveals a compatibility with those obtained under dark skies already from 160 GeV onward. Such observations could then increase the duty cycle of the future CTA and are strongly recommended.

2.3 Conclusion

This Chapter presented some ongoing progresses to better characterise and enhance the possibilities of the H.E.S.S. experiment, by using improved trigger schemes to achieve a better overlap in the stereoscopic regime between H.E.S.S. I and CT5, as well as with the development of run-wise simulations. Altogether, some possibilities remain to push further the capabilities of H.E.S.S. before CTA supersedes the performances of the current instruments. Concerning the simulations performed in the context of CTA, we have demonstrated the advantage of observing when the Moon is up, at the expense of a slightly increased energy threshold with respect to observations under dark conditions. The duty cycle of CTA can thus be substantially increased by this mean, which will beneﬁt to several key science programs, such as the monitoring of AGN, or responses to ToO alerts. Even though the site choices are now in the conclusion phase for CTA, with La Palma in the Canary Islands for the Northern site at an altitude of 2200 m, and in Chile close to the Paranal Observatory for the Southern site at an altitude of 2100 m, we studied in 2011 the eﬀect of implementing CTA at high altitude, an option which was considered back then. Establishing an IACT instrument at higher altitude could provide a better sensitivity in the low energy range.

6In the bottom panel, the eﬀect seen at the highest energies is due to a lack of statistics in the simulations.

Cosmic-ray electron-positron spectrum

There is a theory which states that if ever anyone discovers exactly what the Universe is for and why it is here, it will instantly disappear and be replaced by something even more bizarre and inexplicable. There is another theory which states that this has already hap- pened. — Douglas Noel Adams (1952–2001)

Contents 3.1 Context & motivation 57 3.2 Updated cosmic-ray e± spectrum with H.E.S.S. 58

This chapter summarises a part of the work undertaken by Daniel Kerszberg during his PhD thesis from 2014 to 2017 (Kerszberg 2017), which I co-supervised with + Pascal Vincent, and focuses on the analysis of the e− + e spectrum with H.E.S.S. with increased statistics with respect to previous results. A paper by the H.E.S.S. collaboration on this subject, led by Daniel Kerszberg, is currently in preparation. The results were presented at the ICRC 2017 conference1.

3.1 Context & motivation

he study of cosmic-ray electron and positron spectra is currently tantalizing: data T from PAMELA (Adriani et al. 2009) and AMS-02 (Aguilar et al. 2013, 2014) on the positron fraction revealed an excess with respect to standard predictions for secondary production in the interstellar medium. A signiﬁcant part of these positrons

1Link to the talk presented by Daniel Kerszberg at the ICRC 2017.

57 3. Cosmic-ray electron-positron spectrum should thus be produced as primary cosmic rays. Due to the rapid cooling of HE electrons and positrons through synchrotron and inverse Compton processes, these primary positrons should be produced relatively nearby. The origin of the HE electrons and positrons is one of the current key questions in astroparticle physics. The main scenarii involve acceleration and injection of e± by nearby pulsars (e.g. Hooper et al. 2009; Yüksel et al. 2009; Malyshev et al. 2009; Grasso et al. 2009; Profumo 2012; Linden & Profumo 2013; Cholis & Hooper 2013), dark matter (e.g. Bergström et al. 2008; Cholis et al. 2009c; Arkani-Hamed et al. 2009; Cholis et al. 2009b,a; Harnik & Kribs 2009; Fox & Poppitz 2009; Pospelov & Ritz 2009; Nelson & Spitzer 2010; Chang & Goodenough 2011), or acceleration of secondary particles in supernova remnants (e.g. Blasi 2009; Mertsch & Sarkar 2009; Ahlers et al. 2009; Kachelrieß et al. 2011; Kachelrieß & Ostapchenko 2013; Cholis & Hooper 2014; Kruskal et al. 2016; Cholis et al. 2017). Even though the prime application of IACT is the detection of γ rays, they are of course sensitive to Čerenkov radiation from showers initiated by cosmic-ray hadrons, positrons and electrons. The hadronic showers in fact constitute the overwhelming background signal in astrophysical source analyses for IACT. Once this background is discarded via event reconstruction/classiﬁcation, the remaining events are dominated by e± induced showers. In classical source analyses, this component can also be removed by applying so-called standard ON-OFF subtraction techniques (see e.g. Berge et al. 2007, ) to extract the γ-ray-like signal from astrophysical sources. However, with special treatments, IACT data can also be used to derive measurements + on the combined e−/e spectrum (e.g. Aharonian et al. 2008, 2009; Borla Tridon 2011; Staszak & VERITAS Collaboration 2015).

3.2 Updated cosmic-ray e± spectrum with H.E.S.S.

Along with an almost doubled statistics accumulated with respect to previous measurements with H.E.S.S. reported in Aharonian et al. (2008, 2009), we also take advantage of more sophisticated event reconstruction algorithms developed in the meantime (e.g. de Naurois & Rolland 2009), which can directly be applied to the data. A simple strategy is adopted here. H.E.S.S. I data are considered, using observations with fields of view outside the Galactic plane (see Fig. 3.1), to prevent any contamination from the diffuse Galactic emission (Aharonian et al. 2006d; H.E.S.S. Col- laboration et al. 2017c), and cutting out regions of known VHE-emitting AGN or other galaxies (such as NGC 253). The remaining events are treated as being the studied signal, presumably mainly made of electrons and positrons. Indeed, the incident electrons, positrons or photons generate very similar air showers in the atmosphere, and are barely distinguishable in IACT data. Detailed investigations of strategies for possible distinctions between e± and γ-ray showers are for instance documented in Garrigoux(2015) and Kerszberg(2017). In such extragalactic fields, from the extrapolation of recent measurements with Fermi-LAT (Ackermann et al. 2015), the

58 3.2. Updated cosmic-ray e± spectrum with H.E.S.S.

Figure 3.1: H.E.S.S. observations used for the derivation of the electron spectrum with H.E.S.S., overlaid on a Fermi-LAT all-sky count map. The Galactic plane is excluded from the data selection.

Figure 3.2: Electron/proton discrimination with the model reconstruction. Credits: Daniel Kerszberg. expected contribution from the diﬀuse extragalactic γ-ray background is negligible at VHE, as compared with the electron/positron one, which is in agreement with the upper limit provided with H.E.S.S.(Garrigoux 2015, see).

Details on the reconstruction First introduced in the CAT experiment (Le Bohec et al. 1998), and now widely used

59 3. Cosmic-ray electron-positron spectrum in H.E.S.S. (de Naurois & Rolland 2009), the principle of the model analysis used here is to fit the calibrated camera images using a bank of pre-simulated shower images, accounting for NSB fluctuations. This method thus uses the information of the whole field of view, as opposed to Hillas-based analysis which is based on a computation of the moments of the ellipse-shaped images, obtained after an image cleaning. The data are compared to the modelled images using the following estimator, considering all pixels in the camera as independent:

Npixels Npixels X X ln Ltel = ln Li = 2 ln Pi (si µi , σp, σγ) (3.1) i=1 − i=1 | with N the number of pixels in the camera, and where Pi (si µi , σp, σ ) is the pixels | γ probability density of observing a signal of intensity si in pixel i, with an expectation µi . σγ is the width of a unique photo-electron peak and σp is the pedestal width. This probability density, with n the photoelectron number, is given by:

n µi 2 ! X µi e− (si n) Pi (si µi , σp, σγ) = q exp − (3.2) 2 2 2(σ2 + σ2) | n n! 2π(σp + nσγ) − p n γ

In order to compare model predictions with the actual recorded shower images, and thus discriminate between γ-ray and hadronic showers, a goodness of ﬁt is used to measure the ﬁt quality, with:

Npixels 1 X G = [ln Li ln L µi ] (3.3) √2NdF i=1 − h i| By construction, if the pixel likelihoods behave as independent random variables, G is normally distributed. Similarly, pixels belonging to the shower core can be used to build a ShowerGoodness variable. The remaining pixels are used to construct the BackgroundGoodness variable, sensitive to hadronic clusters, or other irregularities. The MeanScaledShowerGoodness (MSSG) variable, the main one used in the model analysis to discriminate γ-like events from hadronic showers, is a scaled mean of the ShowerGoodness over the diﬀerent telescopes. The electron/positron data set is then selected using this analysis cut, as depicted in Fig. 3.2, among others such as the depth of the ﬁrst interaction in the shower development.

Results From 460 346 321 events, the data set is reduced to 480 739 remaining events after analysis cuts. Among these, the contamination from γ-like hadronic showers is evaluated to be 15% at 1 TeV, and <10% above 5 TeV, from Monte Carlo simulations ∼ of incident protons (Kerszberg 2017). The resulting electron/positron spectrum is shown in red in Fig. 3.3, along with the statistical and systematic errors associated with 2 the H.E.S.S. measurements . The H.E.S.S. e± spectrum is best ﬁt with a smoothly

2The details on the evaluation of the systematic errors are documented in Kerszberg(2017), and not discussed here.

60 3.2. Updated cosmic-ray e± spectrum with H.E.S.S.

+ Figure 3.3: HE e− + e cosmic-ray spectrum as observed with diﬀerent experiments, including the updated H.E.S.S. results included. Results are also shown from AMS-02 (Aguilar et al. 2014), MAGIC (Borla Tridon et al. 2011), VERITAS (Staszak & VERITAS Collaboration 2015), Fermi-LAT (Abdollahi et al. 2017) and the previous H.E.S.S. ones (Aharonian et al. 2008, 2009). Credits: Daniel Kerszberg.

+0.29 broken power-law, with an energy break at Eb = (0.94 0.02(stat.) 0.26(syst.)) TeV, ± +0.10 − with an electron index going from Γ1 = 3.04 0.01(stat.) 0.18(syst.) below Eb to +0.17 ± − Γ2 = 3.78 0.02(stat.) 0.06(syst.) in the highest energy range, and is dominated by ± − systematic uncertainties. A remarkable feature of the H.E.S.S. result is the steeply falling spectrum at the highest energies, up to 20 TeV, which can already tightly ∼ constrain the parameter space of comic-ray propagation models where the e± originate from nearby pulsars (Kobayashi et al. 2004). The energy break could be the sign of a transition between two regimes, from a large number of sources contributing to the spectrum, to a regime at the highest energies where a few, only the closest, contribute. Very recently, the DAMPE (Chang et al. 2017) collaboration confirmed the spectral characteristics (break around 0.9 TeV, and compatible electron indices) observed with H.E.S.S. (Ambrosi et al. 2017, see also the recent results from CALET, Adriani et al. 2017). Hooper et al. (2017) and Hooper & Linden (2017) discussed about new results by HAWC (Abeysekara et al. 2017b,a), in particular their measurements on close pulsars such as Geminga and Monogem, which radial extents observed with HAWC would imply a very low electron diffusion coefficient, lower than expected from models of diffusion into the local interstellar medium (Abeysekara et al. 2017a). If generalised, this would mean in turn that the positron excess observed by PAMELA and AMS-02 may originate from a more exotic mechanism, such as dark matter. Combined with

61 3. Cosmic-ray electron-positron spectrum the hereby presented H.E.S.S. results, Hooper & Linden (2017) argue that the HAWC result should rather only reflect local conditions near Geminga and Monogem, because the inefficient diffusion involved by HAWC measurements would imply the highest energy events observed with H.E.S.S. would only have travelled 10–20 pc before ∼ loosing their energy, a distance from Earth within which no plausible source of VHE cosmic rays is found. In any case, this subject is intensely debated in the community, and for instance analyses including H.E.S.S. II data, or further results from DAMPE or CALET, and then CTA, could help shedding light on this important question.

62 4

Prospects

Equipped with his ﬁve senses, man explores the universe around him and calls the adventure Science. — Edwin Powell Hubble (1889–1953)

he study of AGN at VHE is fascinating, as it links together astrophysical and T particle physics aspects, as well as fundamental physics. For instance, it allows us to study mechanisms of particle acceleration, to probe intergalactic magnetic fields or to test Lorentz invariance. Many recent discoveries have helped us understand a bit more how these objects work. On the astroparticle front, and maybe most notably, recent discoveries (diffuse neutrino emission, gravitational waves) have truly opened new multi-messenger windows in our studies of the cosmos. However, there is still a long way to go. In the following, I list some aspects which appears to me as key topics in the field.

Theoretical prospects Understanding the physics of AGN implies to deal with many diﬀerent topics: particle acceleration in jets e.g. through particle-in-cell simulations and the understanding of relativistic shocks (e.g. Sironi et al. 2015; Pelletier et al. 2017); the link between accretion and ejection via GRMHD modelling; the emission processes at work; the potential link with ultra-high energy cosmic rays for which AGN are good candidates (Takami & Horiuchi 2011; Ptuskin et al. 2013). A decisive AGN model should thus encompass all these aspects, and supersede the current limitations. For instance, particle-in-cell simulations can not yet reach individual Lorentz factor of 106 required ∼ to explain the very observation of VHE γ rays. GRMHD models can not yet explain how relativistic jets in AGN are collimated on such large scales.

Multi-messenger astronomy Studies of the HE emission in AGN relates to their link with astrophysical neutrino

63 4. Prospects sources and as potential contributors to the ultra high energy cosmic rays, in hadronic frameworks. Following the discovery by IceCube of diffuse astrophysical neutrino emission (IceCube Collaboration 2013), this question is even more timely, and thus questions on the nature (leptonic or hadronic) of the γ-ray emission in AGN. Future facilities should be able to help further with this discrimination, notably with e- ASTROGAM which will probe pion-decay bumps in different source classes (De Angelis et al. 2017). CTA will allow the measurements of hadronic spectral signatures, or lack of, in the spectra of VHE BL Lac objects (Zech et al. 2017). Also, any firm association of neutrino events seen with IceCube or future neutrino detectors coincident with AGN flares would unequivocally sign a hadronic origin of the emission. One cannot help but mention the birth of multi-messenger astronomy in the last few years, with the addition of direct detection of gravitational waves in the game, by the LIGO/Virgo consortium, first with the discovery of black hole mergers (Abbott et al. 2016). Even more spectacular are the very recent discovery of gravitational waves from a binary neutron star inspiral (Abbott et al. 2017a), and the associated global observational campaign leading to the direct link between γ-ray bursts and mergers of neutron stars (Abbott et al. 2017b), in which H.E.S.S. took part (H.E.S.S. Collaboration et al. 2017d). The future eLISA mission (Amaro-Seoane et al. 2012) will be able to detect gravitational waves from merging supermassive black holes from days to months before the final coalescence, and provide precise localisation. With such information at hand, observations at all wavelengths could be planned accordingly, and a detection of a neutrino and electromagnetic counterpart would be just magnificent! Interestingly, IceCube issued an alert in September 2017 on the detection of a high-energy, track-like neutrino candidate (Blaufuss 2017). This event was followed up at different wavelengths, including Fermi-LAT which detected increased γ-ray activity from the AGN TXS 0506+056, lying within the IceCube event error region (Tanaka et al. 2017), confirmed by AGILE (AGILE collaboration 2017). Subsequently, MAGIC claimed the detection of a significant γ-ray signal from TXS 0506+056 (Mirzoyan & for the MAGIC collaboration 2017). In parallel, VERITAS reported no detection of VHE γ-ray from this region (Mukherjee & for the VERITAS collaboration 2017), neither did H.E.S.S. (de Naurois & for the H.E.S.S. collaboration 2017). Further investigations are currently in progress. These activities reveal high interest from the community on linking the different multi-messenger information starting to be available to unveil the nature of extreme objects in the Universe.

Near-real time alerts Connected with multi-messenger aspects, but not only, time-domain HE astrophysics is a currently growing field. A global effort is being put forward to quickly, or even automatically, share information of ongoing flares and optimise strategies for follow-up observations. Notably, AMON (Smith et al. 2013; Cowen et al. 2016) is currently ramping up and will enable searches for multi-messenger transients in near real-time. Ongoing efforts in all current HE collaborations are on their way, by developing and deploying real-time analysis systems, such as in H.E.S.S. (Schüssler et al. 2017), IceCube (Aartsen et al. 2017) or HAWC (Abeysekara et al. 2017c). Naturally, such

64 a system is also being thought of within CTA (Bulgarelli et al. 2013), to be able to deliver self triggers and refocus the whole array on target, as well as processing incoming alerts from external facilities so as to automatically re-schedule observations, and to deliver alerts on VHE transients to the community. On the implementation side, GPU programming and/or machine-learning techniques are certainly a way to go to enable fast event reconstruction of IACT data, in order to identify transient events as fast as possible. About alerts coming from other wavelengths, one could mention for instance SVOM (see e.g. Cordier et al. 2015), for which one of the core programs is about autonomous transient detection of events such as AGN flares or γ-ray bursts, relevant for the future CTA science, as well as the need for a space mission operating in the MeV–GeV range in the lifetime of CTA. Methodologies should be studies also concerning the filtering, classification and decision trees for the numerous alerts that the next-generation of radio or optical telescopes will deliver, such as SKA, or LSST for which 107 transients are expected per night (Kantor 2014). ∼

Population studies Last but not least, as Carl Sagan’s saying goes, “extraordinary claims require extraordinary evidence”. An overall understanding, if any, of HE processes occurring in the Universe, generalised to whole source classes or fundamental questions, requires multiple positive observations, and thus, population studies are needed. Some examples are:

whether leptonic emission is characteristic of AGN flares and hadronic of the • quiescent states (or the other way around), or if one process dominates all the time; association of neutrinos or ultra-high energy cosmic rays to astrophysical sources; • detection of dark matter, axion-like particles, or Lorentz invariance violation... • All these questions need multiple, concordant observations before any definitive claim can be made. In VHE astrophysics, from a mere 15 sources detected by the previous generation of instruments such as HEGRA, CAT, Whipple and CANGAROO, the bestiary now amonts to more than 200 objects1, from all source types, still too few though to firmly see any emerging global trend. Future larger, thus more precise and sensitive, instruments such as CTA, will open up population studies in this part of the electromagnetic spectrum, as was the case before at lower energies.

In the next years, apart from the ongoing improvements on the characterisation of the H.E.S.S. instrument response functions, I would like to work on the building of such population studies of AGN with CTA, as well as on probing further their variable and transient nature, with an accent put on the multi-messenger aspects. I also would like to devote eﬀorts towards automatised ToO alert system for H.E.S.S. and CTA,

1cf. TeVCat: http://tevcat.uchicago.edu.

65 4. Prospects both for incoming and outgoing alerts, and on the development of analysis tools for transient events.

I am convinced that the advent of next-generation instruments promises exciting times ahead in extragalactic HE astronomy, in all channels (electromagnetic, neutrino, cosmic rays, gravitational waves) and with the help of close partnerships currently ﬂourishing between multi-wavelength and multi-messengers observatories and experiments. Exciting times ahead indeed...

Paris, December 15th, 2017 Jean-Philippe LENAIN

66 Bibliography

Aartsen, M. G., Ackermann, M., Adams, J., et al. 2017, Astroparticle Physics, 92, 30 (page 64.) Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2016, Physical Review Letters, 116, 061102 (pages 20 and 64.) Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017a, Physical Review Letters, 119, 161101 (page 64.) Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017b, ApJ, 848, L12 (pages 20 and 64.) Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010a, A&A, 523, L2+ (pages 12 and 14.) Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010b, ApJ, 709, L152 (page 12.) Abdollahi, S., Ackermann, M., Ajello, M., et al. 2017, Phys. Rev. D, 95, 082007 (page 61.) Abeysekara, A. U., Albert, A., Alfaro, R., et al. 2017a, Science, 358 [arXiv:1711.06223] (page 61.) Abeysekara, A. U., Albert, A., Alfaro, R., et al. 2017b, ApJ, 843, 40 (page 61.) Abeysekara, A. U., Alfaro, R., Alvarez, C., et al. 2017c, ApJ, 843, 116 (page 64.) Abeysekara, A. U., Archambault, S., Archer, A., et al. 2015, ApJ, 815, L22 (page 22.) Abeysekara, A. U., Archambault, S., Archer, A., et al. 2017d, ApJ, 836, 205 (page 22.) Abraham, J., Abreu, P., Aglietta, M., et al. 2010, Physics Letters B, 685, 239 (page6.) Abramowski, A., Aharonian, F., Ait Benkhali, F., et al. 2014, Phys. Rev. D, 90, 122007

67 Bibliography

(page9.) Abramowski, A., Aharonian, F., Ait Benkhali, F., et al. 2015, ApJ, 802, 65 (pages 17 and 22.) Acciari, V. A., Aliu, E., Arlen, T., et al. 2009, ApJ, 693, L104 (page6.) Acero, F., Ackermann, M., Ajello, M., et al. 2015, ApJS, 218, 23 (page 18.) Acero, F., Aharonian, F., Akhperjanian, A. G., & et al. 2009, Science, 326, 1080 (page 12.) Ackermann, M., Ajello, M., Albert, A., et al. 2015, ApJ, 799, 86 (page 58.) Ackermann, M., Ajello, M., Albert, A., et al. 2017a, ApJ, 840, 43 (page 15.) Ackermann, M., Ajello, M., Albert, A., et al. 2017b, ApJ, 836, 208 (page 15.) Ackermann, M., Ajello, M., Allafort, A., et al. 2013, ApJS, 209, 34 (page 18.) Ackermann, M., Ajello, M., Allafort, A., et al. 2012a, ApJ, 755, 164 (pages 12 and 15.) Ackermann, M., Ajello, M., Allafort, A., et al. 2012b, Science, 338, 1190 (page6.) Ackermann, M., Ajello, M., Atwood, W. B., et al. 2016a, ApJS, 222, 5 (page 18.) Ackermann, M., Anantua, R., Asano, K., et al. 2016b, ApJ, 824, L20 (pages 22 and 24.) Adriani, O., Akaike, Y., Asano, K., et al. 2017, Physical Review Letters, 119, 181101 (page 61.) Adriani, O., Barbarino, G. C., Bazilevskaya, G. A., et al. 2009, Nature, 458, 607 (page 57.) AGILE collaboration. 2017, The Astronomer’s Telegram, 10801 (page 64.) Aguilar, M., Aisa, D., Alvino, A., et al. 2014, Physical Review Letters, 113, 121102 (pages 57 and 61.) Aguilar, M., Alberti, G., Alpat, B., et al. 2013, Physical Review Letters, 110, 141102 (page 57.) Aharonian, F., Akhperjanian, A. G., Anton, G., et al. 2009, A&A, 508, 561 (pages 58 and 61.) Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. 2004a, A&A, 425, L13 (page9.) Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. 2004b, Astroparticle Physics, 22, 109 (page 19.)

68 Bibliography

Aharonian, F., Akhperjanian, A. G., Barres de Almeida, U., et al. 2008, Physical Review Letters, 101, 261104 (pages 58 and 61.) Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2007, ApJ, 664, L71 (page 21.) Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006a, A&A, 457, 899 (page9.) Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006b, Nature, 440, 1018 (page6.) Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006c, ApJ, 636, 777 (page9.) Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006d, Nature, 439, 695 (pages9 and 58.) Aharonian, F. A., Coppi, P. S., & Voelk, H. J. 1994, ApJ, 423, L5 (page 21.) Ahlers, M., Mertsch, P., & Sarkar, S. 2009, Phys. Rev. D, 80, 123017 (page 58.) Ahnen, M. L., Ansoldi, S., Antonelli, L. A., et al. 2015, ApJ, 815, L23 (page 22.) Ahnen, M. L., Ansoldi, S., Antonelli, L. A., et al. 2017, Astroparticle Physics, 94, 29 (page 53.) Aleksić, J., Ansoldi, S., Antonelli, L. A., et al. 2014a, Science, 346, 1080 (page 21.) Aleksić, J., Antonelli, L. A., Antoranz, P., et al. 2014b, A&A, 563, A91 (page 21.) Amaro-Seoane, P., Aoudia, S., Babak, S., et al. 2012, Classical and Quantum Gravity, 29, 124016 (page 64.) Ambrosi, G., An, Q., Asfandiyarov, R., et al. 2017, ArXiv e-prints [arXiv:1711.10981] (page 61.) Amelino-Camelia, G., Ellis, J., Mavromatos, N. E., Nanopoulos, D. V., & Sarkar, S. 1998, Nature, 393, 763 (page 22.) Antonucci, R. 1993, ARA&A, 31, 473 (page2.) Araudo, A. T., Bosch-Ramon, V., & Romero, G. E. 2010, A&A, 522, A97 (page 23.) Archambault, S., Archer, A., Benbow, W., et al. 2017, Astroparticle Physics, 91, 34 (page 53.) Arkani-Hamed, N., Finkbeiner, D. P., Slatyer, T. R., & Weiner, N. 2009, Phys. Rev. D,

69 Bibliography

79, 015014 (page 58.) Arlen, T., Aune, T., Beilicke, M., et al. 2013, ApJ, 762, 92 (page 22.) Arrabito, L., Barbier, C., Graciani Diaz, R., et al. 2012, Journal of Physics Conference Series, 396, 032007 (page 37.) Barkov, M. V., Aharonian, F. A., Bogovalov, S. V., Kelner, S. R., & Khangulyan, D. 2012, ApJ, 749, 119 (page 21.) Barkov, M. V., Aharonian, F. A., & Bosch-Ramon, V. 2010, ApJ, 724, 1517 (page 21.) Barnacka, A., Moderski, R., Behera, B., Brun, P., & Wagner, S. 2014, A&A, 567, A113 (page 23.) Bastieri, D., Bettini, M., Bigongiari, C., et al. 2001, Nuclear Instruments and Methods in Physics Research A, 461, 521 (page 43.) Begelman, M. C. & Sikora, M. 1987, ApJ, 322, 650 (page5.) Berge, D., Funk, S., & Hinton, J. 2007, A&A, 466, 1219 (pages 19 and 58.) Bergström, L., Bringmann, T., & Edsjö, J. 2008, Phys. Rev. D, 78, 103520 (page 58.) Bernlöhr, K. 2008, Astroparticle Physics, 30, 149 (pages 32 and 49.) Bernlöhr, K., Barnacka, A., Becherini, Y., et al. 2013, Astroparticle Physics, 43, 171 (pages 32, 36, 49, and 51.) Blandford, R. D. & Rees, M. J. 1978, Phys. Scr, 17, 265 (page4.) Blasi, P. 2009, Physical Review Letters, 103, 051104 (page 58.) Blaufuss, E. 2017, GCN, 21916 (page 64.) Bolmont, J. 2016, Habilitation à diriger des recherches, Université Pierre et Marie Curie; https://tel.archives-ouvertes.fr/tel-01388037 (page 22.) Bolmont, J., Corona, P., Gauron, P., et al. 2014, Nuclear Instruments and Methods in Physics Research A, 761, 46 (page 41.) Borla Tridon, D. 2011, International Cosmic Ray Conference, 6, 47 (page 58.)

70 Bibliography

Borla Tridon, D., Colin, P., Cossio, L., et al. 2011, ArXiv e-prints [arXiv:1110.4008] (page 61.) Böttcher, M. 2005, ApJ, 621, 176 (page 21.) Bulgarelli, A., Fioretti, V., Contreras, J. L., et al. 2013, ArXiv e-prints [arXiv:1307.6489] (pages 29 and 65.) Bulian, N., Daum, A., Hermann, G., et al. 1998, Astroparticle Physics, 8, 223 (page 43.) Bulik, T., Puerto-Gimenez, I., & CTA Collaboration. 2011, International Cosmic Ray Conference, 9, 223 (page 49.) Carrigan, S., Brun, F., Chaves, R. C. G., et al. 2013, ArXiv e-prints [arXiv:1307.4690] (page9.) Cerruti, M., Benbow, W., Chen, X., et al. 2017a, A&A, 606, A68 (page5.) Cerruti, M., Böttcher, M., Chakraborty, N., et al. 2017b, in American Institute of Physics Conference Series, Vol. 1792, 6th International Symposium on High Energy Gamma-Ray Astronomy, 050029 (page 24.) Cerruti, M., Lenain, J.-P., Prokoph, H., & for the H.E.S.S. Collaboration. 2017c, ArXiv e-prints [arXiv:1708.00658] (pages 24 and 27.) Chang, J., Ambrosi, G., An, Q., et al. 2017, Astroparticle Physics, 95, 6 (page 61.) Chang, S. & Goodenough, L. 2011, Phys. Rev. D, 84, 023524 (page 58.) Chaves, R. C. G. & for the H.E.S.S. Collaboration. 2009, ArXiv e-prints [arXiv:0907.0768] (page9.) Cholis, I., Dobler, G., Finkbeiner, D. P., Goodenough, L., & Weiner, N. 2009a, Phys. Rev. D, 80, 123518 (page 58.) Cholis, I., Finkbeiner, D. P., Goodenough, L., & Weiner, N. 2009b, J. Cosmology Astropart. Phys., 12, 007 (page 58.) Cholis, I., Goodenough, L., Hooper, D., Simet, M., & Weiner, N. 2009c, Phys. Rev. D, 80, 123511 (page 58.) Cholis, I. & Hooper, D. 2013, Phys. Rev. D, 88, 023013 (page 58.) Cholis, I. & Hooper, D. 2014, Phys. Rev. D, 89, 043013

71 Bibliography

(page 58.) Cholis, I., Hooper, D., & Linden, T. 2017, Phys. Rev. D, 95, 123007 (page 58.) Connolly, S. 2015, ArXiv e-prints [arXiv:1503.06676] (page 26.) Cordier, B., Wei, J., Atteia, J.-L., et al. 2015, ArXiv e-prints [arXiv:1512.03323] (page 65.) Cortina, J. 2012, The Astronomer’s Telegram, 4069 (page 22.) Costamante, L. & Ghisellini, G. 2002, A&A, 384, 56 (page5.) Couturier, C. 2014, Thèse de Doctorat, Université Pierre et Marie Curie; https://tel.archives-ouvertes.fr/tel-01219651 (page 22.) Cowen, D. F., Keivani, A., & Tešić, G. 2016, in European Physical Journal Web of Conferences, Vol. 116, European Physical Journal Web of Conferences, 10001 (pages 20 and 64.) CTA Consortium, Acharya, B. S., Actis, M., et al. 2013, Astroparticle Physics, 43, 3 (page 11.) CTA Consortium, Acharya, B. S., Agudo, I., et al. 2017, ArXiv e-prints [arXiv:1709.07997] (pages6, 11, and 21.) Cumani, P., Hassan, T., Arrabito, L., et al. 2017, ArXiv e-prints [arXiv:1709.00206] (page 49.) De Angelis, A., Tatischeﬀ, V., Grenier, I. A., et al. 2017, ArXiv e-prints [arXiv:1711.01265] (page 64.) de Naurois, M. & for the H.E.S.S. collaboration. 2017, The Astronomer’s Telegram, 10787 (page 64.) de Naurois, M. & Rolland, L. 2009, Astroparticle Physics, 32, 231 (pages 16, 33, 58, and 60.) Dubus, G., Contreras, J. L., Funk, S., et al. 2013, Astroparticle Physics, 43, 317 (pages6 and 11.) Ellis, J., Mavromatos, N. E., & Nanopoulos, D. V. 2008, Physics Letters B, 665, 412 (page 22.) Emmanoulopoulos, D., McHardy, I. M., & Papadakis, I. E. 2013, MNRAS, 433, 907 (page 26.) Fanaroﬀ, B. L. & Riley, J. M. 1974, MNRAS, 167, 31P (page3.) Farnier, C. & Lenain, J.-P. 2011, Performance studies for an altitude of 3700 m and 2000 m with Moonlight, CTA internal note, CTA consortium

72 Bibliography

(page 49.) Feng, Q., for the VERITAS Collaboration, Jorstad, S. G., et al. 2017, ArXiv e-prints [arXiv:1708.06386] (page 22.) Fossati, G., Maraschi, L., Celotti, A., Comastri, A., & Ghisellini, G. 1998, MNRAS, 299, 433 (page5.) Fox, P. J. & Poppitz, E. 2009, Phys. Rev. D, 79, 083528 (page 58.) Franceschini, A., Rodighiero, G., & Vaccari, M. 2008, A&A, 487, 837 (pages 18 and 29.) Funk, S., Hermann, G., Hinton, J., et al. 2004, Astroparticle Physics, 22, 285 (page9.) Funk, S., Hinton, J. A., & CTA Consortium. 2013, Astroparticle Physics, 43, 348 (pages 20 and 21.) Gajdus, M., Djannati-Ataï, A., Chrétien, M., et al. 2015, in International Cosmic Ray Conference, Vol. 34, 34th International Cosmic Ray Conference (ICRC2015), ed. A. S. Borisov, V. G. Denisova, Z. M. Guseva, E. A. Kanevskaya, M. G. Kogan, A. E. Morozov, V. S. Puchkov, S. E. Pyatovsky, G. P. Shoziyoev, M. D. Smirnova, A. V. Vargasov, V. I. Galkin, S. I. Nazarov, & R. A. Mukhamedshin, 841 (page 35.) Gallimore, J. F., Axon, D. J., O’Dea, C. P., Baum, S. A., & Pedlar, A. 2006, AJ, 132, 546 (page 14.) Gallimore, J. F., Baum, S. A., & O’Dea, C. P. 2004, ApJ, 613, 794 (page 14.) Garrigoux, T. 2015, Thèse de Doctorat, Université Pierre et Marie Curie; https://tel.archives-ouvertes.fr/tel-01208228 (pages 58 and 59.) Gast, H., Brun, F., Carrigan, S., et al. 2011, International Cosmic Ray Conference, 7, 158 (page9.) Gerard, L. & for the CTA Consortium. 2015, in International Cosmic Ray Conference, Vol. 34, 34th International Cosmic Ray Conference (ICRC2015), ed. A. S. Borisov, V. G. Denisova, Z. M. Guseva, E. A. Kanevskaya, M. G. Kogan, A. E. Morozov, V. S. Puchkov, S. E. Pyatovsky, G. P. Shoziyoev, M. D. Smirnova, A. V. Vargasov, V. I. Galkin, S. I. Nazarov, & R. A. Mukhamedshin, 725 (page 11.) Ghisellini, G. & Tavecchio, F. 2008, MNRAS, 387, 1669 (page5.) Ginzburg, V. L. & Syrovatskii, S. I. 1965, ARA&A, 3, 297 (page4.)

73 Bibliography

Giommi, P., Padovani, P., Polenta, G., et al. 2012, MNRAS, 420, 2899 (pages3 and5.) Grasso, D., Profumo, S., Strong, A. W., et al. 2009, Astroparticle Physics, 32, 140 (page 58.) Greisen, K. 1966, Physical Review Letters, 16, 748 (page7.) Guy, J. 2003, Thèse de Doctorat, Université Paris 6 - Pierre et Marie Curie; https://tel.archives-ouvertes.fr/tel-00003488 (pages9 and 32.) Harnik, R. & Kribs, G. D. 2009, Phys. Rev. D, 79, 095007 (page 58.) Hassan, T., Arrabito, L., Bernlör, K., et al. 2015, ArXiv e-prints [arXiv:1508.06075] (page 49.) Hauser, M. G. & Dwek, E. 2001, ARA&A, 39, 249 (page6.) Hayashida, M., Stawarz, Ł., Cheung, C. C., et al. 2013, ApJ, 779, 131 (page 14.) Heck, D., Knapp, J., Capdevielle, J. N., Schatz, G., & Thouw, T. 1998, CORSIKA: a Monte Carlo code to simulate extensive air showers. (CORSIKA: a Monte Carlo code to simulate extensive air showers., by Heck, D.; Knapp, J.; Capdevielle, J. N.; Schatz, G.; Thouw, T.. Forschungszentrum Karlsruhe GmbH, Karlsruhe (Germany)., Feb 1998, V + 90 p., TIB Hannover, D-30167 Hannover (Germany).) (pages 32 and 49.) H.E.S.S. Collaboration, Abdalla, H., Abramowski, A., et al. 2017a, A&A, 600, A89 (pages 16, 17, 18, 19, and 35.) H.E.S.S. Collaboration, Abdalla, H., Abramowski, A., et al. 2017b, ArXiv e-prints [arXiv:1702.08280] (page9.) H.E.S.S. Collaboration, Abdalla, H., Abramowski, A., et al. 2017c, ArXiv e-prints [arXiv:1706.04535] (pages9 and 58.) H.E.S.S. Collaboration, Abdalla, H., Abramowski, A., et al. 2017d, ArXiv e-prints [arXiv:1710.05862] (page 64.) H.E.S.S. Collaboration, Abramowski, A., Acero, F., et al. 2013, A&A, 559, A136 (page6.) H.E.S.S. Collaboration, Abramowski, A., Acero, F., et al. 2013a, A&A, 554, A107 (pages5 and 23.) H.E.S.S. Collaboration, Abramowski, A., Acero, F., et al. 2013b, A&A, 550, A4 (page6.) H.E.S.S. Collaboration, Abramowski, A., Acero, F., et al. 2012, A&A, 539, A149 (page 21.)

74 Bibliography

H.E.S.S. Collaboration, Abramowski, A., Aharonian, F., et al. 2014, A&A, 562, A145 (page 21.) H.E.S.S. Collaboration, Abramowski, A., Aharonian, F., et al. 2015, A&A, 573, A31 (page6.) H.E.S.S. Collaboration, Abramowski, A., Aharonian, F., et al. 2016, Nature, 531, 476 (page9.) Hillas, A. M. 1984, ARA&A, 22, 425 (pages ix,7, and8.) Hillas, A. M. 1985, in International Cosmic Ray Conference, Vol. 3, International Cosmic Ray Conference, ed. F. C. Jones, 445–448 (pages 33, 49, and 51.) Holler, M., Balzer, A., Chalmé-Calvet, R., et al. 2015a, ArXiv e-prints [arXiv:1509.02896] (page 16.) Holler, M., Berge, D., Hahn, J., et al. 2017a, ArXiv e-prints [arXiv:1707.04196] (pages 40 and 41.) Holler, M., Chevalier, J., Lenain, J.-P., de Naurois, M., & Sanchez, D. 2017b, ArXiv e-prints [arXiv:1711.01118] (pages 37, 39, and 40.) Holler, M., de Naurois, M., Zaborov, D., Balzer, A., & Chalmé-Calvet, R. 2015b, in International Cosmic Ray Conference, Vol. 34, 34th International Cosmic Ray Conference (ICRC2015), ed. A. S. Borisov, V. G. Denisova, Z. M. Guseva, E. A. Kanevskaya, M. G. Kogan, A. E. Morozov, V. S. Puchkov, S. E. Pyatovsky, G. P. Shoziyoev, M. D. Smirnova, A. V. Vargasov, V. I. Galkin, S. I. Nazarov, & R. A. Mukhamedshin, 980 (pages 33 and 34.) Hooper, D., Blasi, P., & Dario Serpico, P. 2009, J. Cosmology Astropart. Phys., 1, 025 (page 58.) Hooper, D., Cholis, I., Linden, T., & Fang, K. 2017, Phys. Rev. D, 96, 103013 (page 61.) Hooper, D. & Linden, T. 2017, ArXiv e-prints [arXiv:1711.07482] (pages 61 and 62.) IceCube Collaboration. 2013, Science, 342 [arXiv:1311.5238] (page 64.) Jaﬀe, W., Meisenheimer, K., Röttgering, H. J. A., et al. 2004, Nature, 429, 47 (pages 12 and 14.) Kachelrieß, M. & Ostapchenko, S. 2013, Phys. Rev. D, 87, 047301 (page 58.) Kachelrieß, M., Ostapchenko, S., & Tomàs, R. 2011, ApJ, 733, 119 (page 58.) Kantor, J. 2014, in The Third Hot-wiring the Transient Universe Workshop, ed. P. R.

75 Bibliography

Wozniak, M. J. Graham, A. A. Mahabal, & R. Seaman, 19–26 (page 65.) Katarzyński, K., Sol, H., & Kus, A. 2001, A&A, 367, 809 (page 14.) Kaur, N., Chandra, S., Baliyan, K. S., Sameer, & Ganesh, S. 2017, ArXiv e-prints [arXiv:1706.04411] (page 22.) Kennel, C. F. & Coroniti, F. V. 1984, ApJ, 283, 710 (page 41.) Kerszberg, D. 2017, Thèse de Doctorat, Université Pierre et Marie Curie; https://tel.archives-ouvertes.fr/tel-01682782 (pages 57, 58, and 60.) Kertzman, M. P. & Sembroski, G. H. 1994, Nuclear Instruments and Methods in Physics Research A, 343, 629 (page 32.) Klepser, S., Ashton, T., Backes, M., et al. 2017, ArXiv e-prints [arXiv:1707.04415] (page 43.) Kobayashi, T., Komori, Y., Yoshida, K., & Nishimura, J. 2004, ApJ, 601, 340 (page 61.) Konigl, A. 1981, ApJ, 243, 700 (page4.) Kotera, K. 2014, Habilitation à diriger des recherches, Université Pierre et Marie Curie (pages7 and8.) Krawczynski, H., Hughes, S. B., Horan, D., et al. 2004, ApJ, 601, 151 (page 21.) Kruskal, M., Ahlen, S. P., & Tarlé, G. 2016, ApJ, 818, 70 (page 58.) Kusunose, M. & Takahara, F. 2006, ApJ, 651, 113 (page 21.) Lacki, B. C., Horiuchi, S., & Beacom, J. F. 2014, ApJ, 786, 40 (page 14.) Lamastra, A., Fiore, F., Guetta, D., et al. 2016, A&A, 596, A68 (page 15.) Le Bohec, S., Degrange, B., Punch, M., et al. 1998, Nuclear Instruments and Methods in Physics Research A, 416, 425 (page 59.) Lenain, J.-P. 2009, Thèse de Doctorat, LUTH, Observatoire de Paris, CNRS, Université Paris Diderot; https://tel.archives-ouvertes.fr/tel-00431288 (pages 14, 21, and 36.) Lenain, J.-P. 2017a, ArXiv e-prints [arXiv:1709.04065] (pages 25 and 26.) Lenain, J.-P. 2017b, FLaapLUC: Fermi-LAT automatic aperture photometry light

76 Bibliography

curve, Astrophysics Source Code Library (page 25.) Lenain, J.-P., Boisson, C., Sol, H., & Katarzyński, K. 2008, A&A, 478, 111 (pages 14 and 25.) Lenain, J.-P. & de Naurois, M. 2013, in Journées SUCCES 2013 (page 37.) Lenain, J.-P., Ricci, C., Türler, M., Dorner, D., & Walter, R. 2010, A&A, 524, A72+ (pages 12, 13, 14, and 15.) Lenain, J.-P. & Walter, R. 2011, A&A, 535, A19 (pages 12 and 14.) Leroy, N. 2004, Thèse de Doctorat, Ecole Polytechnique; https://tel.archives- ouvertes.fr/tel-00117166 (page 19.) Lester, D. F., Joy, M., Harvey, P. M., Ellis, Jr., H. B., & Parmar, P. S. 1987, ApJ, 321, 755 (pages 12 and 14.) Li, T.-P. & Ma, Y.-Q. 1983, ApJ, 272, 317 (page 51.) Liberati, S. 2013, Classical and Quantum Gravity, 30, 133001 (page 22.) Linden, T. & Profumo, S. 2013, ApJ, 772, 18 (page 58.) MAGIC Collaboration, Albert, J., Aliu, E., et al. 2008, Science, 320, 1752 (pages5 and6.) Maier, G., Arrabito, L., Bernlöhr, K., et al. 2017, ArXiv e-prints [arXiv:1709.01381] (page 49.) Maier, G., Arrabito, L., Bernlöhr, K., et al. 2015, ArXiv e-prints [arXiv:1508.06042] (page 49.) Malyshev, D., Cholis, I., & Gelfand, J. 2009, Phys. Rev. D, 80, 063005 (page 58.) Mannheim, K., Biermann, P. L., & Kruells, W. M. 1991, A&A, 251, 723 (page5.) Maraschi, L., Foschini, L., Ghisellini, G., Tavecchio, F., & Sambruna, R. M. 2008, MNRAS, 391, 1981 (page5.) Martínez, M. & Errando, M. 2009, Astroparticle Physics, 31, 226 (page 22.) Massaro, F., Thompson, D. J., & Ferrara, E. C. 2015, A&A Rev., 24, 2 (page 14.) Matt, G., Bianchi, S., Guainazzi, M., & Molendi, S. 2004, A&A, 414, 155 (page 12.) Medina, M. C. & Zech, A. 2010, CTA Design Study - Working Package MC Final

77 Bibliography

report from the SHOWER subtask, CTA internal note, CTA consortium (page 36.) Mertsch, P. & Sarkar, S. 2009, Physical Review Letters, 103, 081104 (page 58.) Meyer, E. T., Fossati, G., Georganopoulos, M., & Lister, M. L. 2011, ApJ, 740, 98 (page5.) Mirzoyan, R. & for the MAGIC collaboration. 2017, The Astronomer’s Telegram, 10817 (page 64.) Moralejo, A., Domínguez, A., Fallah Ramazani, V., et al. 2017, ArXiv e-prints [arXiv:1709.02238] (page6.) Mücke, A. & Protheroe, R. J. 2001, Astroparticle Physics, 15, 121 (page5.) Mukherjee, R. & for the VERITAS collaboration. 2017, The Astronomer’s Telegram, 10833 (page 64.) Nalewajko, K. 2013, MNRAS, 430, 1324 (page 25.) Nelson, A. E. & Spitzer, C. 2010, Journal of High Energy Physics, 10, 66 (page 58.) Neronov, A. & Vovk, I. 2010, Science, 328, 73 (page 21.) Nieppola, E., Valtaoja, E., Tornikoski, M., Hovatta, T., & Kotiranta, M. 2008, A&A, 488, 867 (page5.) Oikonomou, F., Murase, K., & Kotera, K. 2014, A&A, 568, A110 (page 21.) Oya, I., Fuessling, M., Schwanke, U., et al. 2015, ArXiv e-prints [arXiv:1509.01164] (page 11.) Padovani, P. 2007, Ap&SS, 309, 63 (page5.) Padovani, P. 2017, Nature Astronomy, 1, 0194 (pages2 and3.) Padovani, P., Alexander, D. M., Assef, R. J., et al. 2017, A&A Rev., 25, #2 (pages2 and3.) Padovani, P., Giommi, P., & Rau, A. 2012, MNRAS, 422, L48 (page5.) Parsons, R. D. & Hinton, J. A. 2014, Astroparticle Physics, 56, 26 (page 33.) Pavlidou, V. & Fields, B. D. 2001, ApJ, 558, 63 (pages 12 and 14.)

78 Bibliography

Pelletier, G., Bykov, A., Ellison, D., & Lemoine, M. 2017, ArXiv e-prints [arXiv:1705.05549] (page 63.) Peng, F.-K., Wang, X.-Y., Liu, R.-Y., Tang, Q.-W., & Wang, J.-F. 2016, ApJ, 821, L20 (page 14.) Persic, M. & Rephaeli, Y. 2014, A&A, 567, A101 (page 14.) Pita, S., Goldoni, P., Boisson, C., et al. 2014, A&A, 565, A12 (page6.) Pospelov, M. & Ritz, A. 2009, Physics Letters B, 671, 391 (page 58.) Profumo, S. 2012, Central European Journal of Physics, 10, 1 (page 58.) Ptuskin, V., Rogovaya, S., & Zirakashvili, V. 2013, Advances in Space Research, 51, 315 (page 63.) Pueschel, E. 2017, in American Institute of Physics Conference Series, Vol. 1792, 6th International Symposium on High Energy Gamma-Ray Astronomy, 060002 (page6.) Rolland, L. 2005, Thèse de Doctorat, Université Paris 6 - Pierre et Marie Curie; https://tel.archives-ouvertes.fr/tel-00010019 (pages9, 41, and 42.) Romoli, C., Zacharias, M., Meyer, M., et al. 2017, ArXiv e-prints [arXiv:1708.00882] (page 24.) Sahu, S., Oliveros, A. F. O., & Sanabria, J. C. 2013, Phys. Rev. D, 87, 103015 (page 21.) Saito, S., Stawarz, Ł., Tanaka, Y. T., et al. 2015, ApJ, 809, 171 (page 23.) Sanchez, D. A., Brun, F., Couturier, C., et al. 2015, ArXiv e-prints [arXiv:1502.02920] (page 22.) Santander, M. & for the VERITAS Collaboration. 2017, ArXiv e-prints [arXiv:1709.02365] (page 22.) Schüssler, F., Backes, M., Balzer, A., et al. 2017, ArXiv e-prints [arXiv:1708.00466] (pages 21 and 64.) Schüssler, F., Balzer, A., Brun, F., et al. 2015, ArXiv e-prints [arXiv:1509.03035] (page 21.) Sironi, L., Keshet, U., & Lemoine, M. 2015, Space Sci. Rev., 191, 519 (page 63.) Smith, M. W. E., Fox, D. B., Cowen, D. F., et al. 2013, Astroparticle Physics, 45, 56

79 Bibliography

(page 64.) Sol, H., Zech, A., Boisson, C., et al. 2013, Astroparticle Physics, 43, 215 (page6.) Staszak, D. & VERITAS Collaboration. 2015, in International Cosmic Ray Conference, Vol. 34, 34th International Cosmic Ray Conference (ICRC2015), 411 (pages 58 and 61.) Takami, H. & Horiuchi, S. 2011, Astroparticle Physics, 34, 749 (page 63.) Tanaka, Y. T., Buson, S., Kocevski, D., & for the Fermi-LAT collaboration. 2017, The Astronomer’s Telegram, 10791 (page 64.) Thronson, Jr., H. A., Hereld, M., Majewski, S., et al. 1989, ApJ, 343, 158 (page 12.) Tsaregorodtsev, A., Bargiotti, M., Brook, N., et al. 2008, in Journal of Physics Conference Series, Vol. 119, Journal of Physics Conference Series, 062048 (page 37.) Tsujimoto, S., Vazquez Acosta, M., Lindfors, E., et al. 2017, ArXiv e-prints [arXiv:1709.05063] (page 22.) Urry, C. M. & Padovani, P. 1995, PASP, 107, 803 (page2.) van Eldik, C., Holler, M., Berge, D., et al. 2015, in International Cosmic Ray Conference, Vol. 34, 34th International Cosmic Ray Conference (ICRC2015), ed. A. S. Borisov, V. G. Denisova, Z. M. Guseva, E. A. Kanevskaya, M. G. Kogan, A. E. Morozov, V. S. Puchkov, S. E. Pyatovsky, G. P. Shoziyoev, M. D. Smirnova, A. V. Vargasov, V. I. Galkin, S. I. Nazarov, & R. A. Mukhamedshin, 847 (pages 34 and 35.) VERITAS Collaboration, Acciari, V. A., Aliu, E., et al. 2009, Nature, 462, 770 (page 12.) Vernet, J., Dekker, H., D’Odorico, S., et al. 2011, A&A, 536, A105 (page6.) Wojaczyński, R., Niedźwiecki, A., Xie, F.-G., & Szanecki, M. 2015, A&A, 584, A20 (page 15.) Yoast-Hull, T. M., Gallagher, III, J. S., Zweibel, E. G., & Everett, J. E. 2014, ApJ, 780, 137 (page 15.) Yüksel, H., Kistler, M. D., & Stanev, T. 2009, Physical Review Letters, 103, 051101 (page 58.) Zaborov, D., Romoli, C., Taylor, A., et al. 2015, in International Cosmic Ray Conference, Vol. 34, 34th International Cosmic Ray Conference (ICRC2015), ed. A. S. Borisov, V. G. Denisova, Z. M. Guseva, E. A. Kanevskaya, M. G. Kogan, A. E. Morozov, V. S. Puchkov, S. E. Pyatovsky, G. P. Shoziyoev, M. D. Smirnova, A. V. Vargasov,

80 Bibliography

V. I. Galkin, S. I. Nazarov, & R. A. Mukhamedshin, 808 (page 34.) Zacharias, M., Böttcher, M., Jankowsky, F., et al. 2017a, ArXiv e-prints [arXiv:1711.06117] (pages 22 and 23.) Zacharias, M., Jankowsky, F., Mohamed, M., et al. 2017b, ArXiv e-prints [arXiv:1708.00623] (page 23.) Zacharias, M., Sitarek, J., Dominis Prester, D., et al. 2017c, ArXiv e-prints [arXiv:1708.00653] (page 23.) Zatsepin, G. T. & Kuz’min, V. A. 1966, Soviet Journal of Experimental and Theoretical Physics Letters, 4, 78 (page7.) Zech, A., Cerruti, M., & Mazin, D. 2017, A&A, 602, A25 (page 64.)

Appendices

Selected publications

1. FLaapLUC: A pipeline for the generation of prompt alerts on transient Fermi-LAT γ-ray sources

2. Cloud ablation by a relativistic jet and the extended ﬂare in CTA 102 in 2016 and 2017

3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 −

4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

5. Seyfert 2 galaxies in the GeV band: jets and starburst

85 A. Selected publications

Astronomy and Computing 22 (2018) 9–15

Contents lists available at ScienceDirect

Astronomy and Computing

journal homepage: www.elsevier.com/locate/ascom

Full length article FLaapLUC: A pipeline for the generation of prompt alerts on transient Fermi-LAT γ -ray sources✩ J.-P. Lenain Sorbonne Universités, UPMC Université Paris 06, Université Paris Diderot, Sorbonne Paris Cité, CNRS, Laboratoire de Physique Nucléaire et de Hautes Energies (LPNHE), 4 place Jussieu, F-75252, Paris Cedex 5, France article info a b s t r a c t

Article history: The large majority of high energy sources detected with Fermi-LAT are blazars, which are known to Received 31 July 2017 be very variable sources. High cadence long-term monitoring simultaneously at different wavelengths Accepted 21 November 2017 being prohibitive, the study of their transient activities can help shedding light on our understanding Available online 28 November 2017 of these objects. The early detection of such potentially fast transient events is the key for triggering follow-up observations at other wavelengths. A tool, , built on top of the Science Tools Keywords: Python FLaapLUC Methods: data analysis provided by the Fermi Science Support Center and the Fermi-LAT collaboration, has been developed using Methods: numerical a simple aperture photometry approach. This tool can effectively detect relative flux variations in a set of Gamma rays: general predefined sources and alert potential users. Such alerts can then be used to trigger target of opportunity Galaxies: active observations with other facilities. It is shown that FLaapLUC is an efficient tool to reveal transient events in Fermi-LAT data, providing quick results which can be used to promptly organise follow-up observations. Results from this simple aperture photometry method are also compared to full likelihood analyses. The FLaapLUC package is made available on GitHub and is open to contributions by the community. © 2017 Elsevier B.V. All rights reserved.

1. Introduction useful to monitor the whole sky in the high energy range (from 20 MeV up to 300 GeV and above), with full sky snapshots obtained The sky is not as immutable and quiet as it first seems when every three hours. In the case of transient events, a prompt reaction seen with the naked eye. Once studied in detail with sensitive to trigger multiwavelength observations is essential. instruments, variable sources are detected at all wavelengths and The Fermi-LAT collaboration developed the FAVA (Fermi All-sky on different time scales. This notably applies at high energies, and Variability Analysis) tool (Ackermann et al., 2013), which has the more particularly to non-thermal emission from sources such as huge advantage of blindly searching for transient events all across active galactic nuclei (AGN), pulsars, binaries, micro-quasars or the sky, but the latency time of about one week before releasing the cataclysmic variables. data1 prevents using it for prompt, quick alerts and subsequent ob- The Fermi-LAT γ -ray instrument (Atwood et al., 2009) has re- servations before a flare subsides. The typical duration of flares in vealed many new high energy sources (Abdo et al., 2010a; Nolan blazars is indeed shorter than a week (see e.g. H.E.S.S. Collaboration et al., 2012; Acero et al., 2015), many of which are constituted et al., 2013a,b; Wehrle et al., 2016; Ackermann et al., 2016b; Britto by AGN (Abdo et al., 2010b; Ackermann et al., 2011, 2015). AGN et al., 2016; Rani et al., 2017; Abeysekara et al., 2017). The Fermi- are highly variable by nature, and a quick identification of any LAT collaboration also provides to the community a set of light ongoing unusual activity in these sources is crucial to ensure multi- curves, updated daily, on a list of bright sources,2 and the Fermi wavelength follow-up observations, to better characterise the na- Science Support Center makes available aperture photometry light ture of their emission. curves3 for all sources belonging to the 3FGL catalogue (Acero et A full picture of the behaviour of AGN could in principle be obtained with simultaneous, high-cadence monitoring at all avail- al., 2015). However, for sources not reported in the 3FGL catalogue able wavelengths. However, such observational campaigns are and/or if one wants to consider light curves with a different time hardly practically achievable on long time scales. Instead, a grasp of binning, a custom pipeline is necessary. knowledge can be picked up during flaring events, if several facilities follow up on the flare simultaneously. Fermi-LAT is particularly 1 fermi.gsfc.nasa.gov/ssc/data/access/lat/FAVA. 2 fermi.gsfc.nasa.gov/ssc/data/access/lat/msl_lc. ✩ This code is registered at the ASCL with the code entry ascl:1709.011. 3 fermi.gsfc.nasa.gov/ssc/data/access/lat/4yr_catalog/ap_lcs and https://fermi. E-mail address: [email protected]. gsfc.nasa.gov/ssc/data/analysis/user/aperture.pl. https://doi.org/10.1016/j.ascom.2017.11.002 2213-1337/© 2017 Elsevier B.V. All rights reserved.

86 A.1. FLaapLUC: A pipeline for the generation of prompt alerts on transient

10 J.-P. Lenain / Astronomy and Computing 22 (2018) 9–15 Fermi-LAT γ-ray sources

Following major interests by the H.E.S.S. collaboration in tar- photon index of 2.5 is assumed. During the process of light curve get of opportunity (ToO) observations, FLaapLUC (Fermi-LAT computation with FLaapLUC, the photon index of the source of automatic aperture photometry Light C↔Urve) has thus been de- interest is thus set constant, with a value either corresponding to veloped, built in Python on top of the Science Tools provided by the 3FGL catalogue or fixed to 2.5. No potential spectral temporal the Fermi Science Support Center and the Fermi-LAT collaboration, variation is considered. and based on the simple aperture photometry technique (Lenain, FLaapLUC can then be used to generate triggers in two different 2017). FLaapLUC has been extensively tried and tested within the ways: H.E.S.S. collaboration. This tool is able to quickly analyse a predefined list of sources and automatically send alerts within H.E.S.S. 1. One can manually define a fixed flux threshold, on a source in the case of sufficiently bright events occurring at high energies. by source basis; This allows the H.E.S.S. collaboration to promptly trigger follow-up 2. Alternatively, if a long-term light curve using all the avail- ToO observations. Moreover, an advantage of developing a custom able Fermi-LAT data has been pre-computed, FLaapLUC can tool such as FLaapLUC resides in the fact that the trigger criteria dynamically assess the flux threshold above which a source are under full control. activity will generate a trigger (see below). In the following, the method used for the Fermi-LAT data anal- In the latter case, such a pre-computed long-term light curve ysis with FLaapLUC is described in Section2. The performance, can typically be generated using the following command: such as its false alarm trigger rate, and comparisons with classic full likelihood results, is discussed in Section3, before concluding flaapluc --merge-long-term in Section4. --config-file=config/ 2. Description of the method where config_file.cfg is a configuration file where several FLaapLUC uses the aperture photometry approach4 to analyse options can be set. An example configuration file is provided in Fermi-LAT data. The aperture photometry is a simple method con- FLaapLUC/config on GitHub. sisting in summing up photon counts in a region of interest, and The main difficulty in the procedure resides in the definition weighting the results by the instrumental exposure to evaluate a of a flare, while the event is actually ongoing. Specifically, how flux. No background modelling or subtraction is performed. The to dynamically compute such a flux threshold with respect to goal of FLaapLUC is to provide alerts on ongoing activities in the the long-term average flux level. Even once the entire data set is Fermi-LAT data from a predefined list of sources being monitored. available, the definition of what a flare is, is subject to debate. For Indeed, contrary to FAVA, to keep computing resources at a rea- instance, Nalewajko(2013) proposes that a flare consists in finding sonable usage, a blind search of transient events across the full sky the peak in a light curve and to consider the contemporaneous using aperture photometry is not performed. temporal window when the flux is at least half of the peak flux. The implementation of FLaapLUC is based on the standard Here, one cannot use such a definition, which requires having the Science Tools. gtselect is used to extract the events around a full observation data set at hand. Indeed, the aim of FLaapLUC is to source of interest, within a radius of 1◦ in the case of aperture alert for an ongoing event, without knowing whether the last flux photometry. This is because this method assumes that the data set measurement still corresponds to a trend of rising flux, or whether is background-free. This assumption is of course wrong in the case the flare is already on its decay. Instead, the following approach of Fermi-LAT data which are highly contaminated by Galactic and is proposed. For a given source, a weekly-binned long-term light extragalactic diffuse emission. However, these diffuse components curve is pre-computed, thus currently using more than 9 years are not supposed to vary, and their presence will not impede of Fermi-LAT data. To assess whether the source is experiencing detecting relative flux variations. The considered energy range is an active state, another light curve is computed every day, using 100 MeV–500 GeV, and a cut on the maximal zenith angle of bins of N1 days of duration, and the last flux bin measurement 90◦ is applied, as recommended by the Fermi-LAT collaboration is compared to the long-term flux mean. If the last flux bin is for point-source (event class 128) analyses using the Pass 8 in- significantly higher than the average flux, FLaapLUC identifies strument response functions.5 Good time intervals are selected the current flux state as active. Following this, another, finer, light using gtmktime using the standard filter (LAT_CONFIG==1 && curve is generated with bins of N2 days (with N2 < N1). If the new DATA_QUAL>0). Additionally, only time intervals during which more finely binned flux is significantly above the long-term flux the Sun is at least 5◦ away from the region of interest are kept, mean, FLaapLUC issues an alert. so as to avoid contamination by potential solar flares. The Moon More quantitatively, and accounting for flux errors, the aver- being a bright γ -ray emitter as well (Ackermann et al., 2016a), aged flux (FLT ) on long-term data is computed. Let us denote FN1,2 ◦ data when the Moon is closer than 5 from the region of interest the last N1,2-days binned flux value, and δFN1,2 its error. A two-level could also be filtered out as recommended by Corbet et al.(2013), criterion on the flux is set, based on the N1-days binned and N2-day if the input spacecraft file has been previously processed with the binned light curves. The trigger threshold on the flux is such that: moonpos script.6 This last procedure is left to the discretion of the FN − δFN > FLT + αN RMS(FLT ) user. The evolution of the count rate is then computed using the LC 1 1 1 F − δF > F + α RMS(F ) (1) method of gtbin. To correct for the time-dependent exposure on a N2 N2 LT N2 LT source with gtexposure and obtain a flux, a model of the source of To speed the processing of potentially many sources every day, interest has to be provided. To this end, the user-contributed script the N2-day binned light curve is only computed if the first criterion make3FGLxml is used, thus accounting for all sources included in on FN is fulfilled. The trigger threshold, and thus probability (see the 3FGL catalogue (Acero et al., 2015) near the source of interest. 1 Section 3.2), then depends on the settings on αN1 and αN2 , which If the source of interest does not belong to the 3FGL catalogue, a are chosen by the user so as to tune the alert rate for a particular source class. The chosen value of N2 thus limits the minimum time 4 fermi.gsfc.nasa.gov/ssc/data/analysis/scitools/aperture_photometry. scale of a flare FLaapLUC can probe in Fermi-LAT data. Prompt 5 fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Cicerone/Cicerone_ alerts are further limited by the latency time required to downlink Data_Exploration/Data_preparation. the data and reconstruct them. This will be further discussed in 6 https://fermi.gsfc.nasa.gov/ssc/data/analysis/user/moonpos-1.1.tgz. Section 3.3. In any case, it is unfeasible to react on the fly to rapid

87 A. Selected publications

J.-P. Lenain / Astronomy and Computing 22 (2018) 9–15 11

(here usually the most permissive zenith angle at culmination is considered). FLaapLUC takes photon files from Fermi-LAT data as input. The pipeline can be run using e.g. an all-sky photon file encompassing all the γ -ray-like events recorded with Fermi LAT for the whole mission. This is necessary in order to pre-compute a long-term light curve for multiple sources at once. Alternatively, for a daily running, one can use an all-sky file from a subset of the last data acquired with Fermi-LAT to speed up the computation and limit the input/output usage in case many sources are to be processed. A roll back time of typically one month is used by the H.E.S.S. collaboration. Such input files can easily be generated, as well as an automatic retrieval of spacecraft and photon files, using for instance enrico (Sanchez and Deil, 2013). FLaapLUC is actually using enrico to generate those input files on the fly in case the user does not provide them. The daily running instance of FLaapLUC is typically run using the following command: flaapluc-allsources --daily --custom-threshold Fig. 1. Two-dimensional criterion applied on the redshift and the zenith angle --with-history at culmination of a source to accept or veto a latent alert from FLaapLUC. These specific cut values are used in H.E.S.S. for the Fermi-LAT data analysis of extragalactic --config-file=config/ sources. where flaapluc-allsources is just a wrapper of the script flaapluc looping through all the sources listed in the input file. events such as the giant outburst detected from 3C 279 in June Within the H.E.S.S. collaboration, whenever FLaapLUC issues 2015 (Ackermann et al., 2016b), which exhibited doubling times an alert, with the last bit of data on a monitored source fulfilling the double-pass threshold from Eq. (1), as well as constraints on of less than 5 min, or the equivalent at high energies of the very visibility, redshift and zenith angle at culmination, a more detailed high energy flares of PKS 2155−304 seen in 2006 (Aharonian et al., likelihood analysis is automatically performed on the last N days 2007) or Mrk 501 as observed in 2005 (Albert et al., 2007) which 1 of available data. This procedure leads to a computational economy varied on similar time scales. with respect to a scheme where all monitored sources would have Moreover, if one wants to monitor the high energy sky in order been analysed with the full likelihood approach. to trigger ToO observations at a particular site, it is useful to FLaapLUC has been used in H.E.S.S. since 2012 success- additionally filter alerts on the visibility of the sources in the next fully, having given rise to quick reaction follow-ups, such as in hours/days. FLaapLUC can thus perform such a filtering, depend- February 2015 on PKS 0736+01 (see Fig. 2) which resulted in ing on the visibility of a source at a given site and observation time, the detection of this source during this event with H.E.S.S. at using the pyephem package.7 As an example, the common set of very high energies (Cerruti et al., 2016). FLaapLUC issued an trigger criteria used within the H.E.S.S. extragalactic working group alert even before public information was available on this flare. is the following: Fig. 3 shows the energy versus arrival time for each event from the alert on PKS 0736+01. The colours depict the density of • the source should have its last flux measurement fulfilling events in the data with a Gaussian kernel-density estimate using the criteria described in Eq. (1), with N1 = 3 days, N2 = 1 scipy.stats.gaussian_kde. This also allows an assessment of = = the energy of the highest energy photon received during a flaring day, αN1 2 and αN2 3; • the source should be visible the next night at the H.E.S.S. event. Such information can be useful for deciding whether or site (Lon. 23◦16′18′′S, Lat. 16◦30′00′′E), and its zenith angle not ToO observations should be triggered at higher energies, with at culmination should be less than a certain value which e.g. the H.E.S.S. experiment. depends on the redshift of the source, due to the absorption Apart from AGN, FLaapLUC is also used internally in H.E.S.S. to produce alerts on a predefined list of -ray binaries or binary can- by the extragalactic background light (Hauser and Dwek, γ didates, in this case using different criteria on the flux thresholds 2001) of the observed source spectrum at very high energies. and observability, with N1 = 2 days, N2 = 1 day, αN = 2, αN = 3 1 2 ◦ The reasoning behind the last criterion is the following. The and a fixed maximum allowed zenith angle at culmination of 60 . As a third application, a systematic survey of the Galactic plane very high energy γ -ray photons experience absorption on their is performed daily at high energies with FLaapLUC, with a scan propagation path due to the extragalactic background light (Hauser of 540 regions of 1◦ of radius, in the Galactic latitude band |b| < and Dwek, 2001). This absorption depends on the photon energy, ◦ 3 . Again, even though the Galactic plane is largely dominated by and is more severe at the highest energies which imaging at- the Galactic diffuse emission which thus hampers any absolute mospheric Čerenkov telescopes (IACT) are sensitive to. Since the flux determination with the aperture photometry method, any energy threshold of IACT also increases with the observation zenith significant relative flux variation could be detected with this tool. angle, for a similar flux, further away sources should be observed For this application, the trigger criteria are N1 = 2 days, N2 = 1 at smaller zenith angles (i.e. higher elevation) than closer ones to = = day, αN1 3, αN2 5 and a fixed maximum allowed zenith angle reach the same detection probability. This last cut is modular and at culmination of 60◦. programmable. It can be implemented as a simple scalar value on the maximal acceptable zenith angle and/or redshift, or can 3. Performance be mapped as a two-dimensional criterion, as depicted in Fig. 1. For sources whose redshift is unknown, a value of z = 0 is used The performance and limitations of FLaapLUC are hereafter developed. Table 1 gives a summary of the main points discussed in this paper, as well as the operational settings used by H.E.S.S. as 7 rhodesmill.org/pyephem. mentioned above.

88 A.1. FLaapLUC: A pipeline for the generation of prompt alerts on transient

12 J.-P. Lenain / Astronomy and Computing 22 (2018) 9–15 Fermi-LAT γ-ray sources

Table 1 Summary of the operational settings, performances and limitations of FLaapLUC for the applications in use in H.E.S.S. Extragalactic sources Binary candidates Galactic plane survey

N1 3 days 2 days 2 days

αN1 2 2 3 N2 1 day 1 day 1 day

αN2 3 3 5 False alarm probability < 0.3% < 0.3% < 0.05% Minimum time scale probed ∼ 1 day ∼ 1 day ∼ 1 day

procedure consisting in attributing all photons from an analysed region to a given source of interest. In this section, the light curve obtained using FLaapLUC is compared to the one computed using the binned likelihood scheme, for the same object. Data obtained from August 4, 2008 to July 4, 2017 are analysed, for two sources, 3C 279 (a flat spectrum radio quasar, FSRQ) and PKS 2155−304 (a high-frequency-peaked BL Lac object, HBL), as an illustration. For the likelihood analyses, events in a region of interest of 10◦ radius were selected. The PASS 8 instrument response functions (event class 128 and event type 3) corresponding to the P8R2_SOURCE_V6 response were used together with a zenith angle cut of 90◦. The model of the region of interest was built based on the 3FGL catalogue (Acero et al., 2015). The Galactic diffuse emission has been modelled using the file gll_iem_v06. fits (Acero et al., 2016) and the isotropic background using iso_ P8R2_SOURCE_V6_v06.txt. The fit is performed iteratively: in a first step, sources from the 3FGL catalogue within 15◦ around the source of interest are included, with parameters fixed for Fig. 2. FLaapLUC light curve output on PKS 0736+01 issued on Feb. 18, 2015. ◦ The blue points show the 3-day binned light curve, while the red points show the those more than 10 away to account for the large point spread daily-binned one. The horizontal blue line represents the long-term flux average of function at low energies. In a second step, parameters of sources the source, and the horizontal dotted blue lines are the flux levels plus or minus contributing to less than a test statistic (TS, Mattox et al., 1996) of · = 2 RMS(FLT ) away from this average (αN1 2). The horizontal bold red line shows = 9 and to less than 5% of the total number of counts in the region the flux threshold for αN2 3 above which FLaapLUC issues an alert on this source. (For interpretation of the references to colour in this figure legend, the reader is of interest are frozen. In a third step, the only free parameters are referred to the web version of this article.) those of sources less than 3◦ away from our source of interest (if not frozen in the previous step), the source of interest itself, and the normalisations of the Galactic and isotropic diffuse emissions. In the different steps, the spectral parameters (photon and curvature indices, since both PKS 2155−304 and 3C 279 are described with log-parabolic spectra in the 3FGL catalogue) are fixed to the catalogue values. This is to ensure a proper comparison with FLaapLUC results, since the latter does not account for potential spectral evolution as a function of time. The results of the likelihood analyses of 3C 279 and PKS 2155−304 are shown in Figs. 4 and5 respectively. Weekly-binned light curves are shown in the top panel for both FLaapLUC results and the likelihood analysis. The middle left panel represents the relative error between the two analysis methods and the middle right panel displays the distribution of this error. FLaapLUC sys- tematically overshoots the resulting flux compared with a proper likelihood analysis, which is inherent to the original assumption of a data set free of any background (see middle right panel in Fig. 4). This is especially the case for low fluxes (e.g. for 3C 279, see

Fig. 3. Energy versus arrival time plot on PKS 0736+01 of each individual Fermi- middle left panel in Fig. 4, e.g. around MJD 55300 or MJD 56000– LAT event as of Feb. 18, 2015. The colour code depicts a simple Gaussian kernel- 56300) where the contribution from the diffuse emission compo- density estimate for visualisation purposes. Qualitatively, a vertical clustering of nents (Galactic and extragalactic) is not negligible at all. The error yellow points would denote a flare. (For interpretation of the references to colour distributions show that the aperture photometry overestimates in this figure legend, the reader is referred to the web version of this article.) the fluxes by ∼ 30–50% on average, and up to a factor ∼ 2 in case of low activity. However, it can be seen from Fig. 4 and5 that the 3.1. Comparison with the likelihood method global trends of the light curves are well reproduced in the aperture photometry results compared to the likelihood ones. This is further The aperture photometry method provides a fast way to ob- strengthened in the bottom panels, which show a comparison of tain relative results, but is obviously not the best choice when it the FLaapLUC and the likelihood results once debiased from their comes to reliable, absolute flux measurements, because of the basic respective average.

89 A. Selected publications

J.-P. Lenain / Astronomy and Computing 22 (2018) 9–15 13

Fig. 4. Top: Example of a long-term light curve computed with FLaapLUC for 3C 279 in blue, and the same computed with the binned likelihood approach in red. For instance, the June 2015 flare is well visible. Middle left: Relative errors of the aperture photometry analysis with respect to the likelihood results. Middle right: Distribution of the relative errors. Bottom left: Ratio between the aperture photometry and the likelihood results, once debiased from their respective average. Bottom right: Corresponding distribution.

Fig. 5. Same as Fig. 4 for PKS 2155−304.

90 A.1. FLaapLUC: A pipeline for the generation of prompt alerts on transient

14 J.-P. Lenain / Astronomy and Computing 22 (2018) 9–15 Fermi-LAT γ-ray sources

transfer the data from the Fermi spacecraft to the ground, to digest them and then retrieve photon and spacecraft pointing files from the NASA servers, to generate an all-sky file which is used as input for FLaapLUC, and to analyse all the monitored sources, the total latency time of the process, i.e. the delay between the last bit of data and the generation of an alert with FLaapLUC, is about 8 h. However, since this daily processing is usually run between 3:30 UTC and 6:00 UTC in H.E.S.S., some time is left to assess whether ToO observations should be triggered with H.E.S.S. for the next night, and with other multiwavelength facilities.

4. Conclusions and prospects

FLaapLUC, a tool designed to provide alerts on the fly on transient high energy sources using Fermi-LAT data, was presented. This pipeline can provide quick results to allow the prompt organisation of follow-up, multiwavelength observations. The method is Fig. 6. False alarm trigger probability on the first iterative light curve generation based on aperture photometry, which is not well suited to provide step as a function of αN1 . The setting used for extragalactic sources in H.E.S.S. is absolute flux measurements of Fermi-LAT data, but can be used to shown with the dotted grey lines and red point. (For interpretation of the references assess relative time variations from high energy emitting objects. to colour in this figure legend, the reader is referred to the web version of this article.) It has been shown that FLaapLUC is quick and efficient, and thus useful for providing alerts on flaring events from Fermi-LAT data. This can help in the organisation of follow-up ToO observations of transient γ -ray sources, for example with IACT such as 3.2. False alarm rate VERITAS, MAGIC and H.E.S.S. FLaapLUC results are compared to full likelihood analyses, The false alarm probability of the FLaapLUC pipeline is which show good agreement on relative flux variations. An eval- determined by simulating light curves of AGN following Em- uation of the associated false alarm probability reveals that this manoulopoulos et al.(2013). To do so, a scan is performed through tool is robust and efficient to detect transient events. A limitation the α parameter using N = 3 days to assess when FLaapLUC N1 1 comes from the latency of the overall data processing, of about would generate a finer binned light curve of N days. First, for 2 8 preventing the possibility of generating useful prompt alerts on each scanned value of α , sources from the monitored list which N1 events occurring on shorter time scales. have never triggered FLaapLUC so far in the first iterative step As long as the background is approximately constant, the aper- on the flux criterion were identified. For each of those sources, ture photometry method can be used to quickly detect active states 1000 mock light curves were simulated preserving the underlying from sources in data acquired by any instrument producing event probability density function and power spectral density from real data using the method described in Emmanoulopoulos et al.(2013) lists. For instance, it is conceivable to adapt such a system for online and adapted to Python by Connolly(2015). FLaapLUC was run triggering alerts for the future CTA observatory (CTA Consortium using the simulated light curves as inputs, and the false alarm et al., 2013), if events could be reconstructed fast enough (see e.g. Bulgarelli et al., 2014). probability on αN1 was taken from the rate at which N2-day binned light curve are generated. Fig. 6 presents the false alarm probability FLaapLUC (Lenain, 2017) has been made publicly available on GitHub at github.com/jlenain/flaapluc, and contributions from the when varying the threshold parameter αN1 . The operation point used in the H.E.S.S. extragalactic working group is shown in red, community are warmly welcome. corresponding to wrongly generating the finer light curve in 0.3% of the cases. The false alarm probability of the whole double-pass Acknowledgements procedure is not evaluated, being too resource consuming to be properly computed. However, it seems safe to state that with such I am very thankful to my colleagues within the H.E.S.S. collabo- = = running settings on αN1 2 and αN2 3, this false alarm rate ration for very fruitful discussions which led to the implementation is well below ∼ 0.1% for the AGN monitored within the H.E.S.S. of FLaapLUC and further improvements, and especially to Stefan collaboration. Wagner, Michael Punch, Heike Prokoph, Matteo Cerruti, Bruno Khélifi, Pol Bordas, Víctor Zabalza and Julien Bolmont. I am grateful 3.3. Computing time and latency to Agnieszka Jacholkowska, Matteo Cerruti, Julien Bolmont and Andrew Taylor for their careful reading of this manuscript. I would Processing a single source with FLaapLUC takes slightly more like to thank the anonymous referee for constructive inputs. than 5 min, when the input data set is limited to the last 30 days This research made use of Enrico, a community-developed of available data. In H.E.S.S., the FLaapLUC production instance Python package to simplify Fermi-LAT analysis (Sanchez and Deil, monitors about 900 sources every day, including about 300 AGN, 2013). This research has made use of NASA’s Astrophysics Data 540 regions in the Galactic plane, and about 60 binaries and other System. This research has made use of the SIMBAD database, op- Galactic sources. The whole processing typically takes 2 h (wall erated at CDS, Strasbourg, France (Wenger et al., 2000). I gratefully clock time) when 60 concurrent jobs are run at the IN2P3 comput- acknowledge CC-IN2P3 (cc.in2p3.fr) for providing a significant 8 ing cluster (CC-IN2P3). amount of the computing resources and services needed for this As stated above, to organise follow-up observations on a flare, work. a prompt reaction is essential. Summing up the time needed to This work is dedicated to the memory of my missed friend, Jean- Claude Rouffignat, who introduced me to astronomy when I was a 8 cc.in2p3.fr/en. child. You are not forgotten.

91 A. Selected publications

J.-P. Lenain / Astronomy and Computing 22 (2018) 9–15 15

References Bulgarelli, A., et al., 2014. A prototype for the real-time analysis of the Cherenkov Telescope Array. In: Ground-Based and Airborne Telescopes V. p. 91452X.doi: Abdo, A.A., et al., 2010a. Fermi large area telescope first source catalog. Astrophys. J. 10.1117/12.2054744. Suppl. Ser. 188, 405–436.doi:10.1088/0067-0049/188/2/405. arXiv:1002.2280. Cerruti, M., et al.2016. Target of Opportunity observations of blazars with H.E.S.S.. Abdo, A.A., et al., 2010b. The first catalog of active galactic nuclei detected by the ArXiv E-Prints arXiv:1610.05523. Connolly, S., 2015. A Python Code for the Emmanoulopoulos et al. [arXiv:1305. fermi large area telescope. Astrophys. J. 715, 429–457.doi:10.1088/0004-637X/ 0304]. Light Curve Simulation Algorithm. ArXiv e-prints. arXiv:1503.06676. 715/1/429. arXiv:1002.0150. Corbet, R., et al.2013. Looking for Stars and Finding the Moon: Effects of Lunar Abeysekara, A.U., et al., 2017. A luminous and isolated gamma-ray flare from the Gamma-ray Emission on Fermi LAT Light Curves. ArXiv e-prints arXiv:1302. Blazar B2 1215+30. Astrophys. J. 836, 205.doi:10.3847/1538-4357/836/2/205. 5141. arXiv:1701.01067. CTA Consortium et al., 2013. Introducing the CTA concept. Astropart. Phys. 43, 3–18. Acero, F., et al., 2015. Fermi large area telescope third source catalog. Astrophys. J. doi:10.1016/j.astropartphys.2013.01.007. Suppl. Ser. 218, 23.doi:10.1088/0067-0049/218/2/23. arXiv:1501.02003. Emmanoulopoulos, D., McHardy, I.M., Papadakis, I.E., 2013. Generating artificial Acero, F., et al., 2016. Development of the model of galactic interstellar emission for light curves: revisited and updated. Mon. Not. R. Astron. Soc. 433, 907–927.doi: standard point-source analysis of Fermi large area telescope data. Astrophys. J. 10.1093/mnras/stt764. arXiv:1305.0304. Suppl. Ser. 223, 26.doi:10.3847/0067-0049/223/2/26. arXiv:1602.07246. Hauser, M.G., Dwek, E., 2001. The cosmic infrared background: Measurements and Ackermann, M., et al., 2011. The second catalog of active galactic nuclei detected by implications. Ann. Rev. Astron. Astrophys. 39, 249–307.doi:10.1146/annurev. the Fermi large area telescope. Astrophys. J. 743, 171.doi:10.1088/0004-637X/ astro.39.1.249. astro-ph/0105539. 743/2/171. arXiv:1108.1420. H.E.S.S. Collaboration et al., 2013a. Discovery of very high energy γ -ray emission Ackermann, M., et al., 2013. The Fermi all-sky variability analysis: A list of flaring from the BL Lacertae object PKS with H.E.S.S.. Astron. Astrophys 559, A136.doi: gamma-ray sources and the search for transients in our galaxy. Astrophys. J. 10.1051/0004-6361/201321639. arXiv:1309.6174. 771, 57.doi:10.1088/0004-637X/771/1/57. arXiv:1304.6082. H.E.S.S. Collaboration et al., 2013b. H.E.S.S. discovery of VHE γ -rays from the quasar Ackermann, M., et al., 2015. The third catalog of active galactic nuclei detected by PKS. Astron. Astrophys 554, A107.doi:10.1051/0004-6361/201321135. arXiv: the fermi large area telescope. Astrophys. J. 810, 14.doi:10.1088/0004-637X/ 1304.8071. 810/1/14. arXiv:1501.06054. Lenain, J.-P., 2017. FLaapLUC: Fermi-LAT automatic aperture photometry light Ackermann, M., et al., 2016a. Measurement of the high-energy gamma-ray emission curve. Astrophysics Source Code Library. doi:10.5281/zenodo.906991. 1709.011. from the Moon with the Fermi Large Area Telescope. Phys. Rev. D 93, 082001. Mattox, J.R., et al., 1996. The likelihood analysis of EGRET data. Astrophys. J. 461, doi:10.1103/PhysRevD.93.082001. arXiv:1604.03349. 396–407.doi:10.1086/177068. Nalewajko, K., 2013. The brightest gamma-ray flares of blazars. Mon. Not. R. Astron. Ackermann, M., et al., 2016b. Minute-timescale >100 MeV γ -ray variability during Soc. 430, 1324–1333.doi:10.1093/mnras/sts711. arXiv:1211.0274. the giant outburst of quasar 3C 279 Observed by Fermi-LAT in June. Astrophys. Nolan, P.L., et al., 2012. Fermi large area telescope second source catalog. Astrophys. J. 824, L20.doi:10.3847/2041-8205/824/2/L20. arXiv:1605.05324. J. Suppl. Ser. 199, 31.doi:10.1088/0067-0049/199/2/31. arXiv:1108.1435. Aharonian, F., et al., 2007. An exceptional very high energy gamma-ray flare of PKS Rani, B., et al., 2017. Probing the gamma-ray variability in 3C 279 using broad- 2155-304. Astrophys. J. 664, L71–L74.doi:10.1086/520635. arXiv:0706.0797. band observations. Mon. Not. R. Astron. Soc. 464, 418–427.doi:10.1093/mnras/ Albert, J., et al., 2007. Variable very high energy γ -ray emission from Markarian 501. stw2342. arXiv:1609.04024. Astrophys. J. 669, 862–883.doi:10.1086/521382. astro-ph/070. Sanchez, D.A., Deil, C., 2013. Enrico: a Python package to simplify Fermi-LAT analy- Atwood, W.B., et al., 2009. The large area telescope on the Fermi gamma-ray space sis. ArXiv e-prints arXiv:1307.4534. telescope mission. Astrophys. J. 697, 1071–1102.doi:10.1088/0004-637X/697/ Wehrle, A.E., et al., 2016. Erratic Flaring of BL Lac in 2012-2013: Multiwavelength 2/1071. arXiv:0902.1089. Observations. Astrophys. J. 816, 53.doi:10.3847/0004-637X/816/2/53. Britto, R.J., et al., 2016. Fermi-LAT Observations of the 2014 May-July Outburst from Wenger, M., et al., 2000. The SIMBAD astronomical database. The CDS reference 3C 454.3. Astrophys. J. 830, 162.doi:10.3847/0004-637X/830/2/162. arXiv: database for astronomical objects. Astron. Astrophys. Suppl. Ser. 143, 9–22.doi: 1511.02280. 10.1051/aas:2000332. astro-ph/0002110.

92 A.2. Cloud ablation by a relativistic jet and the extended ﬂare in CTA 102 in 2016 and 2017 The Astrophysical Journal, 851:72 (9pp), 2017 December 20 https://doi.org/10.3847/1538-4357/aa9bee © 2017. The American Astronomical Society. All rights reserved.

Cloud Ablation by a Relativistic Jet and the Extended Flare in CTA 102 in 2016 and 2017

M. Zacharias1 , M. Böttcher1 , F. Jankowsky2, J.-P. Lenain3 , S. J. Wagner2, and A. Wierzcholska4 1 Centre for Space Science, North-West University, Potchefstroom, 2520, South Africa; [email protected] 2 Landessternwarte, Universität Heidelberg, Königstuhl, D-69117 Heidelberg, Germany 3 Sorbonne Universités, UPMC Université Paris 06, Université Paris Diderot, Sorbonne Paris Cité, CNRS, Laboratoire de Physique Nucléaire et de Hautes Energies (LPNHE), 4 place Jussieu, F-75252, Paris Cedex 5, France 4 Institute of Nuclear Physics, Polish Academy of Sciences, PL-31342 Krakow, Poland Received 2017 September 29; revised 2017 November 15; accepted 2017 November 16; published 2017 December 14

Abstract In late 2016 and early 2017, the flat spectrum radio quasar CTA 102 exhibited a very strong and long-lasting outburst. The event can be described by a roughly two-month long increase of the baseline flux in the monitored energy bands (optical to γ-rays) by a factor 8, and a subsequent decrease over another two months back to pre-flare levels. The long-term trend was superseded by short but very strong flares, resulting in a peak flux that was a factor 50 above pre-flare levels in the γ-ray domain and almost a factor 100 above pre-flare levels in the optical domain. In this paper, we explain the long-term evolution of the outburst by the ablation of a gas cloud penetrating the relativistic jet. The slice-by-slice ablation results in a gradual increase of the particle injection until the center of the cloud is reached, after which the injected number of particles decreases again. With reasonable cloud parameters, we obtain excellent fits of the long-term trend. Key words: galaxies: active – quasars: individual (CTA 102) – radiation mechanisms: non-thermal – relativistic processes

1. Introduction interaction, where the penetration process is included, reveal that the obstacle is (partially) ablated, and a significant amount Blazars, the relativistically beamed, radio-loud version of of matter is mixed into the jet flow. active galactic nuclei (Blandford & Rees 1974), are historically This is easy to see for a gas cloud, given that it is mainly categorized into two classes depending on the width of their confined by its own, rather weak gravity. The ram pressure of optical emission lines: BL Lacertae objects with line equivalent the jet will immediately start to ablate the outer layers of the width EW<5 Å, and flat spectrum radio quasars (FSRQs) cloud while it starts to penetrate the jet. The mass loss of the with EW>5 Å. The latter case indicates the presence of a strong broad-line region (BLR) surrounding the central super- cloud will weaken its structural integrity even before it has massive black hole on scales of ∼0.1 pc. The origin of the fully penetrated the jet. As we will discuss below, the cloud double-humped spectral energy distribution (SED) is regarded will be ablated and carried along by the jet. Depending on the by most authors to be synchrotron and inverse-Compton (IC) cloud parameters, such as size and velocity, this might lead to emission of particles within the relativistic jet, with electrons pronounced and prolonged jet activity, when the additional and positrons being responsible for the emission, and protons material in the jet reaches an internal shock located down- serving as a cold background. Especially in FSRQs, seed stream of the cloud penetration site. We apply this model to a fl fl fi photon fields for the IC process are abundant. Apart from the recent are in CTA 102, where uxes varied signi cantly over emission region’s internal synchrotron emission (resulting in several months. synchrotron self-Compton, SSC, flux), the external fields from CTA 102 is an FSRQ at a redshift zred = 1.037, roughly the accretion disk, the BLR, or the dusty torus are also potential halfway across the observable universe. The accretion disk ¢ 46- 1 ( targets depending on the distance of the emission region from luminosity is Ldisk =´3.8 10 erg s Zamaninasab et al. ) the black hole. 2014 . The mass of the central black hole is estimated at M ∼ × 8 M ( ) Blazars are strongly variable in all energy bands. The large bh 8.5 10 e Zamaninasab et al. 2014 , giving an 47- 1 variety in flaring events has led to a similarly large number of Eddington luminosity of LEdd¢ ~´1.1 10 erg s . The BLR models. A particularly interesting case is the interaction of the properties were derived by Pian et al. (2005) using UV jet with an obstacle, such as a star (Blandford & Königl 1979; spectroscopy observations with the Hubble Space Telescope, 45- 1 Komissarov 1994; Perucho et al. 2014; Bosch-Ramon 2015), resulting in a luminosity of LBLR¢ =´4.14 10 erg s and a 17 ( its wind (Araudo et al. 2009; de la Cita et al. 2017), or a gas radius of RBLR¢ =´6.7 10 cm all quantities given in the cloud (Araudo et al. 2010; Bosch-Ramon et al. 2012). What AGN frame). most of these models have in common is that the obstacle is Long-term observations in radio bands since 1980 (Fromm already fully inside the jet before the start of the interaction. et al. 2011) revealed a rather dormant source until ∼1997, after However, given the strong pressure of the relativistically which it showed a few radio outbursts with a particularly strong moving matter of the jet, interactions will start as soon as the one in 2006. Fromm et al. (2011) favor a shock–shock obstacle hits the jet, since the jet will look like a strong shock. interaction scenario to explain the observed evolution of the Simulations of shock/cloud interactions have shown that a latter event. Similarly, in the high-energy (HE; E>100 MeV) cloud will be quickly ripped apart (Klein et al. 1994; γ-ray band, scanned continuously by the Fermi satellite since Poludnenko et al. 2002). Recent simulations of a jet/cloud mid-2008, CTA 102 showed low fluxes in the first almost four (Bosch-Ramon et al. 2012) or jet/star (Perucho et al. 2017) years of Fermi-LAT operation with an average flux above

93 A. Selected publications

The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al. 1 GeV of (5.0 ± 0.2)×10−9 ph cm−2 s−1 and a photon index of Γ=2.34±0.03 (Acero et al. 2015). In the second half of 2012, CTA 102 exhibited a strong γ-ray outburst with a peak − − − flux above 100 MeV of ∼8×10 6 ph cm 2 s 1. This outburst along with the correlated optical variability led Larionov et al. (2016) to propose the helical motion and the accompanied variation of the Doppler factor of a plasma blob (Schramm et al. 1993) as the main driver of the flare. Since 2012, CTA 102 has remained active without long returns to pre-flare levels in both the γ-ray and optical bands. However, all of these outbursts have been rather short-lived on the order of a few days, with fast rises to the maximum and subsequent quick decays. This behavior changed in late 2016, when CTA 102 entered into a prolonged activity phase, which saw both the γ-ray and optical fluxes, as well as the X-ray flux, rising continuously for about two months. The peak fluxes were obtained at the end of 2016 December, which were in all cases significantly higher Figure 1. Light curves of (a) Fermi-LAT data, (b) Swift-XRT data, and (c) fl fl optical data from ATOM as labeled. The vertical thin black and red lines mark than any previously observed uxes. The optical uxes the dates when the spectra were extracted. exhibited clear intranight variability (Bachev et al. 2017). Subsequently, the γ-ray flux decreased over the course of about The Fermi-LAT data are analyzed using the public two months to 2016 October levels. Unfortunately, this ScienceTools v10r0p5.5 Events in a circular region of decrease of flux could not be observed in optical or X-ray interest of 10° in radius are extracted, centered on the nominal observations due to Sun constraints. position of 3FGL J2232.5+1143. To probe the active state In this paper, we present the multiwavelength data of this reported here, only data between 2015 August 8 (MJD 57235) roughly four-month-long outburst and explain it by the ablation and 2017 May 1 (MJD 57874), in the 100 MeV–500 GeV of a gas cloud by the relativistic jet. The initial density increase energy range, are considered. The P8R2_SOURCE_V6 instru- in ablated material causes the rise of the light curve, while the ment response functions (event class 128 and event type 3) ablation of the second half of the cloud exhibits a decrease in were used, together with a zenith angle cut of 90° to avoid ablated material, resulting in the subsequent drop of the light contamination by the γ-ray-bright Earth-limb emission. The curve. Our focus is on the explanation of the long-term trend, model of the region of interest was built based on the 3FGL and we do not deal with the fast variability on top of the longer catalog (Acero et al. 2015). The Galactic diffuse emission was trend. The paper is organized as follows. First, we present the modeled using the file gll_iem_v06.fits (Acero data analysis in Section 2. Section 3 describes the theoretical et al. 2016) and the isotropic background using iso_- model of cloud ablation, followed by a summary of the code P8R2_SOURCE_V6_v06.txt. In the following, the source used and the modeling in Section 4. We discuss and conclude spectrum will be investigated both with a power-law shape, in Section 5. In the following sections, primed quantities are in the AGN dN ⎛ E ⎞-G frame, quantities marked with the superscript “obs” are in the = N0⎜ ⎟ ,1 dE ⎝ E ⎠ () observer’s frame, and unmarked quantities are in the comoving 0 jet frame. We use a standard, flat cosmology with and a log-parabola, −1 −1 H0=69.6 km s Mpc and ΩM=0.27, which gives a 28 ⎛ ⎞-G+(())b log EEb luminosity distance dL=2.19×10 cm. dN E = N ⎜ ⎟ ,2 0⎝ ⎠ () dE Eb

2. Data Analysis with Eb=308 MeV fixed to the value reported in the 3FGL catalog. The flare in CTA 102 was extensively observed by a large For the considered period between 2015 August and 2017 number of observatories. Here, we analyze and report the May, CTA 102 is detected with a Test Statistic (TS; Mattox detailed observations of Fermi-LAT in the γ-ray band, Swift- et al. 1996) of 163,879, i.e., ∼405σ. The spectrum of CTA 102 X-ray Telescope (XRT) in the X-ray band, as well as the Swift- is significantly curved with a photon index of Γ=2.068± UVOT and Automatic Telescope for Optical Monitoring 0.008 and a curvature index of β=0.064±0.003. The (ATOM) in the optical band. −6 −2 −1 average flux is F=(2.27 ± 0.01)×10 ph cm s . To further study the activity of CTA 102 at high energies, a light curve has been produced with a time binning of 1 day. 2.1. Fermi-LAT Data Analysis Since on these timescales the preference for a log-parabola is The LAT instrument (Atwood et al. 2009) on board the not guaranteed, the spectrum has been modeled with a simple Fermi satellite surveys the high-energy γ-ray sky every 3 hr, power law in each time bin, leaving the photon index free to with energies between 20 MeV and above 300 GeV, thus vary. The resulting light curve is shown in Figure 1(a). making it an ideal instrument to monitor the activity of CTA From this data set, spectra were derived for two particular 102. This AGN has been reported in all of the available Fermi- dates: MJD 57670 and MJD 57745, which are representative of LAT catalogs and is identified as 3FGL J2232.5+1143 in the third Fermi-LAT source catalog (Acero et al. 2015). 5 See http://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation.

94 A.2. Cloud ablation by a relativistic jet and the extended flare in CTA 102 in 2016 and 2017 The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al. the pre-flare state and the flare state around the maximum. For correspond to the observations taken nearest to MJD 57745, MJD 57670, CTA 102 is detected with TS=161 (∼12σ), and which is data with ObsId 00033509109. Apparently, only the the observed spectrum is well-described by a power law with normalization of the spectra changes. − − − F=(1.19 ± 0.23)×10 6 ph cm 2 s 1 and Γ=2.08±0.14. Testing a log-parabola only yielded a log-likelihood ratio 0.2 2.3. Optical/UV Analysis with respect to a power law. In order to validate that the nondetection of curvature is independent of the detection Simultaneously with XRT, CTA 102 was monitored with the significance, we derived a 10 day spectrum starting on UVOT instrument on board Swift. The observations were taken MJD 57670. Despite the increased significance of the source in the UV and optical bands with the central wavelengths with TS=683 (∼26σ), the spectrum is still compatible with a of UVW2 (188 nm), UVM2 (217 nm), UVW1 (251 nm), power law, since the log-parabola is only preferred with 0.95σ. U (345 nm), B (439 nm), and V (544 nm). The instrumental For MJD 57745, the detection level of CTA 102 reaches magnitudes were calculated using the uvotsource task TS=4558 (∼67σ), and the observed source spectrum is including all photons from a circular region with radius 5″. The significantly curved, with a log-likelihood ratio of 9.9 for a log- background was determined from a circular region with a parabolic spectrum with respect to a power law. The radius of 5″near the source region that is not contaminated corresponding spectrum results in F=(1.10 ± 0.05)× with signal from any nearby source. The optical and ultraviolet − − − 10 5 ph cm 2 s 1, Γ=1.797±0.061 and β=0.077± data points were corrected for dust absorption using the 0.025. The two one-day spectra are shown in Figure 4 as the reddening E(B − V )=0.0612 mag (Schlafly & Finkbeiner black and red butterflies, respectively. The displayed spectra 2011) and the ratios of the extinction to reddening, were corrected for absorption by the extragalactic background Aλ/E(B − V )(Giommi et al. 2006). light (EBL) following the model of Franceschini et al. (2008), Further optical data in the R- and B-band filters were which has, however, only a minor influence at the highest obtained with the ATOM, which is a 75 cm optical telescope energies. located at the H.E.S.S. site in the Khomas Highland in Namibia The change in spectral shape can be interpreted as a move of (Hauser et al. 2004). It regularly observes roughly 250 γ-ray the peak energy during the flare toward higher energies. emitters. Although the peak of the IC component cannot be determined ATOM has monitored CTA 102 since 2008. During the before the flare (somewhere between 10 keV and 100 MeV), visibility period presented in this paper, R-band monitoring during the peak of the flare it is at about 3 GeV. This points lasted from 2016 June until 2017 January. Additional B-band toward a significant hardening of the underlying particle observations were taken from 2016 October until 2016 distribution. December. Most of the high-flux period is covered by at least one B-band and several R-band measurements per night. The 2.2. X-Ray Analysis data were analyzed using the fully automated ATOM Data Reduction and Analysis Software and were manually quality The Swift Gamma-Ray Burst Mission (hereafter Swift; checked. The resulting flux was calculated via differential Gehrels et al. 2004) is a multifrequency space observatory that photometry using five custom-calibrated secondary standard allows targets in the optical, ultraviolet, and X-ray energy stars in the same field of view. bands to be monitored. The XRT (Burrows et al. 2005) has Using measurements from a calm period between 2008 and monitored CTA 102 since 2005 in 137 pointing observations 2011, the baseline flux of CTA 102 can be established as taken in the energy range of 0.3–10 keV. In this work, the light R=16.90±0.02 mag. An outburst in 2012 September curve (Figure 1(b)) presents data collected between MJD 57668 reached R=14.6±0.1 mag before returning to previous and MJD 57821, which correspond to the ObsIDs of levels. In late 2015, ATOM detected CTA 102 at 00033509084–00033509120. R=16.54±0.08 mag. Beginning in mid-2016, CTA 102 All data collected were analyzed using version 6.20 of the showed increasing activity, with a first outburst in August HEASOFT package.6 The data were recalibrated using the reaching R=14.20±0.02 mag. Toward the end of visibility, standard procedure xrtpipeline. For the spectral fitting, CTA 102 started to steadily brighten, culminating in XSPEC v.12.8.2 was used (Arnaud 1996). All data were binned R=10.96±0.05 mag on 2016 December 29 (MJD 57751). to have at least 30 counts per bin. Each observation was fitted We find significant intranight variability, similar to the results using the power-law model, Equation (1), with the Galactic reported in Bachev et al. (2017). Both R- and B-band light 20 −2 absorption value of NH=4.76×10 cm (Kalberla curves are shown in Figure 1(c). et al. 2005) set as a frozen parameter. In each observation, We confirmed that the color of the optical/UV spectra is we also checked if a broken power-law model can result in a constant in time, which implies that the peak of the synchrotron better description of the spectrum. According to reduced χ2 component does not move significantly from its initial, values, a simple power law is the best model for all data in unknown position in the infrared toward bluer, optical our set. frequencies. This has the unfortunate side effect that we cannot The two observations presented in the global SED (Figure 4) determine the peak synchrotron energy during this flare. On the are described with the following spectral parameters: Γ57670 = other hand, one can deduce that neither the maximum Lorentz −3 −2 −1 −1 1.3 ± 0.2 and N57670 = (1.17 ± 0.16)×10 cm s keV factor of the electrons nor the magnetic field increases and Γ57745=1.52±0.06 and N57745=(3.93 ± 0.18)× significantly. 10−3 cm−2 s−1 keV−1. The spectrum shown with black symbols corresponds to observations taken nearest to MJD 57670, 2.4. Flux Evolution After 2017 March which is data with ObsId 00033509084, while red symbols Between mid-January and late April, the source is not visible 6 http://heasarc.gsfc.nasa.gov/docs/software/lheasoft to optical and X-ray observatories, since CTA 102 is too close

95 A. Selected publications

The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al. to the Sun during these months. Hence, the downward trend quantity. Rj is the radius of the jet. Hence, the number of visible in the γ-ray light curve could not be observed in any particles in the cloud equals the difference in the particle other band. Swift and ATOM resumed observations of CTA number in the jet at the peak of the event to the beginning of 102 in late April. the event: The optical flux was still highly variable between ¢ R=16 mag and R=13 mag while displaying a general trend NNc ==c 2() Nj,max - N j,min of becoming faint. The behavior in the X-ray band was similar. 8p 3 In the γ-ray domain, fluxes became variable again in early =+-RaUnj 11.5je,,min 3 ()() () April, exhibiting day-long outbursts similar to the behavior before 2016 October. Therefore, we conclude that the optical The factor 2 takes into account that the maximum of the event and X-ray activity at that time is unrelated to the γ-ray activity takes place when the center of the cloud is ablated and the between 2016 October and 2017 March and of no concern for second half of the cloud is still to be ablated. our modeling. U = nnje,,max je ,,min marks the ratio of the electron densities at the peak and the beginning of the flare. Naturally, in the 3. Cloud Ablation by the Relativistic Jet cloud, a=1. Hence, the addition of cloud material into the jet The potential cause of a strong outburst is the accumulation should raise the value of a in the jet emission region. For ease of more matter than usual in a standing shock within the jet. If a of computation, we ignore this effect here. For an initial jet gas cloud on its orbit around the black hole happens to plasma with a1, the influence is negligible. penetrate the jet, it will be ablated and carried along by the jet. The jet ablates the cloud due to its ram pressure, which is in fi Hence, this is an ef cient process for the jet to pick up a large the black hole frame, amount of material and to cause prolonged jet activity, if the ¢¢ ¢ cloud is ablated at a steady pace while it enters the jet. Given pnmcnmc=G-11g 2 +G- 2 the changing density within the cloud, the jet ablates different ram ()j je,,min ee () j jp,,min p ⎛ ⎞ amounts of matter at a given time while the cloud penetrates the mp =G G -11g mc2 ⎜ + a ⎟n , 6 jet. This leads to a gradual increase and decrease of the light jj() ee je,,min () ⎝ geem ⎠ curve during the flare over the timescale the cloud is ablated. Density fluctuations within the cloud and instabilities during introducing the bulk Lorentz factor Γj of the jet and the speed the process might lead to a more chaotic ablation, which could of light c. Not surprisingly, for fractions of protons result in strong and fast fluctuations related to the size of these 1  amm> g e p, with the average electron Lorentz factor fluctuations on top of the longer trend. In the following, we will ¯e g , the electron mass me, and the proton mass mp, the ram concentrate on the long-term trend and discuss the influence of ¯e pressure is dominated by protons. Since the ram pressure of the density fluctuations elsewhere. jet is provided by particles already present in the jet, it remains The situation is that a spherical cloud approaches the jet with fl orbital speed around the central black hole constant throughout the are and can be reconstructed by pre- flare parameters. vGMzc¢ = bh ¢ ,3() The gravitational pressure that keeps the cloud together is where G is the gravitational constant and z¢ the distance Fr¢¢ GM r¢ m pr¢ = gc()= c ()c H ,7 g ()c 2 2 () between the cloud and the black hole. The radius of the cloud AH prr¢ ( Hc can be derived from the rising time tf¢ that is, from the 2 -17 2 beginning to the peak) of the event: where AH =~´prH 8.8 10 cm is the cross-section and mH∼mp is the mass of a hydrogen atom, which constitutes the obs tvfc¢ bulk of the particles in the cloud. Mr¢ is the enclosed mass at Rtv¢ = ¢¢= .4 c ( c ) cfc () ¢ ()1 + zred cloud radius rc . ¢ ¢ The cloud will be ablated if ppram > g. Hence, with Apart from the redshift correction, the frame of the cloud and ( ) ( ) ’ Equations 6 and 7 , and a slight redistribution, we can the observer s frame are identical, since the motion of the cloud construct a lower limit on the initial jet electron density: is nonrelativistic. Hence, the observed duration of the flare is indeed the same as the cloud penetration time. GmH Mrc ()c¢ nje,,min> .8 The number of particles in the cloud follows from the 2 2 m () prmcH e p 2 GG-jj()11g e +arc¢ increase in particles in the jet, under the assumption that ¯ g¯eme the cloud is fully ablated. We can calculate the particles in the () cloud if we take the difference between the particles at the peak In order to get an estimate on the required jet electron density, and at the beginning of the flare. This includes the simplifying we chose the outer layer of the cloud rcc¢ = R¢ as an example. m ∼m fi assumption that the cloud contains a pure hydrogen plasma. Approximating g¯e  mmpeand H p,we nd Within the emission region of the jet, the density of electrons ( ) ⎛ ⎞-1⎛ GG-⎞--11⎛ 1 ⎞ and possibly positrons is nje, . The electron charge is balanced -12⎜⎟a jj nje,,min 2.8´ 10 ⎜ ⎟ ⎜ ⎟ by a fraction a1 of protons, depending on the number of ⎝ 0.1⎠ ⎝ 10 ⎠ ⎝ 9 ⎠ positrons in the jet. The total density of particles in the jet is ⎛ ⎞⎛ ⎞-2 Mc R ¢ nj =+()1 anj,e. The number of particles in the emission ´ ⎜ ⎟⎜ c ⎟ cm-3 . 9 4 ⎝ ⎠⎝ 15 ⎠ () region obviously is N = pRn3 , which is an invariant 0.01M 10 cm j 3 j j

96 A.2. Cloud ablation by a relativistic jet and the extended ﬂare in CTA 102 in 2016 and 2017 The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al.

( ) ( ) ( ) ( ) Figure 2. a Numerical solid and approximate dashed solution for the cloud density distribution nc¢, Equation 11 , as a function of cloud radius rc¢ for two values of 10 −3 ( ) ( ) ( ) the cloud temperature Tc¢ as labeled. The central density is set to n0¢ = 10 cm . b Integration of the numerical solution solid and the analytical solution dashed , ( ) x′ ( ) Equation 17 , of the cloud density distribution as a function of slice position for two values of its temperature Tc¢ as labeled. Parameters as in a . In both panels, we dropped the primes for clarity.

Obviously, the cloud cannot withstand destruction. Even a of the cloud that first touches the jet. That is, x¢=0 where the fi ’ solar-like star with much higher surface gravity could be cloud rst touches the jet, xR¢= c¢ is the cloud s center, and stripped of its outer layers while penetrating the jet, which xR¢=2 c¢ marks the rear end of the cloud. With the speed of the −2 −3 t′ typically exhibits electron densities exceeding 10 cm . cloud, it can be written as xvt¢= c¢ ¢, where is the time that has However, this estimate might not hold for the inner, dense passed since first contact in the AGN frame. core of a star. The number of particles ablated in each slice is the integral ¢¢ Given that the cloud penetrates the jet gradually, the number over the density ncc(r ) with respect to the slice volume. In the x′ of particles injected into the jet changes over time. In order to case of a sphere, the volume of a slice between positions and ( ) calculate the correct injection term, the density distribution of xdx¢+ ¢ is Zacharias & Schlickeiser 2013 the cloud n¢¢r must be known. We consider a profile based on cc( ) ¢ ¢ ¢ 2 hydrostatic equilibrium. The simplest ansatz would be to dVssc() x¢ =¢ dxò dA () x¢ = p (2,14 R x¢ - x¢ ) dx¢ ( ) assume that the cloud consists of an isothermal ideal gas with ¢ ¢ ¢¢ where As¢ x is the cross-section of a slice and dx¢ its width. temperature Tc , so that the thermal pressure pkTmT = rc B c p, () ¢ ¢ ’ k The particle number in each slice then becomes where rc = mnp c is the cloud s mass density, and B is the Boltzmann constant. In this case, the equation of hydrostatic ¢ ¢¢ ¢ equilibrium reads dNsccs() x¢ =¢ dxò n () r dA () x¢ .15 ( ) kT¢ drr¢¢ Writing the integral in cylindrical coordinates with B c c ()c =-gr¢¢¢r r ()c c ()c ¢ 22¢ ¢ ¢ 2 mp drc¢ rcc()xRx¢=w + (-¢ ) and wcc()xRxx¢ = 2 ¢ - ¢ , ¢ Equation (15) becomes Grr¢¢ rc c ()c 2 =-4.10p drrr¢ r ¢ 2 ò ˜˜c (˜) ( ) wc()x¢ rc¢ 0 ¢ ¢¢ dNs () x¢ =¢2.16pwww dxò ncc (()) r d ( ) fi ¢ ( ) 0 With the de nition tp¢ º kTB c (4 mp G), Equation 10 reduces to Inserting Equation (12) into Equation (16), the integral can be ⎛ ⎞ easily solved, giving d r¢2 dr¢ t¢ ⎜ c ⎟ =-r¢ r¢2.11 ⎛ ⎞ c () rR¢22+ ¢ drc¢ ⎝ r¢ drc¢ ⎠ 2 ⎜ 0 c ⎟ dNs¢() x¢ =¢p dx r0¢¢ n0 ln .() 17 ⎝ rRx¢2 + ¢ - ¢ 2 ⎠ The numerical solution to this nonlinear differential equation is 0 ()c ( ) plotted in Figure 2 a for two values of Tc¢. As is also shown in This function is shown in Figure 2(b) for two cases of Tc along that plot, the numerical solution is well approximated by with an integration of the numerical solution of Equation (11). n ¢ The analytical approximation and the exact result match nicely. nr¢¢= 0 .12The injection of particles in the jet, which get dragged along cc() ¢¢2 () 1 + ()rrc 0 and cause the flare at a shock somewhere downstream, can then be described by The normalization n0¢ can be determined by integrating ( ) ( ) Equation 12 and equating it to Equation 5 , and ⎛ ¢22¢ ⎞ ⎛ ⎞ ⎜ rR0 + c ⎟ ⎜ x¢ ⎟ Qtinj()µ lnd t - .() 18 3t¢ ⎝ rRx¢2 + ¢ - ¢ 2 ⎠ ⎝ v¢ ⎠ r¢ = .13 0 ()c c 0 ¢ () mnp 0 ’ δ Here, d ()q is Dirac s function, which describes the slice-by- Naturally, the density drops to zero for rc¢ ¥. In order to slice ablation in time. make progress, we approximate the cloud as a sphere with We stress that the entire mass of the cloud is not added to the ¢ ¢ ¢¢ ¢ outer boundary Rc > r0 and set ncc()rR c= 0. Once the jet at once, but gradually over about four months in cloud hits the jet, it is ablated slice by slice beginning with a the observer’s frame. Hence, the impact of the added mass low particle-number region at the front, through the dense on the jet’s bulk Lorentz factor at any given time is minor central region, and ending again at a low-density region at the compared to the case where the entire cloud mass would be rear end. Therefore, we define all quantities of the cloud as a added at once. In the following, we assume a constant jet bulk function of x′, the slice position with respect to the outer edge Lorentz factor.

97 A. Selected publications

The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al. 4. Modeling In order to model the long-term trend of the CTA 102 flare, we use the code by Diltz & Böttcher (2014) and adapt it slightly to accommodate the variability induced by the cloud ablation as discussed in Section 3. The code calculates the electron distribution and photon emission spectra in the comoving frame of the emission region, and subsequently transforms it to the observer’s frame taking into account the Doppler factor δj, which we assume here to be equal to the bulk Lorentz factor Γj, and the redshift zred. The electron distribution function ne(γ, t) is calculated with a Fokker–Planck-type differential equation that takes into account injection, stochastic acceleration, cooling, and escape. The injection electron distribution is of the form -st Qt()ggggg,;,,19= Qt0minmax ()() H [ ()()] t t () γ s where is the electron Lorentz factor, the electron spectral ( ) Fermi ( ) Swift ( ) ’ Figure 3. Light curves of the a -LAT data, b -XRT data, and c index, and H [gg;,min g max ] denotes Heaviside s step function ATOM/R data. The thick red lines are the modeling result, while the vertical with H=1 for γminγγmax, and H=0 otherwise. The thin black and red lines mark the dates when the spectra have been extracted. injection normalization is derived from input parameters as Note the logarithmic scaling of the y-axis. ⎧ 2 - st() ifs ¹ 2 ⎪ gg2--st()- 2 st () Ltje,,inj()⎨ max min Qt0 = ,20 () 2 g -1 () Vmje c ⎪ lnmax ifs = 2 ⎩ g ()min with the electron injection luminosity Lje,,injand the comoving volume V = 4 pR3 of the emission region. Since the input j 3 j parameters can be time dependent, the injection distribution might change in every time step. The acceleration and escape terms are parameterized independent of energy. The escape timescale is defined by tesc=ηesc R/c, namely, a multiple ηesc of the light-crossing timescale. The acceleration timescale in turn is defined as a multiple ηacc of the escape timescale: tacc=ηacctesc. The cooling term takes into account all radiative processes, namely synchrotron radiation in a randomly oriented magnetic field Bj, SSC, and IC emission on potential external fields, such  fl as the accretion disk, the BLR, or a dusty torus. The IC process Figure 4. Two representative spectra of CTA 102 during the are: MJD 57670 – (black symbols and butterfly) and MJD 57745 (red symbols and butterfly). The takes into account the full Klein Nishina cross-section. The Fermi-LAT spectra have been corrected for EBL absorption using the model of accretion disk spectrum is assumed to be of the Shakura & Franceschini et al. (2008). The thick black and red solid lines show model Sunyaev (1973) type, which basically depends on the mass of spectra for the beginning and the peak of the flare, while the thin solid lines (magenta, green, orange, yellow, blue) show the evolution of the model the central black hole and the Eddington ratio ηEdd of the disk ¢ ¢ spectrum in roughly 10 day steps toward the maximum. The other lines give luminosity Ldisk = hEdd LEdd. The BLR spectrum is assumed to example curves of the composition of the spectrum: accretion disk (dashed be a blackbody spectrum of effective temperature TBLR¢ magenta), BLR (dashed green), synchrotron (dotted black), SSC (dashed– normalized to the measured BLR luminosity. The size of the dotted black), and IC/BLR (dashed–double-dotted black). BLR is important to calculate the BLR energy density and potential absorption of γ-rays from the emission region. Similar to being representative of the respective flux levels, the data definitions are possible for the dusty torus, but we ignore that taken in all bands is contemporaneous. The spectra are shown photon field here due to a lack of observational evidence. in Figure 4. Most obvious are the significant flux changes An implicit Crank–Nichelson scheme is used to solve the between the two states and the change in peak energy of the IC – fi Fokker Planck equation. With the solution for ne ()g, t , the component. The parameters of the t to the low state are given radiation spectra are derived, which consider internal absorp- in Table 1. tion through synchrotron self-absorption and external absorp- A few of these parameters are constrained by observations. tion of γ-rays through the external soft photon fields. The size of the emission region modulo the Doppler factor Before starting to model the light curve, we first derived two Rj/δj is constrained by the variability timescale in our data as fl obs exemplary spectra of the low state before the are in October Rjj D+tcd (1 zred). Due to the measured optical intra- and of the high state in late December in order to derive the night variability, the emission region must be smaller than a baseline parameter sets that needed to be matched at these two lightday in the observer’s frame, corresponding to less than 16 states. The dates are MJD 57670 and MJD 57745, and are 4.5×10 cm for a Doppler factor of δj=35. The chosen 16 marked by the black and red vertical lines in the light curves of value of Rj=2.5×10 cm is a compromise between the Figure 3, respectively. We chose these dates, since, in addition aforementioned limit and the necessity of a rather large

98 A.2. Cloud ablation by a relativistic jet and the extended ﬂare in CTA 102 in 2016 and 2017 The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al.

Table 1 Model Parameter Description, Symbol, and Value

Deﬁnition Symbol Value Emission region distance to black hole z′ 6.5×1017 cm Doppler factor of emission region δj 35 16 Emission region radius Rj 2.5×10 cm Magnetic ﬁeld of emission region Bj 3.7 G 43 −1 Electron injection luminosity Lj,e,inj 2.2×10 erg s 1 Minimum electron Lorentz factor γmin 1.3×10 3 Maximum electron Lorentz factor γmax 3.0×10 Electron spectral index s 2.4

Escape time scaling ηesc 10.0 Acceleration to escape time ratio ηacc 1.0 × 4 Effective temperature of the BLR TBLR¢ 5.0 10 K

43 −1 Electron luminosity variation ΔLj,e,inj 1.75×10 erg s Electron spectral index variation Δs −0.6 Time between onset and peak of flare tobs 60 days Figure 5. (a) Electron distribution function γ2 n(γ, t) as a function of the f γ ( ) ¢ × 14 electron Lorentz factor for the same time steps as in Figure 4. b Electron Cloud scale height r0 1.6 10 cm -2 γ cooling term gg∣˙∣ as a function of the electron Lorentz factor . Note. Values below the horizontal line mark parameters for the induced variability. electron Lorentz factor γmin is not constrained by the observations, but has been chosen in such a way that the IC/BLR emission region in order to keep the SSC emission low. The spectrum fits well the hard X-ray spectrum. In principle, the latter would quickly overproduce the X-ray flux for smaller γ-ray spectrum could be used to constrain the spectral index s. source radii especially during the variable period. However, as one can see in Figure 5, the Klein–Nishina effect in The magnetic field Bj is constrained from the Compton the cooling changes the particle spectrum considerably at the dominance parameter W, which is defined as the ratio of the particle energies that correspond to the γ-ray spectrum probed by fl fl peak uxes of the two spectral components. The peak uxes Fermi-LAT (see also the discussion below). Hence, the standard are directly proportional to the underlying energy densities, relations between photon spectra and (un)cooled particle namely the magnetic energy density and the energy density in distributions do not work, and we chose the spectral index to the BLR photon field transformed to the comoving frame. Fermi 2 match well the -LAT spectrum. ¢ B obs 48 −1 Hence, W =G43j uuBLR B. Solving for j, one obtains The observed luminosity of L ∼10 erg s of the ground state is at the high end of FSRQ luminosities (e.g., 8G2L ¢ B = j BLR Ghisellini et al. 1998). In order to reduce the required particle j 2 3cRBLR¢ W energy densities, we chose a relatively high Doppler factor of δ = ⎛ ⎞ -12⎛ ⎞12 j 35. However, observations of the MOJAVE program Gj ⎛ W ⎞ L ¢ = 2.9⎜ ⎟⎜⎟ ⎜ BLR ⎟ revealed radio knots moving with apparent speeds of ~18c ⎝ ⎠ - ⎝10⎠ 10⎝ 4.14´ 1045 erg s 1 ⎠ (Lister et al. 2016), which permit Doppler and Lorentz factors -1 in the chosen order of magnitude. ⎛ R¢ ⎞ ´ ⎜ BLR ⎟ G. 21 Although there is no observational constraint on the distance ⎝ 17 ⎠ () 6.7´ 10 cm of the emission region from the black hole z¢, it is chosen in such a way that the emission region is immersed in BLR Unfortunately, the peak fluxes of both the synchrotron and the photons, but still the attenuation of γ-rays by the BLR photons IC component are not well-defined in the low state. Hence, W is is minimal. Closer to the black hole, the attenuation would start not particularly well constrained, and values of at least 10 are to become important even in the HE domain, resulting in softer plausible. We follow the standard assumption of the one-zone spectra and a much poorer fit. A greater distance from the black fi model that the magnetic eld is tangled. Although this is a hole would result in an inefficient IC/BLR process. Hence, the simplification, since one expects an ordered guide magnetic emission region should be located around the outer edge of field in the jet, we have no observational constraints in hand the BLR. fi that could constrain the geometry of the magnetic eld during The results are insensitive to the escape time scaling ηesc and this particular event. Larionov et al. (2016) modeled their the acceleration to escape time ratio ηacc, since the strong polarimetry data of the 2012 flare assuming a helical magnetic cooling (see Figure 5) dominates over the escape and field and a helical motion of the emission region, which is acceleration for all energies. The effective temperature of the different from our model. BLR TBLR¢ is also not constrained by observations, but it – The maximum electron Lorentz factor γmax is constrained by impacts the onset of the Klein Nishina domain in the IC the soft optical synchrotron spectrum. While a soft electron process. The chosen value implies that the Klein–Nishina distribution could also account for the soft synchrotron spectrum, domain already sets in for electron Lorentz factors of ∼100. this would be inconsistent with the harder (than the optical Lower values of TBLR¢ would increase the electron turnover spectrum) γ-ray spectrum. Hence, the soft optical spectrum can energy slightly. be interpreted as an exponential cutoff induced by a maximum We note that the chosen parameter set is not unique, and electron Lorentz factor significantly below 104. The minimum other parameter sets might give equivalent results. However,

99 A. Selected publications

The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al. the precise parameters are not important for the evolution of the 5. Discussion event, which is the main concern of this paper. The modeling gives a good representation of the overall flare In order to model the evolution of the flare, we varied the profile. We can safely conclude that the long-term activity of electron injection luminosity following Equation (18) as CTA 102 is consistent with the addition of a large amount of ⎛ 22⎞ mass to the jet over a time period of a few months. We modeled tt0 + f LtL=+D Lln⎜ ⎟ , 22 this by the penetration of the jet by a gas cloud, for which we je,,inj() je ,,inj je ,,inj 2 2 () ⎝ ttt0 +-()f ⎠ only made the assumption of being in hydrostatic equilibrium. Below, we will discuss the potential origin of the cloud. where all parameters are considered in the comoving frame, Given that we use the IC/BLR process to model the high- obs ¢ ¢ energy component of the spectrum, the cloud–jet interaction implying ttfj=+d f ()1 zred , and the timescale tvr0 = dj c 0 should take place within the BLR. We set the emission region is related to the cloud’s scale height r¢. There is no 0 close to the outer edge of the BLR, allowing the IC/BLR observational constraint on the latter value. We know that the process to operate, while the absorption at γ-rays is kept low. ’ ¢ ’ cloud s scale height r0 must be smaller than the cloud s radius Hence, the cloud could originate from the BLR itself. Rc¢. We tested a few values and found that the value related to From the above modeling, we can deduce that the electron fi the scale height given in Table 1 gives the best t. A larger density at the beginning of the flare is nj,e,min=2.32× 4 −3 4 −3 scale height than the one used underpredicts the fluxes, while a 10 cm , which rises to nj,e,max=4.0×10 cm at the peak smaller scale height produces a narrow peak, which is also of the flare. With Equation (5), a=0.1, and the parameters inconsistent with the observations. given in Table 1, we can calculate the number of particles 54 In order to account for the changing peak energy of the IC in the cloud to be Nc¢ =´2.34 10 or a mass of Mc = 30 component, we also change the electron spectral index. Due to Nc¢mMp =´3.9 10 g ~ 0.1% . The speed of the cloud ( ) 81- a lack of constraints, we assume a linear change as is, Equation 3 , vc¢ =´5.12 10 cm s , and its radius, Equation (4), R¢ =´1.3 1015 cm. The average particle ttt-- c ff∣∣ n¢ =´83- st()=+D s s .23() density in the cloud is, thus, c 2.54 10 cm . The scale t 14 f height of the cloud is r0¢ =´1.6 10 cm. This value, along with Equation (5), Equation (13), and an integration of Δs ( ) With being negative see Table 1 , the injection spectrum Equation (12), implies a temperature hardens until the maximum of the flare and subsequently returns to the pre-flare value. We further assume that the bulk Gm2 N¢ T¢ = pc Lorentz factor of the emission region is constant. c 6arctankrRrB 00¢¢¢[(cc )- ( Rr¢¢ 0 )] The resulting model spectra are shown in Figure 4. The fitof ~ 0.5 K. 24 the pre-flare and high-state spectra (black and red curves) is () quite good, taking into account that we do not aim for a precise This temperature is clearly too low, since a gas cloud cannot fit. The colored spectra show the evolution of the spectrum become colder than the cosmic microwave background. from the low state toward the maximum in roughly 10 day Additionally, standard parameters for BLR clouds suggest − − intervals. The lack of evidence of a broken power-law spectrum a radius of ∼1013 cm and an average density of 109 11 cm 3 in the X-ray domain gives us confidence that the seemingly (Dietrich et al. 1999; Peterson 2006). Hence, the size of our poor fit at low X-ray energies is not a big concern. The upturn model cloud is too large, while the density is too low. of the model curves around 100 MeV is due to a change in the However, all of these parameters were derived under the cooling behavior at these and higher energies, as shown in assumption that the entire cloud is devoured by the jet, and this Figure 5(b). At low energies, the cooling is dominated by the does not include potential higher density regions responsible IC/BLR process, but reduces for electron Lorentz factors for the fast but bright flares on top of the long-term trend. These γ>100 due to the Klein–Nishina effect. This hardens the higher density regions would exhibit higher temperatures, electron distribution, as can be seen in Figure 5(a), where we likely raising the temperature of the entire cloud. Additionally, show the underlying particle distribution of each photon the collision of the cloud with the jet might induce a shock spectrum of Figure 4. For electron Lorentz factors γ>104, wave running through the former (e.g., Poludnenko et al. 2002; synchrotron cooling becomes dominant, which is however Araudo et al. 2009), which could lead to the ejection of cloud unimportant for the present study, since we do not consider material away from the interaction site. Then, our estimate is electrons with these energies. Since neither the BLR nor the only a lower limit on the matter content of the cloud, and the magnetic field is assumed to vary, the electron cooling term is particle number, and hence the temperature, could be constant in time. The small wiggles in the γ-ray spectra, the significantly higher. Furthermore, we assumed that the particle distributions, and the cooling term are due to numerical hydrostatic equilibrium is solely mediated by an isothermal inaccuracies. gas. If the cloud contains a significant magnetic field, it will The resulting model light curves are shown as thick red lines stabilize the cloud even if the temperature is exceeding the in Figure 3. We present the light curves with a logarithmic isothermal temperature derived above. y-axis in order to highlight the significant change in flux and Although these considerations could lead to a density and the details of the theoretical light-curve evolution. We model temperature of the cloud that more closely resemble parameters γ R fl the -ray, X-ray, and optical band. We ignore the other of BLR clouds, it does not in uence our estimate of the size Rc¢, optical/UV bands, since the R band is the most detailed which solely depends on the speed of the cloud. Since we synchrotron light curve, and the constant color implies that the assumed Keplerian motion of the cloud, the size of the cloud behavior in the other bands is very similar. The model of the depends inversely on the square root of the distance from the general evolution of the flare is very good in all energy bands. black hole. Hence, the size of the cloud can be reduced, if the

100 A.2. Cloud ablation by a relativistic jet and the extended flare in CTA 102 in 2016 and 2017 The Astrophysical Journal, 851:72 (9pp), 2017 December 20 Zacharias et al. ablation takes place farther away from the black hole. Although ORCID iDs this could bring the size closer to the BLR cloud parameters, a M. Zacharias https://orcid.org/0000-0001-5801-3945 BLR cloud can be excluded, since our model is already placed M. Böttcher https://orcid.org/0000-0002-8434-5692 close to the outer edge of the BLR. In such a case, the high- J.-P. Lenain https://orcid.org/0000-0001-7284-9220 / energy component cannot be due to IC BLR, and more likely A. Wierzcholska https://orcid.org/0000-0003-4472-7204 hadronic scenarios need to be invoked. Assuming that a shock at parsec-scale distance from the black hole can efficiently References accelerate protons, the flare could be proton induced, since the cloud provides the jet with the same amount of protons as Acero, F., Ackermann, M., Ajello, M., et al. 2015, ApJS, 218, 23 fl Acero, F., Ackermann, M., Ajello, M., et al. 2016, ApJS, 223, 26 electrons. We will elaborate on a hadronic scenario for the are Araudo, A. T., Bosch-Ramon, V., & Romero, G. E. 2009, A&A, 503, 673 elsewhere. Araudo, A. T., Bosch-Ramon, V., & Romero, G. E. 2010, A&A, 522, A97 The constraint on the maximum electron Lorentz factor is Arnaud, K. A. 1996, in ASP Conf. Ser. 101, Astronomical Data Analysis particularly strong from the shape of the synchrotron spectrum. Software and Systems V, ed. G. H. Jacoby & J. Barnes (San Francisco, CA: fi ∼ ASP), 17 Hence, the cutoff of the IC component is xed at 20 GeV, Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, ApJ, 697, 1071 which does not even take into account absorption by the EBL. Bachev, R., Popov, V., Strigachev, A., Semkov, E., et al. 2017, MNRAS, If the spectrum of CTA 102 is indeed mainly shaped by the 417, 2216 leptonic model as described here, CTA 102 cannot be detected Blandford, R., & Königl, A. 1979, ApL, 20, 15 Blandford, R., & Rees, M. J. 1974, MNRAS, 169, 395 at very high-energy γ-rays (E > 100 GeV) by ground-based Bosch-Ramon, V. 2015, A&A, 575, A109 Cherenkov experiments. Bosch-Ramon, V., Perucho, M., & Barkov, M. V. 2012, A&A, 539, A69 In summary, we showed that the prolonged and strong Burrows, D. N., Hill, J. E., Nousek, J. A., & Kennea, J. A. 2005, SSRv, 120, 165 activity of the FSRQ CTA 102 could have been caused by the de la Cita, V. M., del Palacio, S., Bosch-Ramnon, V., et al. 2017, A&A, full or partial ablation of a gas cloud colliding with the jet. 604, A39 From the assumption of hydrostatic equilibrium of an Dietrich, M., Wagner, S. J., Courvoisier, T. J.-L., Bock, H., & North, P. 1999, isothermal gas with the gravity of the cloud, we derived the A&A, 351, 31 fl Diltz, C., & Böttcher, M. 2014, JHEAp, 1, 63 density structure of the cloud. This structure is re ected in the Franceschini, A., Rodighiero, G., & Vaccari, M. 2008, A&A, 487, 837 injection of material ablated by the jet causing the several- Fromm, C. M., Perucho, M., Ros, E., et al. 2011, A&A, 531, A95 month-long outburst. Our model light curves are in good Gehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611, 1005 Ghisellini, G., Celotti, A., Fossati, G., Maraschi, L., & Comastri, A. 1998, agreement with the observations. The model parameters MNRAS, 301, 451 suggest that the cloud was not fully ablated, and much of the Giommi, P., Blustin, A. J., Capalbi, M., et al. 2006, A&A, 456, 911 material might have been lost during the collision. Hauser, M., Möllenhoff, C., Pühlhofer, G., et al. 2004, AN, 325, 659 Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005, A&A, 440, 775 Klein, R. I., McKee, C. F., & Colella, P. 1994, ApJ, 420, 213 The authors are grateful for fruitful discussions with Heike Komissarov, S. S. 1994, MNRAS, 269, 394 Prokoph, Moritz Hackstein, and Chris Diltz, as well as for a Larionov, V. M., Villata, M., Raiteri, C. M., et al. 2016, MNRAS, 461, 3047 constructive report by the anonymous referee. Lister, M. L., Aller, M. F., Aller, H. D., et al. 2016, AJ, 152, 12 The work of M.Z. and M.B. is supported through the South Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, ApJ, 461, 396 ( ) Perucho, M., Bosch-Ramon, V., & Barkov, M. V. 2017, A&A, 606, A40 African Research Chair Initiative SARChI of the South Perucho, M., Marti, J. M., Laing, R. A., & Hardee, P. E. 2014, MNRAS, African Department of Science and Technology (DST) and 441, 1488 National Research Foundation.7 F.J. and S.J.W. acknowledge Peterson, B. M. 2006, LNP, 693, 77 Pian, E., Falomo, R., & Treves, A. 2005, MNRAS, 361, 919 support by the German Ministry for Education and Research Poludnenko, A. Y., Frank, A., & Blackman, E. G. 2002, ApJ, 576, 832 (BMBF) through Verbundforschung Astroteilchenphysik grant Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103 05A11VH2. J.-P.L. gratefully acknowledges CC-IN2P3 (cc. Schramm, K.-J., Borgeest, U., Camenzind, M., et al. 1993, A&A, 278, 391 in2p3.fr) for providing a significant amount of the computing Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337 Zacharias, M., & Schlickeiser, R. 2013, ApJ, 777, 109 resources and services needed for this work. A.W. is supported Zamaninasab, M., Clausen-Brown, E., Savolainen, T., & Tchekhovskoy, A. by the Foundation for Polish Science (FNP). 2014, Natur, 510, 126

7 Any opinion, ﬁnding and conclusion, or recommendation expressed in this material is that of the authors, and the NRF does not accept any liability in this regard.

101 A. Selected publications

A&A 600, A89 (2017) DOI: 10.1051/0004-6361/201629427 Astronomy c ESO 2017 & Astrophysics

Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113

? 1 2 − 3, 4, 5 3 6, 5, 10 H.E.S.S. Collaboration , H. Abdalla , A. Abramowski , F. Aharonian , F. Ait Benkhali , A. G. Akhperjanian †, T. Andersson , E. O. Angüner7, M. Arrieta15, P. Aubert23, M. Backes8, A. Balzer9, M. Barnard1, Y. Becherini10, J. Becker Tjus11, D. Berge12, S. Bernhard13, K. Bernlöhr3, R. Blackwell14, M. Böttcher1, C. Boisson15, J. Bolmont16, P. Bordas3, F. Brun25, P. Brun17, M. Bryan9, T. Bulik18, M. Capasso27, J. Carr19, S. Casanova20, 3, M. Cerruti16, N. Chakraborty3, R. Chalme-Calvet16, R. C. G. Chaves21, A. Chen22, J. Chevalier23, M. Chrétien16, S. Colafrancesco22, G. Cologna24, B. Condon25, J. Conrad26, C. Couturier16, Y. Cui27, I. D. Davids1, 8, B. Degrange28, C. Deil3, J. Devin17, P. deWilt14, L. Dirson2, A. Djannati-Ataï29, W. Domainko3, A. Donath3, L. O’C. Drury4, G. Dubus30, K. Dutson31, J. Dyks32, T. Edwards3, K. Egberts33, P. Eger3, J.-P. Ernenwein20, S. Eschbach34, C. Farnier26, 10, S. Fegan28, M. V. Fernandes2, A. Fiasson23, G. Fontaine28, A. Förster3, S. Funk34, M. Füßling35, S. Gabici29, M. Gajdus7, Y. A. Gallant17, T. Garrigoux1, G. Giavitto35, B. Giebels28, J. F. Glicenstein18, D. Gottschall27, A. Goyal36, M.-H. Grondin25, D. Hadasch13, J. Hahn3, M. Haupt35, J. Hawkes14, G. Heinzelmann2, G. Henri30, G. Hermann3, O. Hervet15, 42, A. Hillert3, J. A. Hinton3, W. Hofmann3, C. Hoischen33, M. Holler28, D. Horns2, A. Ivascenko1, A. Jacholkowska16, M. Jamrozy36, M. Janiak32, D. Jankowsky34, F. Jankowsky24, M. Jingo22, T. Jogler34, L. Jouvin29, I. Jung-Richardt34, M. A. Kastendieck2, K. Katarzynski´ 37, U. Katz34, D. Kerszberg16, B. Khélifi29, M. Kieffer16, J. King3, S. Klepser35, D. Klochkov27, W. Klu´zniak32, D. Kolitzus13, Nu. Komin22, K. Kosack18, S. Krakau11, M. Kraus34, F. Krayzel23, P. P. Krüger1, H. Laffon25, G. Lamanna23, J. Lau14, J.-P. Lees23, J. Lefaucheur15, V. Lefranc18, 29 25 16, 33 7 18 3 3 35 3 A. Lemière , M. Lemoine-Goumard , J.-P. Lenain ∗, E. Leser , T. Lohse , M. Lorentz , R. Liu , R. López-Coto , I. Lypova , V. Marandon , A. Marcowith17, C. Mariaud28, R. Marx3, G. Maurin23, N. Maxted14, M. Mayer7, P. J. Meintjes38, M. Meyer26, A. M. W. Mitchell3, R. Moderski32, M. Mohamed24, L. Mohrmann34, K. Morå26, E. Moulin18, T. Murach7, M. de Naurois28, F. Niederwanger13, J. Niemiec20, L. Oakes7, P. O’Brien31, H. Odaka3, S. Öttl13, S. Ohm35, M. Ostrowski36, I. Oya35, M. Padovani17, M. Panter3, R. D. Parsons3, M. Paz Arribas7, N. W. Pekeur1, G. Pelletier30, C. Perennes16, P.-O. Petrucci30, B. Peyaud18, S. Pita29, H. Poon3, D. Prokhorov10, H. Prokoph10, G. Pühlhofer27, 29, 10 24 34 13 13 17 3 3, 39 4, M. Punch , A. Quirrenbach , S. Raab , A. Reimer , O. Reimer , M. Renaud , R. de los Reyes , F. Rieger , C. Romoli ∗, 23 14 32 15 6, 5 40 23, 27 27 S. Rosier-Lees , G. Rowell , B. Rudak , C. B. Rulten , V. Sahakian , D. Salek , D. A. Sanchez ∗, A. Santangelo , M. Sasaki , R. Schlickeiser11, F. Schüssler18, A. Schulz35, U. Schwanke7, S. Schwemmer24, M. Settimo16, A. S. Seyffert1, N. Shafi22, I. Shilon34, R. Simoni9, 15 1 26 2 36 8 33, 35 34, 35 1 H. Sol , F. Spanier , G. Spengler , F. Spies , Ł. Stawarz , R. Steenkamp , C. Stegmann , F. Stinzing †, K. Stycz , I. Sushch , 16 29 4, 29 3 34 2 20 3 41 J.-P. Tavernet , T. Tavernier , A. M. Taylor ∗, R. Terrier , L. Tibaldo , D. Tiziani , M. Tluczykont , C. Trichard , R. Tuffs , Y. Uchiyama , D. J. van der Walt1, C. van Eldik34, B. van Soelen38, G. Vasileiadis17, J. Veh34, C. Venter1, A. Viana3, P. Vincent16, J. Vink9, F. Voisin14, H. J. Völk3, T. Vuillaume23, Z. Wadiasingh1, S. J. Wagner24, P. Wagner7, R. M. Wagner26, R. White3, A. Wierzcholska20, P. Willmann34, 34 18 3 31 ,28 24 32 15 28 34 A. Wörnlein , D. Wouters , R. Yang , V. Zabalza , D. Zaborov∗ , M. Zacharias , A. A. Zdziarski , A. Zech , F. Zefi , A. Ziegler , N. Zywucka˙ 36, and LAT Collaboration, M. Ackermann43, M. Ajello44, L. Baldini45, 46, G. Barbiellini47, 38, R. Bellazzini49, R. D. Blandford46, R. Bonino50, 51, J. Bregeon52, P. Bruel28, R. Buehler43, G. A. Caliandro46, 53, R. A. Cameron46, M. Caragiulo54, 55, P. A. Caraveo56, E. Cavazzuti57, C. Cecchi58, 59, J. Chiang46, G. Chiaro60, S. Ciprini57, 58, J. Cohen-Tanugi52, F. Costanza55, S. Cutini57, 58, F. D’Ammando61, 62, F. de Palma45, 63, R. Desiante64, 40, N. Di Lalla49, M. Di Mauro46, L. Di Venere54, 55, B. Donaggio65, C. Favuzzi54, 55, W. B. Focke46, P. Fusco54, 55, F. Gargano55, D. Gasparrini57, 58, N. Giglietto54, 55, F. Giordano54, 55, M. Giroletti61, L. Guillemot66, 67, S. Guiriec68, 69, D. Horan28, G. Jóhannesson70, T. Kamae71, S. Kensei72, D. Kocevski69, S. Larsson73, 74, J. Li75, F. Longo47, 48, F. Loparco54, 55, M. N. Lovellette76, P. Lubrano58, S. Maldera50, A. Manfreda49, M. N. Mazziotta55, P. F. Michelson46, T. Mizuno77, M. E. Monzani46, A. Morselli78, M. Negro50, 51, E. Nuss52, M. Orienti61, E. Orlando46, D. Paneque79, J. S. Perkins69, M. Pesce-Rollins49, 57, F. Piron52, G. Pivato49, T. A. Porter46, G. Principe80, S. Rainò54, 55, M. Razzano49, D. Simone55, E. J. Siskind81, F. Spada49, P. Spinelli54, 55, J. B. Thayer46, D. F. Torres76, 82, E. Torresi83, E. Troja69, 84, G. Vianello48, and K. S. Wood77 (Affiliations can be found after the references)

Received 29 July 2016 / Accepted 1 December 2016

ABSTRACT

Context. The addition of a 28 m Cherenkov telescope (CT5) to the H.E.S.S. array extended the experiment’s sensitivity to lower energies. The lowest energy threshold is obtained using monoscopic analysis of data taken with CT5, providing access to gamma-ray energies below 100 GeV for small zenith angle observations. Such an extension of the instrument’s energy range is particularly beneﬁcial for studies of active galactic nuclei with soft spectra, as expected for those at a redshift 0.5. The high-frequency peaked BL Lac objects PKS 2155 304 (z = 0.116) and PG 1553+113 (0.43 < z < 0.58) are among the brightest objects≥ in the gamma-ray sky, both showing clear signatures of gamma-ray− absorption at E > 100 GeV interpreted as being due to interactions with the extragalactic background light (EBL). Aims. The aims of this work are twofold: to demonstrate the monoscopic analysis of CT5 data with a low energy threshold, and to obtain accurate measurements of the spectral energy distributions (SED) of PKS 2155 304 and PG 1553+113 near their SED peaks at energies 100 GeV. Methods. Multiple observational campaigns of PKS 2155 304 and PG− 1553+113 were conducted during 2013 and 2014 using the≈ full H.E.S.S. II instrument (CT1–5). A monoscopic analysis of the data− taken with the new CT5 telescope was developed along with an investigation into the systematic uncertainties on the spectral parameters which are derived from this analysis. Results. Using the data from CT5, the energy spectra of PKS 2155 304 and PG 1553+113 were reconstructed down to conservative threshold energies of 80 GeV for PKS 2155 304, which transits near zenith, and− 110 GeV for the more northern PG 1553+113. The measured spectra, well − ? Corresponding author: H.E.S.S. and LAT Collaborations, e-mail: [email protected] † Deceased.

Article published by EDP Sciences A89, page 1 of 13

102 A.3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 − A&A 600, A89 (2017)

fitted in both cases by a log-parabola spectral model (with a 5.0σ statistical preference for non-zero curvature for PKS 2155 304 and 4.5σ for PG 1553+113), were found consistent with spectra derived from contemporaneous Fermi-LAT data, indicating a sharp break in the− observed spectra of both sources at E 100 GeV. When corrected for EBL absorption, the intrinsic H.E.S.S. II mono and Fermi-LAT spectrum of PKS 2155 304 was found to show significant≈ curvature. For PG 1553+113, however, no significant detection of curvature in the intrinsic spectrum could be found− within statistical and systematic uncertainties. Key words galaxies: active – BL Lacertae objects: individual: PKS 2155-304 – BL Lacertae objects: individual: PG 1553+113 – gamma rays: galaxies

1. Introduction The HBL object PG 1553+113 was first announced as a VHE gamma-ray source by H.E.S.S. (Aharonian et al. 2006b) The very high energy (VHE, E & 100 GeV) gamma-ray exper- and independently and almost simultaneously confirmed by iment of the High Energy Stereoscopic System (H.E.S.S.) con- MAGIC using observations from 2005 (Albert et al. 2007). The sists of five imaging atmospheric Cherenkov telescopes (IACTs) H.E.S.S. I measurements (Aharonian et al. 2008) yielded a pho- located in the Khomas Highland of Namibia (23 16 18 S, ton index Γ = 4.5 0.3 0.1 above 225 GeV. At high ener- ◦ 0 00 ± stat ± syst 16◦3000000 E), 1835 m above sea level. From January 2004 to gies (HE, 100 MeV < E < 300 GeV) the source was detected by October 2012, the array was operated as a four telescope instru- Fermi-LAT with a photon index of 1.68 0.03 (Abdo et al. 2009, ment (H.E.S.S. phase I). The telescopes, CT1–4, are arranged in 2010), making PG 1553+113 an active± galactic nucleus (AGN) a square formation with a side length of 120 m. Each of these with one of the largest HE-VHE spectral breaks observed and a telescopes has an effective mirror surface area of 107 m2, a field hint for long-term gamma-ray flux oscillation (Ackermann et al. of view of 5◦ in diameter, capable of detecting cosmic gamma 2015). The redshift of PG 1553+113 is constrained by UV obser- rays in the energy range 0.1–100 TeV (Aharonian et al. 2006a). vations to the range 0.43 < z . 0.58 (Danforth et al. 2010). The In October 2012 a fifth telescope, CT5, placed at the centre of first upper-limits of z < 0.69 (pre-Fermi-LAT) Mazin & Goebel the original square, started taking data. This set-up is referred (2007) and more recently (post-Fermi-LAT) z < 0.61 on the to as H.E.S.S. phase II, or H.E.S.S. II. With its effective mir- source redshift have been obtained Aliu et al.(2015) using TeV ror surface close to 600 m2 and a fast, finely pixelated camera data and of z < 0.53 by Biteau & Williams(2015) using also (Bolmont et al. 2014), CT5 potentially extends the energy range GeV data. Assuming that the difference in spectral indices be- covered by the array down to energies of 30 GeV. tween the HE and VHE regimes is imprinted by the attenua- ∼ In this study, we focus on obtaining high statistic results tion by the extragalactic background light, the redshift was constrained to the range z = 0.49 0.04 (Abramowski et al. 2015). with observations of the high-frequency peaked BL Lac objects ± PKS 2155 304 and PG 1553+113. These blazars are among the This paper reports on the first observations of PKS 2155 304 + − brightest objects− in the VHE gamma-ray sky. Furthermore, the and PG 1553 113 conducted in 2013 and 2014 using the spectra of both these blazars exhibit signatures of gamma-ray H.E.S.S. II instrument (CT5) in monoscopic mode. A descrip- absorption at energies E 100 GeV, due to interactions with the tion of the analysis for both AGNs, using data from this instru- extragalactic background∼ light (EBL). ment, is provided. Systematic errors associated with our results are also estimated. Particular emphasis is placed on the spec- PKS 2155 304 is a high-frequency peaked BL Lac (HBL) tral measurements at low energies and their connection with − object at z = 0.116 (Ganguly et al. 2013; Farina et al. 2016). the Fermi-LAT measurements. Using the H.E.S.S. II mono and This source is located in a galaxy poor cluster (Falomo et al. Fermi-LAT results, the implications on intrinsic source spectrum 1993) and the host galaxy is resolved (Kotilainen et al. 1998). are considered. It was first discovered as a high energy emitter by the ffi HEAO 1 X-ray satellite (Gri ths et al. 1979; Schwartz et al. 2. The H.E.S.S. II experiment 1979). Gamma-ray emission in the energy range 30 MeV to 10 GeV was detected from this blazar by the EGRET instrument The H.E.S.S. II experiment is the first hybrid Cherenkov instru- on board the Compton Gamma Ray Observatory (Vestrand et al. ment and has the ability to take data in different modes. The 1995). The first detection in the VHE range was attained in 1996 H.E.S.S. II system triggers on events detected either by CT5 only by the University of Durham Mark 6 Telescope, with a statis- (mono) or by any combination of two or more telescopes (stereo, tical significance of 6.8σ (Chadwick et al. 1999). Starting from CT5 plus at least one of CT1–4, or at least two of CT1–4). The 2002 the source was regularly observed with H.E.S.S., with the field of view of CT5 is 3.2◦ in diameter, smaller than that for first detection based on the 2002 data subsequently published CT1–4. Consequently, not all stereo triggers include CT5. The with just one telescope of H.E.S.S. phase I (Aharonian et al. standard observation mode of H.E.S.S. II is to collect both mono 2005). After completion of the array, this source was detected and stereo events during the same observation run. in stereoscopic mode in 2003 with high significance (>100σ) at The analysis of CT1–5 stereo data provides a lower energy energies greater than 160 GeV (Aharonian et al. 2005). Strong threshold, better hadron rejection and better angular resolution flux variability with multiple episodes of extreme flaring ac- than with CT1–4 only. The analysis of H.E.S.S. II mono events tivity in the VHE band were reported (Aharonian et al. 2007; potentially provides a factor of approximately four lower energy H.E.S.S. Collaboration et al. 2010; Aleksic´ et al. 2012a). A pho- threshold than CT1–5 stereo. However, the absence of stereo- ton index (Γ, describes the spectral shape of the photon en- scopic constraints makes the rejection of hadronic events more Γ ergy distribution, dN/dE E− .) of 3.53 0.06stat 0.10syst difficult, leading to a larger background and reduced signal-to- was obtained from analysis∝ of observations± during a± low flux background ratio at the analysis level. The low energy threshold state (2005–2007) above 200 GeV (H.E.S.S. Collaboration et al. of H.E.S.S. II mono implies high event rates, and thus small sta- 2010). For average and high flux states the presence of curvature tistical uncertainties on the background, which leads to tight re- or a cut-off was favoured from the spectral fit analysis carried quirements for the accuracy of background subtraction. The an- out (H.E.S.S. Collaboration et al. 2010). gular reconstruction of the monoscopic analysis is significantly

A89, page 2 of 13

103 A. Selected publications

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis less precise than that obtained in the stereoscopic mode, leading – CT5 trigger rate between 1200 and 3000 Hz (its nominal to a reduction of the sensitivity for point-like sources. value depends on the observed field of view and zenith angle) Nevertheless, the H.E.S.S. II mono analysis provides new and stable within 10% during a run; ± opportunities to probe astronomy at energies <100 GeV for – Telescope tracking functioning normally; southern sources, which are complementary to satellite experiments (e.g. Fermi Large Area Telescope, LAT) and to northern 3.3. Data analysis hemisphere facilities such as MAGIC and VERITAS which can detect northern sources below 100 GeV (Aleksic´ et al. 2015a; The data sets were processed with the standard H.E.S.S. analysis Abeysekara et al. 2015). The low energy threshold provided by software using the Model reconstruction (de Naurois & Rolland H.E.S.S. II mono is, consequently, particularly beneficial for 2009) which was recently adapted to work with monoscopic studies of bright variable objects such as gamma-ray bursts and events (Holler et al. 2015). The Model reconstruction performs a AGNs out to high redshifts (z & 0.5), along with associated spec- likelihood fit of the air shower image to a semi-analytical model tral features introduced into the spectra through gamma-ray in- of an average gamma-ray shower parameterised as a function teractions with the extragalactic background light (EBL). of energy, primary interaction depth, impact distance and di- The full performance characterization of the CT1–5 system rection. Gamma-like candidate events are selected based on the will be provided in a forthcoming publication. value of the goodness-of-fit variable and the reconstructed primary interaction depth. In addition, events with an estimated error in direction reconstruction >0.3 are rejected. The low energy 3. H.E.S.S. II mono observations and analysis ◦ threshold is controlled with a dedicated variable NSB Goodness, 3.1. H.E.S.S. II observations which characterises the likelihood of accidentally triggering on fluctuations due to the night sky background. Two cut configu- PKS 2155 304 was monitored with H.E.S.S. II regularly − rations were defined for this analysis, loose and standard, with for two consecutive years: in 2013 (from Apr. 21 to Nov. different settings for the NSB Goodness cut. Loose cuts provide 5, 2013, MJD 56 403–56 601); and 2014 (May 28–Jun. 9, the lowest energy threshold, but may lead to a significant level 2014, MJD 56 805–56 817). PG 1553+113 was observed with of systematic errors in the background subtraction when applied H.E.S.S. II between May 29 and Aug. 9, 2013 (MJD 56 441– to high statistics datasets. Standard cuts provide a better control 56 513). Most of the observations were taken using the full over the background subtraction at the cost of increased thresh- H.E.S.S. II array. This paper only reports on the monoscopic old. The event selection cuts, except for the NSB Goodness cut, analysis of this data, which provides the lowest achievable en- were optimised to maximise the discovery potential for a point ergy threshold. source with a photon index of 3.0 observed at a zenith angle of H.E.S.S. data taking is organised in 28 min blocks, called 18◦ for 5 h. The optimized analysis provides an angular resolu- runs. Observations are usually taken in wobble mode, with the tion of 0.15 (68% containment radius) at 100 GeV and energy ≈ ◦ camera’s field of view centred at a 0.5◦ or 0.7◦ offset from the resolution of 25%. For photon indices harder than 3.0, standard source position, in either direction along the right-ascension or cuts provide a≈ better sensitivity than loose cuts. declination axis. Only runs for which the source position is lo- The background subtraction is performed using the stan- cated between 0.35◦ and 1.2◦ off-axis from camera centre are dard algorithms used in H.E.S.S.– the ring background method used in the present analysis. Runs with non-standard wobble (for sky maps) and the reflected-region background method offests were taken during the commissioning phase to assess the (Berge et al. 2007, with multiple off-source regions, for spec- performance of the instrument. This is to ensure that the source tral measurements). The ring background method uses a zenith- is well within the field of view and allow background subtrac- dependent two-dimensional acceptance model, an inner ring ration using the reflected-region background method (Berge et al. dius of 0.3◦ and outer radius of 0.6◦, and top-hat smoothing 2007). radius of 0.1◦. The acceptance model, which describes the observed distribution of background events in the camera’s field of view in absence of gamma-ray sources, is obtained from 3.2. Data quality selection the data itself, using background events outside of a radius of To ensure the quality of the AGN data sets for the H.E.S.S. II 0.3◦ from any known VHE gamma-ray source (for this analysis, mono analysis the several run quality criteria were applied. PKS 2155 304 and PG 1553+113). The reflected-region back- − ground method uses an on-source region radius of 0.122◦, which – Stable clear sky conditions according to the telescope ra- corresponds to an angular distance cut θ2 < 0.015 deg2. The diometers. We use the narrow field-of-view radiometers in- number of off-source regions was adjusted on a run-by-run ba- stalled on the CT1–4 telescopes, requesting radiometer tem- sis so as to always use the maximum possible number of them, perature to be less than 20 C and stable during the run given the wobble angle. For instance, for a wobble angle of 0.5 − ◦ ◦ within 3 ◦C; nine off-source regions were used. A simple acceptance model, – Relative± humidity <90%; which only corrects for linear gradients in the acceptance, is – Run duration >5 min and live time fraction >90%. A run may used with this method. The significance of the excess after back- be interrupted due to an automated target-of-opportunity ob- ground subtraction is determined using the method described servation of a transient source, deteriorating weather condi- by Li & Ma(1983). Spectral measurements are obtained us- tions, or a technical issue; ing the forward folding technique (Piron et al. 2001), applied to – At least 90% of pixels in CT5 are active (pixels can be tem- the excess events observed with the reflected-region background porarily switched off due to a star in the field of view or re- method. The energy threshold for the spectral fit is defined as the moved from the data due to bad calibration); energy at which the effective area reaches 15% of its maximum – CT5 trigger in standard configuration pixel/sector threshold value, in line with the definition previously adopted in H.E.S.S. =4/2.5, see Aharonian et al.(2006a) for a definition of the analysis (H.E.S.S. Collaboration et al. 2014a). Such a definition trigger pattern; ensures that the systematic uncertainties in the analysis are kept

A89, page 3 of 13

104 A.3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 − A&A 600, A89 (2017)

103 H.E.S.S. H.E.S.S. Mean : 0.00 Mean : -0.07 entries -29° entries -29° E < 100 GeV Sigma: 1.15 Sigma: 1.37 2 102 10

-30° PKS 2155-304 -30° PKS 2155-304 Declination (J2000) Declination (J2000) 4000 300 3500 250 3000 10 200 10 2500 ° ° 150 -31 PSF 2000 -31 PSF 100 1500 50 1000 500 0 -50 -32° 0 1 -32° 1

22h05m00s 22h00m00s 21h55m00s 21h50m00s 0 10 20 30 40 50 22h05m00s 22h00m00s 21h55m00s 21h50m00s -5 0 5 10 15 20 Right Ascension (J2000) significance Right Ascension (J2000) significance

300 3500 H.E.S.S. H.E.S.S. counts counts 250 3000 PKS 2155-304 PKS 2155-304

2500 200 E < 100 GeV

2000 150 1500 100 1000 50 500

0 0 0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3 θ2 (deg2) θ2 (deg2)

Fig. 1. Top:(left) excess map of events observed in the direction of Fig. 2. Top: PKS 2155 304 excess map (left) and significance distri- PKS 2155 304 using the H.E.S.S. II mono analysis (2013–2014 data). bution (right) for events− with reconstructed energy E < 100 GeV The inset represents− the point spread function of the instrument obtained (H.E.S.S. II mono analysis, 2013–2014 data). Bottom: distribution of from simulations. The source position is indicated by a black dot. Right: θ2 (squared angular distance to PKS 2155 304) for gamma-like events. significance distribution that corresponds to the excess map (black his- − togram). The distribution obtained by excluding a circular region of 0.3◦ This offset can be attributed to the systematic errors on the tele- radius around the source is shown in red; the results of a Gaussian fit to 2 scope pointing. Outside the exclusion radius of 0.3◦ the signif- this distribution are also shown. Bottom: distribution of θ (squared an- icance distribution was found to be well fit by a Gaussian with gular distance to PKS 2155 304) for gamma-like events obtained with − σ = 1.149 0.004. This result indicates the presence of a system- the H.E.S.S. II mono analysis (filled histogram) in comparison with the ± normalised θ2 distribution for off-source regions (black points). The ver- atic effect in background subtraction, whose σsyst corresponds tical dashed line shows the limit of the on-source region. The energy to about 57% of the statistical errors (σstat equal one by con- threshold for this analysis is 80 GeV. struction). We here assume that the errors add in quadrature. A ≈ 2 √ value of σ = 1 + σsyst > 2 would then indicate the domi- under control. The H.E.S.S. II mono analysis was applied to nance of background subtraction errors. This effectively reduces all events that include CT5 data (ignoring information from q the observed excess significance from 42σ to 36σ1. This sys- CT1 4). ≈ − tematic effect is currently under investigation as part of a larger effort to understand the mono analysis performance. Repeating 4. Results the analysis using only events with reconstructed energy below 100 GeV leads to a 10σ (7.3σ) significance at the position of 4.1. PKS 2155 304 − PKS 2155 304 in the skymap (Fig.2). The significance distribution outside− the exclusion region has σ = 1.374 0.005, indi- The PKS 2155 304 data set, filtered as explained in Sect. 3.2, ± comprises 138− runs. The total live time of this data set is 56.0 h, cating that the background subtraction errors are slightly smaller 43.7 h taken in 2013 and 12.3 h taken in 2014. During these ob- than the statistical errors. Thus the source is confidently detected at E < 100 GeV. servations, the source zenith angle ranged from 7◦ to 60◦, with The distribution of θ2, the square of the angular difference a median value of 16◦. This data set was analysed using standard cuts as described in Sect. 3.3. The background event counts between the reconstructed shower position and the source posi- obtained for the off-source regions in each run (in the reflected- tion, is shown in the bottom panel of Fig.1 (filled histogram). A 43σ excess over the background (black crosses) is observed region background analysis) were used to perform an additional 2 2 test of the uniformity of the camera acceptance. This was done within the on-source region (θ < 0.015 deg ). The reconstructed spectrum of PKS 2155 304 obtained using a likelihood ratio test (LLRT), with the baseline hypothe- − sis that the event counts observed in all off-source regions come for 2013, and each of the observation years (2013 and from the same Poisson distribution, and a nested model allowing 2014), is shown in Figs.3 and4, respectively. For the for different mean values in each region. The results of this test full data set (2013+2014), a log-parabola model, dN/dE = Γ β log(E/E0) were consistent with an axially-symmetric camera acceptance. Φ0 (E/E0)− − · , better fits the data with respect to a sim- The sky map obtained for PKS 2155 304 using the ple power-law model with a log-likelihood ratio of 25 (i.e. 5σ). − The flux normalisation is found to be Φ0 = (5.11 0.15stat) H.E.S.S. II mono analysis is shown in the top-left panel of Fig.1. 10 2 1 1 2 ± × The analysis found that the source is detected with a significance 10− cm− s− TeV− at a decorrelation energy E0 = 156 GeV, of 42σ, with 4000 excess events. The corresponding distribu- ≈ ≈ 1 From this point forward, significance values are not corrected for this tion of the excess significance of all skymap bins is shown in effect, with the corrected values being quoted within brackets immedi- the top-right panel of Fig.1. The width of the observed excess is ately proceeding these uncorrected values. approximately compatible with the simulated point spread func- 2 For the log-parabola model, the decorrelation energy is the energy tion (PSF; shown in the inset on Fig.1). The best-fit position where the error on the flux is the smallest, that is where the confidence of the excess is found 32 10 from the target position. band butterfly is the narrowest in the graphical representation. 00 ± stat00 A89, page 4 of 13

105 A. Selected publications

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

103 ν [Hz] 13° 23 24 25 26 27 H.E.S.S. 10 10 10 10 10 Mean : -0.03 entries ]

-1 Sigma: 1.22 − 10 10 12° PKS 2155-304 102

[ erg s PG 1553+113

45 ν 10 Declination (J2000) 2500 ] L 11° -1 ν 2000 s −11 10

-2 1500 10 PSF 1000 10° 500 44 10 0 1

− −10 10 12 CT5 mono 10 16h00m00s 15h55m00s 15h50m00s -10 -5 0 5 10 15 20 25 30 35 − Right Ascension (J2000) significance dN/dE [ erg cm Fermi-LAT E>100MeV 11

2 10

E Fermi-LAT E>10GeV 10−12 43 Fermi-LAT E>50GeV 10 − 10 13 H.E.S.S. H.E.S.S. I (CT1-4) counts 2500 −13 − 10 10 1 1 10 102 103 PG 1553+113 2000 − 10 1 1 10 102 103 E [GeV] 1500

Fig. 3. Energy spectrum of PKS 2155 304 obtained from the 1000 H.E.S.S. II mono analysis (2013 data, shown− by blue circles with confidence band) in comparison with the contemporaneous Fermi-LAT data 500 0 with an energy threshold of 0.1 GeV (red triangles and confidence 0 0.05 0.1 0.15 0.2 0.25 0.3 band), 10 GeV (green band), and 50 GeV (purple band) and contem- θ2 (deg2) poraneous CT1–4 data (grey squares). In all cases the confidence bands represent the 1σ region. The right-hand y-axis shows the equivalent Fig. 5. Top:(Left) excess map of events observed in the direction of isotropic luminosity (not corrected for beaming or EBL absorption). PG 1553+113 using the H.E.S.S. II mono analysis (16.8 h live time). The inset compares the H.E.S.S. confidence band with the Fermi-LAT The source position is indicated by a black dot. Right: significance dis- catalogue data (3FGL, 1FHL and 2FHL, see Sect. 4.4.2). tribution that corresponds to the excess map. The meaning of the his- tograms and statistics data is the same as in Fig.1. Bottom: θ2 distribution for PG 1553+113. The meaning of the data shown is the same as in ν [Hz] Fig.1. The vertical dashed line shows the limit of the on-source region. 23 24 25 26 27 10 10 10 10 10 The energy threshold for this analysis is 100 GeV. ]

− -1 ≈ 10 9 PKS 2155-304

10 [ erg s ν

] −10 L -1 10 ν During the observations, the source zenith angle ranged between s -2 45 33◦ and 40◦, with a mean value of 35◦. The sky map obtained 10 for PG 1553+113 using the H.E.S.S. II mono analysis is shown −11 10 in the top-left panel of Fig.5. This analysis found that the source CT5 mono 2013 1044 is detected with a statistical significance of 27σ (21σ), with Fermi-LAT 2013 (E>100MeV) dN/dE [ erg cm −12 2500 excess events. 2 10 E CT5 mono 2014 ≈ 43 The best-fit position of the excess is found to be 36 12 Fermi-LAT 2014 (E>100MeV) 10 00 ± stat00 −13 from the target position, this shift is attributed to the system- 10 atic errors on the telescope pointing. The width of the observed − 10 1 1 10 102 103 excess is compatible with the simulated PSF within a 10% sys- E [GeV] tematic uncertainty on the PSF width. Fig. 4. SED of PKS 2155 304 separated into the 2013 and 2014 obser- . vation periods. Both the H.E.S.S.− II mono and contemporaneous Fermi- The significance distribution in the region outside of the 0 3◦ LAT data are shown. The bands represent the 1σ confidence region. exclusion radius is consistent with a normal distribution (top- right panel of Fig.5). The same holds true when the analysis is repeated in only a low energy bin, with a reconstructed energy range of 100–136 GeV. Within this energy bin, the source is de- with a photon index Γ = 2.63 0.07stat and a curvature parameter tected with a 10σ (8.2σ) significance (Fig.6). The significance β = 0.24 0.06 . The spectral± data points (blue filled circles) distribution outside the exclusion region has σ = 1.219 0.005 stat ± cover the± energy range from 80 GeV to 1.2 TeV (not including and 1.288 0.005, for the full energy range and the first energy ± upper limits). The spectral parameters obtained for the 2013 and bin, respectively, indicating presence of background subtraction 2014 data sets are given in Table1. The isotropic luminosity that errors at a level smaller than the statistical errors. corresponds to the measured SED is shown by the additional The θ2 distribution is shown in the bottom panel of Fig.5. y -axis on the right-hand side of the SED plots. A 27σ (21σ) excess over the background is observed within the on-source region (θ2 < 0.015 deg2). The reconstructed spectrum, 4.2. PG 1553+113 with a threshold of 110 GeV, is found to be well fit by a log- parabola (with a LLRT of 20 over the power-law model, Fig.7), The PG 1553+113 data set, filtered as explained in Sect. 3.2, with a photon index Γ = 2.95 0.23 at decorrelation energy ± stat comprises 39 runs (16.8 h live time), which were analysed us- E0 = 141 GeV, curvature parameter β = 1.04 0.31stat, and 9 ±2 1 1 ing loose cuts as described in Sect. 3.3. This analysis configura- differential flux Φ = (1.48 0.07 ) 10 cm s TeV− at 0 ± stat × − − − tion, providing lower energy threshold than standard cuts, is well E0. The spectral data points (blue filled circles) cover the energy suited for bright soft-spectrum sources, such as PG 1553+113. range from 110 GeV to 550 GeV (not including upper limits).

A89, page 5 of 13

106 A.3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 − A&A 600, A89 (2017)

Table 1. Spectral analysis results of H.E.S.S. II mono observations.

Source Year MJD Livetime E0 Φ0 Γ β 9 2 1 1 [h] [GeV] [10− cm− s− TeV− ] PKS 2155 304 2013 56 403–56 601 43.7 151 0.530 0.018 2.65 0.09 0.22 0.07 − ± stat ± stat ± stat 2014 56 805–56 817 12.3 177 0.532 0.029stat 2.82 0.13stat 0.16 0.10stat 2013+2014 56 403–56 817 56.0 156 0.511 ± 0.015 2.63 ± 0.07 0.24 ± 0.06 ± stat ± stat ± stat PG 1553+113 2013 56 441–56 513 16.8 141 1.48 0.07 2.95 0.23 1.04 0.31 ± stat ± stat ± stat

Notes. For both blazars, the observational period is provided along with the spectral parameters: decorrelation energy E0; differential flux at the decorrelation energy Φ0; photon index Γ; and curvature parameter β. These three parameters describe the log-parabola fit to the spectra.

13° (de Naurois & Rolland 2009) using the loose cuts H.E.S.S. Mean : -0.03 100 < E < 136 GeV entries (Aharonian et al. 2006a) to ensure a low energy threshold. Sigma: 1.29 ° In total, data sets of 27.2 h of live time for PKS 2155 304 and 12 102 9.0 h for PG 1553+113 have been analysed, yielding− a signif- PG 1553+113

Declination (J2000) 500 icance of 46σ for PKS 2155 304 and 9.0σ for PG 1553+113. 11° 400 − 300 10 We note that the live times differ from the corresponding mono PSF 200 analysis live times due to different run qualities and observation 10° 100 0 schedules for the different instruments. For each data set the -100 1 spectrum is well fitted by a power-law model and the resulting 16h00m00s 15h55m00s 15h50m00s -10 -5 0 5 10 15 forward-folded data points for PKS 2155 304 (2013 data) Right Ascension (J2000) significance and PG 1553+113 are shown on Figs.3 and−7, respectively.

700 The CT1–4 results for PKS 2155 304 were found to be in H.E.S.S. − counts 600 PG 1553+113 excellent agreement with the H.E.S.S. II mono results. Due 100 < E < 136 GeV to the limited statistics and relatively high energy threshold 500 of the CT1–4 analysis, the CT1–4 results for PG 1553+113 400 are represented on Fig.7 by 3 data points only. Taking into 300 consideration the systematic uncertainties on the energy scale 200 and flux normalization (see Sect.5), the CT1–4 data were found 100 to be in satisfactory agreement with the CT5 results. 0 0 0.05 0.1 0.15 0.2 0.25 0.3 θ2 (deg2) 4.4. HE gamma-rays observed by Fermi-LAT Fig. 6. Top:(Left) PG 1553+113 excess map and (right) significance distribution for events with reconstructed energy between 100 GeV 4.4.1. Contemporaneous data and 136 GeV (H.E.S.S. II mono analysis). Bottom: distribution of θ2 (squared angular distance to PKS 2155 304) for gamma-like events. The Fermi-LAT detects gamma-ray photons above an en- − ergy of 100 MeV. Data taken contemporaneously with the H.E.S.S. II observations were analysed with the publicly avail- 3 4.3. Cross check analysis able ScienceTools v10r0p5 . Photon events in a circular region of 15◦ radius centred on the position of sources of in- The robustness of the new H.E.S.S. II mono results presented terest were considered and the PASS 8 instrument response above has been tested through an independent analysis using functions (event class 128 and event type 3) correspond- the Image Pixel-wise fit for Atmospheric Cherenkov Telescopes ing to the P8R2_SOURCE_V6 response were used together (ImPACT) method described in Parsons & Hinton(2014). This with a zenith angle cut of 90◦. The analysis was per- independent analysis provides a consistent cross-check with the formed using the Enrico Python package (Sanchez & Deil above results, being successfully applied to the reconstruction of 2013) adapted for PASS 8 analysis. The sky model was con- data coming from CT5-only triggers (Parsons et al. 2015). The structed based on the 3FGL catalogue (Acero et al. 2015). The analysis was equally capable of detecting PKS 2155 304 below − Galactic diffuse emission has been modelled using the file 100 GeV and the derived spectra were found to be in very good gll_iem_v06.fits (Acero et al. 2016) and the isotropic back- agreement with the Model analysis for both PKS 2155 304 and − ground using iso_P8R2_SOURCE_V6_v06.txt. PG 1553+113. Furthermore, the difference between the spectral Three energy ranges were considered with the corresponding parameters derived using ImPACT and the Model analysis was data cuts in this analysis: 0.1 GeV–500 GeV, 10 GeV–500 GeV adopted as an estimate of the systematic uncertainties associated and 50 GeV–500 GeV, with time windows chosen to coincide with the reconstruction and analysis techniques (see Sect.5). with the H.E.S.S. II observation periods (as defined in Sect. 3.1). Additionally, the robustness of the analysis was tested using The spectral fit parameter results are given in Table2. For an alternative cut configuration. Within the statistical and sys- both AGNs a log-parabola fit to the contemporaneous Fermi- tematic uncertainties, the results obtained with the different cut LAT data did not provide a sufficient improvement to the spectral configurations were found to be in good agreement with each fit, with respect to the power-law model. Some evidence for a other. softening of the spectrum with energy in the Fermi-LAT energy The CT1–4 stereoscopic data collected simultaneously with the H.E.S.S. II mono data have been analysed us- 3 See http://fermi.gsfc.nasa.gov/ssc/data/analysis/ ing the H.E.S.S. I version of the Model analysis method documentation/

A89, page 6 of 13

107 A. Selected publications

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

Table 2. Fermi-LAT spectral analysis results for the time intervals contemporaneous with the H.E.S.S. II observations.

Source Year MJD Eth φ0 Γ E0 TS 11 2 1 1 (GeV) 10− (ph cm− s− GeV− ) (GeV) +0.03 PKS 2155 304 2013 56 403–56 601 0.1 557 26 1.82 0.03 1.48 2162.6 − ± − +0.21 10 2.52 0.43 2.00 0.21 25.5 379.7 ± − +0.66 50 0.12 0.05 1.82 0.72 112 52.4 ± − +0.13 PKS 2155 304 2014 56 805–56 817 0.1 996 168 1.79 0.13 1.54 193.5 − ± − +0.45 10 2.36 1.18 1.20 0.45 53.3 52.4 ± − +1.03 50 1.00 0.71 1.53 1.20 115 23.7 ± − +0.07 PG 1553+113 2013 56 403–56 817 0.1 118 13 1.59 0.07 2.95 455.6 ± − +0.26 10 2.04 0.53 1.68 0.21 33.5 169.9 ± − +0.91 50 0.64 0.27 2.97 1.13 80.8 66.8 ± −

Notes. For each data set and energy threshold, Eth, the differential flux φ0 at decorrelation energy E0, photon index Γ, and value of the test statistic (TS), for the power-law fit, are provided.

range, however, was suggested by the analysis of Fermi-LAT ν [Hz] 23 24 25 26 27 data for the scan of energy thresholds shown in Figs.3 and7 10 10 10 10 10 ] whose ﬁt indices are given in Table2. The data points have been -1 obtained by redoing the Fermi-LAT analysis in a restrained en- −10 47 10 10 [ erg s ergy range freezing the spectral index of the power-law model PG 1553+113 ν ] L -1 ν

to the value found for the global fit above 100 MeV. An upper- s limit at 95% confidence level is computed if the TS is found to -2 −11 46 be below 9. 10 10 These Fermi-LAT analysis results are used to pro- −10 CT5 mono 10 vide gamma-ray HE-VHE SEDs of PKS 2155 304 and −12 45 10 −11 10 − dN/dE [ erg cm Fermi-LAT E>100MeV 10 PG 1553+113. In Fig.3, the 2013 H.E.S.S. II data set of 2 E Fermi-LAT E>10GeV 10−12 PKS 2155 304 is presented along with the contemporaneous Fermi-LAT E>50GeV − −13 Fermi H.E.S.S. I (CT1-4) 10 -LAT data analysed above 100 MeV (shaded red), 10 GeV −13 −1 2 3 44 (shaded green) and 50 GeV (shaded magenta) respectively. 10 10 1 10 10 10 10 − These results show very good agreement between the Fermi- 10 1 1 10 102 103 LAT and H.E.S.S. II mono data within the common overlap- E [GeV] 4 ping region , presenting a comprehensively sampled SED over Fig. 7. Energy spectrum of PG 1553+113 obtained from the H.E.S.S. II more than four orders of magnitude in energy. Evidence for a mono analysis (blue) in comparison with the contemporaneous Fermi- strong down-turn spectral feature within this broadband SED, LAT data with an energy threshold of 0.1 GeV (red triangles and confi- occurring near the transition zone between the two instruments, dence band), 10 GeV (green band), and 50 GeV (purple band) and con- is apparent. temporaneous CT1–4 data (grey squares). In all cases the bands shown Figure7 presents the SED of PG 1553 +113 obtained from represent the 1σ confidence region. The right-hand y-axis shows the the contemporaneous Fermi-LAT and H.E.S.S. II data. In equivalent isotropic luminosity (not corrected for beaming or EBL ab- = . this case, again, good agreement between the Fermi-LAT and sorption) assuming redshift z 0 49. The inset compares the H.E.S.S. confidence band with the Fermi-LAT catalogue data (3FGL, 1FHL and H.E.S.S. II mono data is found within the common energy range 2FHL, see Sect. 4.4.2). of the two instruments. Furthermore, evidence of a strong downturn feature within this SED, occurring within the overlapping energy range of the two instruments, is once again apparent. the Fermi-LAT analysis above 100 MeV, while the 1FHL relies on the first 3 years of data with a higher energy cut at 10 GeV. 4.4.2. Catalogue data Moreover, the 2FHL catalogue was built with the highest energy available to Fermi-LAT only, with E > 50 GeV, probing a The H.E.S.S. II mono and contemporaneous Fermi-LAT spec- somewhat different energy range, and thus potentially different tra of PKS 2155 304 and PG 1553+113 obtained in the pre- spectral properties with respect to the FGL source catalogues. vious sections are− compared here to the Fermi-LAT catalogue results. Different catalogues probing different photon statis- The insets in Figs.3 and7 provide a comparison of the tics and energy ranges are considered here, namely the 3FGL H.E.S.S. II mono results (shown by the blue band) with the (Acero et al. 2015), the 1FHL (Ackermann et al. 2013) and the Fermi-LAT catalogue data (red for 3FGL, green for 1FHL, and purple for 2FHL), for PKS 2155 304 and PG 1553+113, respec- 2FHL (Ackermann et al. 2016). The 3FGL catalogue gives an − average state of the sources with 4 years of data integrated in tively. It is worth comparing the Fermi-LAT contemporaneous data 4 80–500 GeV for PKS 2155 304 and 110–500 GeV for obtained in Sect. 4.4.1 and the Fermi-LAT catalogue data dis- PG 1553+113. − cussed here. For PKS 2155 304, it is noted that the Fermi-LAT − A89, page 7 of 13

108 A.3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 − A&A 600, A89 (2017) catalogue flux is slightly above the Fermi-LAT contemporaneous The variability in HE has also been probed on a weekly flux in the high energy band. For PG 1553+113, however, the timescale which gives a good balance between the ability to catalogue flux is in close agreement with the Fermi-LAT con- probe short timescale variations and good statistics. For the temporaneous flux in the high energy band. Since the Fermi- 2013 dataset (the 2014 dataset time range being too short), LAT catalogue data represent the average flux state of the source PKS 2155 304 is found to be variable with Fvar = 37%. since data taking commenced in 2008, the comparable level of For PG− 1553+113, our new H.E.S.S. II mono spectral re- the fluxes (though slightly below for the case of PKS 2155 304) sults are in reasonable agreement with the earlier measurements is suggestive that both sources were in average states of activity− by H.E.S.S. (Aharonian et al. 2008; Abramowski et al. 2015; at during the observational campaign. It has to be noted that the E > 200 GeV), MAGIC (Albert et al. 2007; Aleksic´ et al. 2010, catalogues are based on different time intervals and different en- 2012b) and VERITAS (Aliu et al. 2015), as well as with the ergy ranges. Furthermore, the results of the fits are dominated Fermi-LAT catalogue spectra (at E < 200 GeV). These compar- by the lower energy events and, in particular for the 2FHL, the isons with previous measurements indicate that PG 1553+113 statistics are rather poor at the highest energies. was indeed in a low state during the H.E.S.S. II observation period of the results presented. No significant night-by-night or weekly variability is found in the H.E.S.S. II mono lightcurve. 4.5. Variability The upper limit on Fvar is found to be 21% at the 95% confidence level. In the HE range, PG 1553+113 is not variable and The AGNs considered in this work are known to be vari- F < 110% at 95% CL. able at VHE, both having previously been observed to ex- var hibit major flares (Aharonian et al. 2007; Abramowski et al. 2015). In the case of PKS 2155 304, this variability has 5. Systematic uncertainties been shown to also introduce changes− in the spectral shape (H.E.S.S. Collaboration et al. 2010). The main sources of systematic uncertainties in the H.E.S.S. II mono analysis presented in this publication, and their estimated In both cases, the present observational campaign found contributions to the uncertainty on the spectral parameters, are the AGNs to be in low states. For PKS 2155 304, at E > − summarised in Table3. For each source of uncertainty the table 300 GeV the spectrum level from our new H.E.S.S. II mono gives the flux normalisation uncertainty, the photon index uncer- result agrees with the level reported for the quiescent state tainty and the uncertainty on the curvature parameter β (for the observed by H.E.S.S. from observations during 2005–2007 log-parabola model). In addition, the energy scale uncertainty is < (H.E.S.S. Collaboration et al. 2010). As seen in Fig.3, at E given in the second column. The energy scale uncertainty im- 300 GeV the H.E.S.S. II mono spectrum level lies below the plies an additional uncertainty on the flux normalisation which Fermi-LAT spectra reported in the 3FGL and 1FHL catalogues. depends on the steepness of the spectrum. It is also relevant for These comparisons are all consistent with PKS 2155 304 being − the determination of the position of spectral features such as the in a low flux state during the observations analysed in this work, SED maximum or EBL cutoff. The procedures used here for es- as is also indicated by the Fermi-LAT contemporaneous analysis timating the systematic uncertainties generally repeat the pro- results. cedures used for H.E.S.S. I (Aharonian et al. 2006a). We high- Although observed in a low state, the H.E.S.S. II light that the discussion in this section focuses specifically on the mono lightcurve of PKS 2155 304 did exhibit nightly and − sources and analysis presented. A more general discussion of the monthly variability with a fractional variability amplitude Fvar systematic uncertainties of the H.E.S.S. II mono analysis will be (Vaughan et al. 2003) of, respectively 47% and 59%. Inter- ≈ ≈ part of a future publication. year variability at VHE with a fractional variability amplitude Except for background subtraction, all sources of uncertainty F of 50% has also been found. Analysis of this variability var ≈ listed in Table3 are related to the conversion of the measured in the H.E.S.S. II mono data set revealed that an increase in the event counts into flux. This conversion is done using the in- flux exists between the 2013 and the 2014 dataset by a factor strument response functions (IRF) which are determined from 1.6 0.1 , though without significant change in the spectral ± stat Monte Carlo simulations. The IRF uncertainties show how well parameters. A simple power-law fit to the 2013 (resp. 2014) data the real instrument, after all calibrations, is described by the sim- yields a spectral index Γ2013 = 2.92 0.04stat (resp. Γ2014 = ulation. 2.91 0.08 ). We note, however, that± the statistics of the 2013 ± stat The first group of uncertainties is related to the interac- and 2014 PKS 2155 304 H.E.S.S. II mono data sets are signif- tion of particles and their production and to the absorption of icantly different in size.− Consequently, the 2014 PKS 2155 304 − Cherenkov light in the atmosphere. The estimated uncertainty data set is not sufficient to discriminate between a power-law or a due to the shower interaction model does not exceed 1% (for log-parabola shaped spectrum, whereas the 2013 PKS 2155 304 − photon-induced showers). The atmospheric uncertainties include data set is found to be significantly better fit with a log-parabolic the effects of the atmospheric density profile (which affects the spectrum. height of shower maximum and Cherenkov light production) For comparison, variability analysis of the PKS 2155 304 and the atmospheric transparency (light attenuation by Mie and contemporaneous Fermi-LAT data, discussed in Sect. 4.4.1−, was Rayleigh scattering). These effects were studied extensively dur- carried out. Figure4 shows the PKS 2155 304 2013 and 2014 ing H.E.S.S. phase I (Bernlöhr 2000; Aharonian et al. 2006a; multi-wavelength SED obtained. It is notable− that a brighten- Hahn et al. 2014). The uncertainties were found to be dominated ing of the source flux between these two epochs by about the by the atmospheric transparency, which has direct influence on same level as that seen by H.E.S.S. II mono is also observed the amount of Cherenkov light detected by the telescopes, thus in the Fermi-LAT contemporaneous results, and again without affecting the energy reconstruction. Data from the telescope ra- any corresponding spectral variability. That is the Fermi-LAT diometers and other atmospheric monitoring devices, as well as and H.E.S.S. II mono photon indices are respectively consistent trigger rate data, are used to ensure good atmospheric conditions between the two epochs, but the overall flux increased by about during the observations used in the analysis (see Sect. 3.2). For 60%. zenith angles relevant to this work, the remaining uncertainty on

A89, page 8 of 13

109 A. Selected publications

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

Table 3. Estimated contributions to the systematic uncertainties in the spectral measurements using H.E.S.S. II mono for the analyses presented in this work. Source of uncertainty Energy scale Flux Index Curvature MC shower interactions – 1% – – MC atmosphere simulation 7% – – Instrument simulation/calibration 10% 10% – – Broken pixels – 5% – – Live time – <5% – – Reconstruction and selection cuts 15% 15% 0.1/0.46 0.01/0.8 Background subtraction – 6%/10% 0.14/0.46 0.12/0.6 Total 19% 20%/22% 0.17/0.65 0.12/1.0 Notes. Numbers separated by “/” correspond to PKS 2155 304 and PG 1553+113, respectively. − the absolute energy scale due to the atmosphere is estimated to uncertainties, while the description of the atmosphere and be 7% (Aharonian et al. 2006a, similar to the uncertainty level instrument calibration contribute substantially to the energy reported≈ in). scale and flux normalisation uncertainties. The instrument simulation and calibration uncertainty in- It should lastly be highlighted that the systematic uncertain- cludes all remaining instrumental effects, such as mirror reflec- ties are energy-dependent. In particular, the background sub- tivity and electronics response. These effects are controlled us- traction uncertainties tend to become more important towards ing various calibration devices (Aharonian et al. 2004), as well low energies, where the signal-to-background ratio is usually as Cherenkov light from atmospheric muons (Leroy et al. 2003). smaller. For an analysis aiming at the lowest energies this can The non-operational pixels in the CT5 camera (<5%) and the lead to a large uncertainty in the measurement of spectral index electronics dead time (<5%) contribute only marginally to the and curvature, especially for soft spectrum sources, as is the case overall uncertainty. for PG 1553+113. The event reconstruction and selection uncertainties are de- In the context of variability studies, the uncertainty val- rived from a comparison of the measured spectra with the results ues presented in Table3 can be considered as a conserva- obtained using an alternative analysis chain (see Sect. 4.3). tive upper bound. Preliminary studies of steady sources with Irregularities in the camera acceptance (e.g. due to non- H.E.S.S. II suggest that the rms variability induced by systematic operational pixels) and the night sky background (e.g. bright effects is about 15–20%, a result similar to that for H.E.S.S. I stars) can both have an effect on background subtraction. The (Aharonian et al. 2006a). This suggests that at least some of background subtraction errors are controlled in this study by vi- the spectral measurement uncertainties are constant in time and sually examining the raw and acceptance-corrected skymaps (to could therefore be reduced by means of additional calibrations. ensure that there are no artefacts, e.g. from bad calibration of Variations related to changes in the atmosphere transparency can individual data runs), as well as using additional dedicated tests also be reduced by means of additional corrections (Hahn et al. and run quality selection. As shown already in Sect.4, the width 2014). of the skymap significance distributions is dominated by statistical errors. This is ensured for both objects, PKS 2155 304 and 6. Discussion PG 1553+113, and throughout the entire energy range− covered by this study (see Figs.2 and6). Hence, arguably, the e ffect The successful H.E.S.S. II mono observations and analysis of the background subtraction errors should not exceed the sta- of PKS 2155 304 and PG 1553+113 convincingly demonstrate tistical uncertainties. Consequently, the statistical uncertainties that the low− energy part of the VHE spectrum is accessible to on the spectral parameters represent a reasonably conservative the H.E.S.S. experiment, following the addition of the CT5 in- estimate of the background subtraction uncertainties. It should strument. This fact makes EBL studies of high redshift AGNs by be noted, however, that the reflected-region background method, H.E.S.S. II mono feasible, without the need for strong theoreti- which is used for the spectral measurements, is potentially more cal biases on the intrinsic spectra or the need to rely on spectral sensitive to non-axially symmetric effects in the camera accep- extrapolations using results from other instruments. tance than the ring background maps (which use a 2D accep- Here we consider EBL deabsorbed fits to the H.E.S.S. II tance model). We have investigated this further by splitting the mono and contemporaneous Fermi-LAT spectra for both AGNs. full data set into two subsets, one of which groups the data from Our aim here is twofold. The first is to investigate evidence for runs taken with a wobble offset in right ascension (in either posi- curvature in the two AGN intrinsic spectra, correcting for EBL tive or negative direction) and another one for the remaining runs absorption effects. Second, given the present systematic uncer- (with wobble in declination). The signal-to-background (S/B) ra- tainties derived for these data sets, we determine the correspond- tios obtained with these subsets were compared to the full dataset ing uncertainties on the combined fit parameters. Such consid- S/B ratio. It was found that the S/B ratio varied by 3%, which is erations provide insight into the constraining power of these re- about twice the background subtraction accuracy≈ observed with sults, under the assumption of both a specific EBL model (in this the ring background method ( 1.5% of the background level). work the one of Franceschini et al. 2008) and simple underlying Therefore in Table3 the statistical≈ uncertainties are doubled to spectral shape. obtain the values for the background subtraction uncertainties. The spectra in the H.E.S.S. II mono energy range The net effect of all uncertainties summed in quadrature have been reconstructed with a spectral model corrected for is given in the last row of Table3. It can be noted that the EBL absorption. Furthermore, for PG 1553+113, whose redshift spectral index and curvature uncertainties are dominated by is not well-constrained, we adopt the well-motivated value of the reconstruction, event selection and background subtraction z = 0.49 (Abramowski et al. 2015).

A89, page 9 of 13

110 A.3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 − A&A 600, A89 (2017)

Table 4. Parameters obtained for the combined ﬁt of the Fermi-LAT and H.E.S.S. data.

11 2 1 Source φ0[10− cm− s− ] Γ β log10(Epeak[GeV]) Sig. (σ) PKS 2155 304 2.35 0.10 0.57 2.30 0.04 0.09 0.15 0.02 0.02 0.99 0.19 0.19 5.1 − ± stat ± sys ± stat ± sys ± stat ± sys ± stat ± sys PG 1553+113 5.97 0.25stat 2.19sys 1.68 0.05stat 0.13sys ––– PG 1553+113 6.66 ± 0.42 ± 1.43 1.83 ± 0.08 ± 0.29 0.12 0.05 0.13 2.76 0.45 0.93 2.2 ± stat ± sys ± stat ± sys ± stat ± sys ± stat ± sys Notes. The reference energy E0 used here is 100 GeV. For both blazars, the log-parabola ﬁts values are provided. For PG 1553+113, the values for the power-law model, which was marginally disfavoured, are also given. The last column gives the signiﬁcance, obtained by comparing the χ2 values for the log-parabola model against those for the power-law model, using only statistical errors in the analysis.

ν [Hz] ν [Hz] 1023 1024 1025 1026 1027 1023 1024 1025 1026 1027 ] ]

-1 48 -1 10−9 10−9 10 PKS 2155-304 PG 1553+113 46

10 [ erg s [ erg s ν ν

] −10 L ] −10 47 L -1 -1 10 ν 10 10 ν s s -2 -2 1045 46 10−11 10−11 10

1044 45 dN/dE [ erg cm −12 dN/dE [ erg cm −12 2 10 2 10 10 E E CT5 mono, corrected for EBL CT5 mono, corrected for EBL 1043 Fermi-LAT E>100MeV Fermi-LAT E>100MeV 10−13 10−13 1044

− − 10 1 1 10 102 103 10 1 1 10 102 103 E [GeV] E [GeV]

Fig. 8. Energy spectrum of PKS 2155 304 obtained from the Fig. 9. Energy spectrum of PG 1553+113 obtained from the H.E.S.S. II H.E.S.S. II mono analysis (blue) of the 2013− data corrected for EBL mono analysis (blue) corrected for EBL absorption in comparison absorption in comparison with the contemporaneous Fermi-LAT data with the contemporaneous Fermi-LAT data with a minimal energy of with a minimal energy of 0.1 GeV (red). The black line is the best-fit 0.1 GeV (red). The assumed redshift is z = 0.49. The black line is the log-parabola model to the points and the cyan butterfly indicates the 1σ best-fit log-parabola model fit to the points and the cyan butterfly indi- region using only the statistical errors in the combined data set analy- cates the 1σ (statistical error only) uncertainty region. The right-hand y- sis. The right-hand y-axis shows the equivalent isotropic luminosity (not axis shows the equivalent isotropic luminosity (not beaming corrected). beaming corrected). The power-law fit of the H.E.S.S. II mono 2013 data obtained an In order to look for a possible turnover in the intrinsic spec- intrinsic spectral index of Γ = 2.49 0.05. Such an index ap- trum and, if present, to locate the peak emission in the energy pears somewhat softer than the power-law± analysis of the Fermi- flux (E2dN/dE) representation, the EBL deabsorbed Fermi-LAT LAT contemporaneous data (Γ = 1.82 0.03 see Table2). The and H.E.S.S. II mono data points were fitted both separately and spectral fits found for the combined data± sets, dominated by the as a combined data set with power-law, broken power-law and low energy data points where EBL effects can be neglected, al- log-parabola models. In the combined fit procedure, a consider- lowed the continuity of the source spectrum to be probed. The ation of the systematic uncertainties for each of the data sets was fit of the combined Fermi-LAT and H.E.S.S. II mono data with a taken into account in the analysis. log-parabola model was preferred at the 5.1σ level with respect For the H.E.S.S. systematic uncertainties, the effect of the to the power-law model (See Fig.8). The broken power-law does energy systematic uncertainty on the deabsorbed spectrum fit not significantly improve the fit in this case. The results of the fit results was found to be the dominant contributing systematic. are given in Table4. The peak flux position within the SED was The contribution of this uncertainty on the results was estimated at a moderate energy (around 10 GeV), in agreement with its through the shifting of the data points in the E dN/dE repre- 4-yr averaged position found in the 3FGL. sentation by an energy scale factor of 19% (see Table3) before applying the EBL deabsorbtion. The variation in the best-fit For PG 1553+113, an EBL absorbed power-law fit to the model, introduced via the application of this procedure within H.E.S.S. II mono spectra required an intrinsic spectral index of Γ = 1.91 0.13. For comparison, Table2 shows that the full energy uncertainty range, was then taken as the system- ± atic contribution to the uncertainty on each model parameter (see the Fermi-LAT spectral fits for power laws with thresholds of Table4). An estimate of the size of the Fermi-LAT systematic 100 MeV and 10 GeV give consistent spectral indices to this uncertainties was also obtained, using the effective area system- value. On the other hand, the fit of the combined Fermi-LAT atic uncertainty, derived by the LAT collaboration5. These un- and H.E.S.S. II mono gamma-ray data, however, found a log- certainties were noted to be small in comparison to the statis- parabola model preferred at the 2.2σ level over the power-law tical errors such that their further consideration could be safely model (See Fig.9). The fit values for these two spectral mod- neglected. els are provided in Table4. The parameters that results from In the case of PKS 2155 304, separate fits of the Fermi- fits with a broken power-law being close to one of the sin- LAT and H.E.S.S. II mono EBL− deabsorbed data, the power-law gle power-law model case. The sizeable systematic errors, once model was found to provide a sufficient description in both cases. also taken into account, however, weaken this preference. Thus, this only marginal improvement, brought by the log-parabola 5 See http://fermi.gsfc.nasa.gov/ssc/data/analysis/ model, suggests that the observed softening of the PG 1553+113 scitools/Aeff_Systematics.html spectrum is predominantly introduced by VHE interaction on

A89, page 10 of 13

111 A. Selected publications

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis the EBL, a result consistent with that from other instruments to be within 1σ of its mean quiescent level, as defined in which have searched for intrinsic curvature in the source’s spec- H.E.S.S. Collaboration et al.(2010), during the 2013 H.E.S.S. II tra (Aleksic´ et al. 2015b). Furthermore, the constraint on the in- observations. Temporal analysis of its emission during the cam- +1.0 trinsic peak position, at a value of 0.6 0.4 TeV, also carries sig- paign revealed mild ( 50%) variability in the lightcurve of nificant uncertainties. This limitation is− primarily due to the rel- PKS 2155 304 between∼ the 2013 and 2014 H.E.S.S. II data atively small intrinsic curvature, limited lever arm (energy range sets. No significant− variability was found in the emission of coverage by the measurements), and the very soft observed spec- PG 1553+113. Further analysis of the PKS 2155 304 data, sep- tral index in the H.E.S.S. II mono band, which amplifies the ef- arating the two consecutive years of observations,− revealed an fect of the energy scale uncertainty. This could be improved in enhancement in the flux state, by a factor of 60%, in the 2014 the future via more accurate calibration of the H.E.S.S. II mono data. Interestingly, a similar size increase in∼ the flux level be- energy scale, using bright flaring or stable sources to compare tween the 2013 and 2014 fluxes is seen in the contemporaneous flux measurements with those of Fermi-LAT contemporaneous Fermi-LAT data (see Fig.4). Spectral analysis of the fluxes from measurements as for exemple in Meyer et al.(2010). these two different brightness periods, however, reveals no evi- In summary, the high-energy SED of PG 1553+113, cor- dence for significant alteration of the spectral shape from either rected for EBL with the model of Franceschini et al.(2008), the H.E.S.S. II mono or Fermi-LAT observations. The change in assuming a redshift of 0.49, reveals only marginal evidence for source state between these periods therefore appears to be as- intrinsic curvature once systematic uncertainties are taken into sociated with a broad increase in the source brightness in the account. This result is compatible with a scenario in which the 0.1–1000 GeV energy range. observed spectral downturn at an energy of around 100 GeV Multi-wavelength SED plots containing the new H.E.S.S. II is introduced through the attenuation at the highest energies data points for these observations of PKS 2155 304 and is due to the interaction of VHE photons with the EBL. Con- PG 1553+113, and their comparison with contemporaneous− trary to this, in the case of PKS 2155 304, the EBL corrected − Fermi-LAT observations, are shown in Figs.3,4, and7. Spec- SED is better described by a log-parabola model than by a tral analysis of the H.E.S.S. II mono data indicate that a log- power-law. The addition of intrinsic spectral curvature or break parabola fit is preferred over a simple power-law or a broken is required to account for the data presented. Such a feature is power-law fit in both cases. The measurement of the curvature naturally expected rather generically on physical grounds in the parameter in these fits, however, is marginal for PG 1553+113 high energy region of the particle spectrum for both stochas- once the systematic errors are taken into account. Within their tic and shock acceleration mechanisms (Park & Petrosian 1995; multi-wavelength SEDs, the presence of a strong spectral down- Heavens & Meisenheimer 1987; Kirk et al. 1998). turn feature, at an energy of 100 GeV, is apparent in both ∼ 7. Conclusions cases, consistent with previous multi-wavelength observations made of these objects during low activity states (Aharonian et al. Here we report, for the first time, H.E.S.S. II mono blazar results 2009; H.E.S.S. Collaboration et al. 2014b; Abdo et al. 2010; following observations of PKS 2155 304 in 2013 and 2014 and Aleksic´ et al. 2012b; Aliu et al. 2015). The introduction of such − PG 1553+113 in 2013, taken with the new CT5 instrument in a feature at these energies is expected through gamma-ray ab- monoscopic configuration. The successful analysis of these ob- sorption on the EBL during their transit through extragalac- servations resulted in the detection of these two AGNs at lev- tic space. Adopting a specific EBL model, spectral fitting els of 42σ (36σ) and 27σ (21σ), respectively. For these re- of the data, deabsorbed on the EBL, indicates the presence ∼ ∼ sults, low-energy thresholds of 80 GeV and 110 GeV, respec- of significant curvature in the intrinsic source spectrum for tively, were achieved. These thresholds amount to a reduction PKS 2155 304, with the peak of the intrinsic SED sitting at by a factor of two to three relative to that achieved in the CT1- an energy− of 10 GeV. A similar EBL deabsorbed analysis for 4 cross-check results presented (see Figs.3 and7). Further- PG 1553+113∼ reveals a milder level of curvature in the intrinsic more, we note that the energy threshold achieved by the present spectrum, suggesting that the peak of the intrinsic SED sits at an H.E.S.S. II mono analysis remains limited by the accuracy of the energy of 500 GeV. However, once systematic errors are taken background subtraction method, rather than by the instrument into account,∼ the intrinsic spectrum of PG 1553+113 was found trigger threshold. to be consistent with no curvature. It therefore remains possible Namely, at energies below the respective thresholds achieved that the observed softening in the PG 1553+113 spectra is purely for the PKS 2155 304 and PG 1553+113 datasets, the system- introduced by VHE interaction on the EBL, and is not intrinsic − atic uncertainties in background subtraction become larger than to the source. the statistical uncertainties. The energy at which the transition Our results demonstrate for the first time the successful em- from statistics-dominated to systematics-dominated regime oc- ployment of the monoscopic data from the new H.E.S.S. II in- curs depends on the accuracy of background subtraction and strument (CT5) for blazar and other AGN studies. These re- the size of the dataset being analysed. For the present analy- sults mark a significant step forward in lowering the gamma- sis the level of systematic uncertainties in background subtrac- ray energy range that AGN may be probed in the H.E.S.S. II tion was found to be 1.5% (for skymaps), which corresponds ≈ era. This reduction in the energy threshold opens up the op- to a minimal requirement for the signal-to-background ratio of portunity to probe new low-energy aspects about AGN fluxes, S/B > 7.5% for a 5σ detection (assuming normally distributed their variability, and their attenuation on the EBL out to larger errors). This limitation does not apply to the special case of redshifts than that probed previously in the H.E.S.S. I era. Fur- gamma-ray pulsars, where the pulsar phasogram can be used thermore, coupled with the level of significance obtained for the to define “off regions” for background subtraction. Subsequent detection of both AGNs, the reduction in threshold offers great improvements and reduction in the energy threshold are likely to potential for temporally resolving AGN lightcurves down to un- occur in the future. precedented temporal scales during flaring episodes. A comparison of the emission level of PKS 2155 304 and − PG 1553+113 with their historic observations revealed both Acknowledgements. The support of the Namibian authorities and of the Uni- to be in low states of activity, with PKS 2155 304 found versity of Namibia in facilitating the construction and operation of H.E.S.S. is − A89, page 11 of 13

112 A.3. Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155 304 and PG 1553+113 − A&A 600, A89 (2017) gratefully acknowledged, as is the support by the German Ministry for Edu- Falomo, R., Pesce, J. E., & Treves, A. 1993, ApJ, 411, L63 cation and Research (BMBF), the Max Planck Society, the German Research Farina, E. P., Fumagalli, M., Decarli, R., & Fanidakis, N. 2016, MNRAS, 455, Foundation (DFG), the French Ministry for Research, the CNRS-IN2P3 and the 618 Astroparticle Interdisciplinary Programme of the CNRS, the UK Science and Franceschini, A., Rodighiero, G., & Vaccari, M. 2008, A&A, 487, 837 Technology Facilities Council (STFC), the IPNP of the Charles University, the Ganguly, R., Lynch, R. S., Charlton, J. C., et al. 2013, MNRAS, 435, 1233 Czech Science Foundation, the Polish Ministry of Science and Higher Educa- Griffiths, R. E., Briel, U., Chaisson, L., & Tapia, S. 1979, ApJ, 234, 810 tion, the South African Department of Science and Technology and National Hahn, J., de los Reyes, R., Bernlöhr, K., et al. 2014, Astropart. Phys., 54, 25 Research Foundation, the University of Namibia, the Innsbruck University, the Heavens, A. F., & Meisenheimer, K. 1987, MNRAS, 225, 335 Austrian Science Fund (FWF), and the Austrian Federal Ministry for Science, H.E.S.S. Collaboration, Abramowski, A., Acero, F., et al. 2010, A&A, 520, Research and Economy, and by the University of Adelaide and the Australian A83 Research Council. We appreciate the excellent work of the technical support H.E.S.S. Collaboration, Abramowski, A., Aharonian, F., et al. 2014a, A&A, 564, staff in Berlin, Durham, Hamburg, Heidelberg, Palaiseau, Paris, Saclay, and in A9 Namibia in the construction and operation of the equipment. This work bene- H.E.S.S. Collaboration, Abramowski, A., Aharonian, F., et al. 2014b, A&A, 571, fited from services provided by the H.E.S.S. Virtual Organisation, supported by A39 the national resource providers of the EGI Federation. The Fermi-LAT Collabo- Holler, M., Balzer, A., Chalmé-Calvet, R., et al. 2015, in Proc. 34th Int. Cosmic ration acknowledges generous ongoing support from a number of agencies and Ray Conference, ICRC 2015 [arXiv:1509.02896] institutes that have supported both the development and the operation of the LAT Kirk, J. G., Rieger, F. M., & Mastichiadis, A. 1998, A&A, 333, 452 as well as scientific data analysis. These include the National Aeronautics and Kotilainen, J. K., Falomo, R., & Scarpa, R. 1998, A&A, 336, 479 Space Administration and the Department of Energy in the United States, the Leroy, N., Bolz, O., Guy, J., et al. 2003, Int. Cosmic Ray Conference, 5, 2895 Commissariat à l’Énergie Atomique and the Centre National de la Recherche Li, T.-P., & Ma, Y.-Q. 1983, ApJ, 272, 317 Scientifique/Institut National de Physique Nucléaire et de Physique des Partic- Mazin, D., & Goebel, F. 2007, ApJ, 655, L13 ules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Meyer, M., Horns, D., & Zechlin, H.-S. 2010, A&A, 523, A2 Nucleare in Italy, the Ministry of Education, Culture, Sports, Science and Tech- Park, B. T., & Petrosian, V. 1995, ApJ, 446, 699 nology (MEXT), High Energy Accelerator Research Organization (KEK) and Parsons, R. D., & Hinton, J. A. 2014, Astropart. Phys., 56, 26 Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallen- Parsons, R. D., Gajdus, M., Murach, T., & for the H.E.S.S. Collaboration. 2015, berg Foundation, the Swedish Research Council and the Swedish National Space in Proc. 34th Int. Cosmic Ray Conference (ICRC 2015) Board in Sweden. Additional support for science analysis during the operations Piron, F., Djannati-Atai, A., Punch, M., et al. 2001, A&A, 374, 895 phase is gratefully acknowledged from the Istituto Nazionale di Astrofisica in Sanchez, D. A., & Deil, C. 2013, in Proc. 33rd International Cosmic Ray Italy and the Centre National d’Études Spatiales in France. This research has Conference, ICRC 2013 [arXiv:1307.4534] made use of NASA’s Astrophysics Data System. This research has made use of Schwartz, D. A., Griffiths, R. E., Schwarz, J., Doxsey, R. E., & Johnston, M. D. the SIMBAD database, operated at CDS, Strasbourg, France. This research made 1979, ApJ, 229, L53 use of Enrico, a community-developed Python package to simplify Fermi-LAT Vaughan, S., Edelson, R., Warwick, R. S., & Uttley, P. 2003, MNRAS, 345, analysis (Sanchez & Deil 2013). M. Razzano is funded by contract FIRB-2012- 1271 RBFR12PM1F from the Italian Ministry of Education, University and Research Vestrand, W. T., Stacy, J. G., & Sreekumar, P. 1995, IAU Circ., 454, 93 (MIUR).

1 Centre for Space Research, North-West University, 2520 Potchef- stroom, South Africa References 2 Universität Hamburg, Institut für Experimentalphysik, Luruper Chaussee 149, 22761 Hamburg, Germany Abdo, A. A., Ackermann, M., Ajello, M., et al. 2009, ApJS, 183, 46 3 Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010, ApJ, 708, 1310 Max-Planck-Institut für Kernphysik, PO Box 103980, 69029 Abeysekara, A. U., Archambault, S., Archer, A., et al. 2015, ApJ, 815, L22 Heidelberg, Germany Abramowski, A., Aharonian, F., Ait Benkhali, F., et al. 2015, ApJ, 802, 65 4 Dublin Institute for Advanced Studies, 31 Fitzwilliam Place, Acero, F., Ackermann, M., Ajello, M., et al. 2015, ApJS, 218, 23 Dublin 2, Ireland Acero, F., Ackermann, M., Ajello, M., et al. 2016, ApJS, 223, 26 5 National Academy of Sciences of the Republic of Armenia, Mar- Ackermann, M., Ajello, M., Allafort, A., et al. 2013, ApJS, 209, 34 shall Baghramian Avenue, 24, 0019 Yerevan, Armenia Ackermann, M., Ajello, M., Albert, A., et al. 2015, ApJ, 813, L41 6 Yerevan Physics Institute, 2 Alikhanian Brothers St., 375036 Yere- Ackermann, M., Ajello, M., Atwood, W. B., et al. 2016, ApJS, 222, 5 van, Armenia Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. 2004, Astropart. Phys., 7 22, 109 Institut für Physik, Humboldt-Universität zu Berlin, Newtonstr. 15, Aharonian, F., Akhperjanian, A. G., Aye, K.-M., et al. 2005, A&A, 430, 865 12489 Berlin, Germany 8 Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006a, A&A, 457, University of Namibia, Department of Physics, 13301 Private Bag, 899 Windhoek, Namibia Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006b, A&A, 448, 9 GRAPPA, Anton Pannekoek Institute for Astronomy, Univer- L19 sity of Amsterdam, Science Park 904, 1098 XH Amsterdam, Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2007, ApJ, 664, The Netherlands L71 10 Department of Physics and Electrical Engineering, Linnaeus Uni- Aharonian, F., Akhperjanian, A. G., Barres de Almeida, U., et al. 2008, A&A, versity, 351 95 Växjö, Sweden 477, 481 11 Aharonian, F., Akhperjanian, A. G., Anton, G., et al. 2009, ApJ, 696, L150 Institut für Theoretische Physik, Lehrstuhl IV: Weltraum und Astro- Albert, J., Aliu, E., Anderhub, H., et al. 2007, ApJ, 654, L119 physik, Ruhr-Universität Bochum, 44780 Bochum, Germany 12 Aleksic,´ J., Anderhub, H., Antonelli, L. A., et al. 2010, A&A, 515, A76 GRAPPA, Anton Pannekoek Institute for Astronomy and Institute of Aleksic,´ J., Alvarez, E. A., Antonelli, L. A., et al. 2012a, A&A, 544, A75 High-Energy Physics, University of Amsterdam, Science Park 904, Aleksic,´ J., Alvarez, E. A., Antonelli, L. A., et al. 2012b, ApJ, 748, 46 1098 XH Amsterdam, The Netherlands Aleksic,´ J., Ansoldi, S., Antonelli, L. A., et al. 2015a, J. High Energy Astrophys., 13 Institut für Astro- und Teilchenphysik, Leopold-Franzens- 5, 30 Universität Innsbruck, 6020 Innsbruck, Austria Aleksic,´ J., Ansoldi, S., Antonelli, L. A., et al. 2015b, MNRAS, 450, 4399 14 School of Physical Sciences, University of Adelaide, 5005 Adelaide, Aliu, E., Archer, A., Aune, T., et al. 2015, ApJ, 799, 7 Australia Berge, D., Funk, S., & Hinton, J. 2007, A&A, 466, 1219 15 Bernlöhr, K. 2000, Astropart. Phys., 12, 255 LUTH, Observatoire de Paris, PSL Research University, CNRS, Biteau, J., & Williams, D. A. 2015, ApJ, 812, 60 Université Paris Diderot, 5 place Jules Janssen, 92190 Meudon, Bolmont, J., Corona, P., Gauron, P., et al. 2014, Nucl. Instrum. Meth. Phys. Res. France A, 761, 46 16 Sorbonne Universités, UPMC Université Paris 06, Université Paris Chadwick, P. M., Lyons, K., McComb, T. J. L., et al. 1999, ApJ, 513, 161 Diderot, Sorbonne Paris Cité, CNRS, Laboratoire de Physique Danforth, C. W., Keeney, B. A., Stocke, J. T., Shull, J. M., & Yao, Y. 2010, ApJ, Nucléaire et de Hautes Énergies (LPNHE), 4 place Jussieu, 75252 720, 976 Paris Cedex 5, France de Naurois, M., & Rolland, L. 2009, Astropart. Phys., 32, 231

A89, page 12 of 13

113 A. Selected publications

H.E.S.S. and Fermi-LAT Collaborations: Gamma-ray blazar spectra with H.E.S.S. II mono analysis

17 DSM/Irfu, CEA Saclay, 91191 Gif-Sur-Yvette Cedex, France 48 Dipartimento di Fisica, Università di Trieste, 34127 Trieste, Italy 18 Astronomical Observatory, The University of Warsaw, Al. Ujaz- 49 Istituto Nazionale di Fisica Nucleare, Sezione di Pisa, 56127 Pisa, dowskie 4, 00-478 Warsaw, Poland Italy 19 Aix-Marseille Université, CNRS/IN2P3, CPPM UMR 7346, 13288 50 Istituto Nazionale di Fisica Nucleare, Sezione di Torino, 10125 Marseille, France Torino, Italy 20 Instytut Fizyki Ja¸drowej PAN, ul. Radzikowskiego 152, 31-342 51 Dipartimento di Fisica, Università degli Studi di Torino, 10125 Kraków, Poland Torino, Italy 21 Laboratoire Univers et Particules de Montpellier, Université Mont- 52 Laboratoire Univers et Particules de Montpellier, Université Mont- pellier, CNRS/IN2P3, CC 72, Place Eugène Bataillon, 34095 pellier, CNRS/IN2P3, 34095 Montpellier, France Montpellier Cedex 5, France 53 Consorzio Interuniversitario per la Fisica Spaziale (CIFS), 10133 22 School of Physics, University of the Witwatersrand, 1 Jan Smuts Torino, Italy Avenue, Braamfontein, 2050 Johannesburg, South Africa 54 Dipartimento di Fisica “M. Merlin” dell’Università e del Politecnico 23 Laboratoire d’Annecy-le-Vieux de Physique des Particules, Uni- di Bari, 70126 Bari, Italy versité Savoie Mont-Blanc, CNRS/IN2P3, 74941 Annecy-le-Vieux, 55 Istituto Nazionale di Fisica Nucleare, Sezione di Bari, 70126 Bari, France Italy 24 Landessternwarte, Universität Heidelberg, Königstuhl, 69117 56 INAF-Istituto di Astroﬁsica Spaziale e Fisica Cosmica, 20133 Heidelberg, Germany Milano, Italy 25 Université Bordeaux, CNRS/IN2P3, Centre d’Études Nucléaires de 57 Agenzia Spaziale Italiana (ASI) Science Data Center, 00133 Roma, Bordeaux Gradignan, 33175 Gradignan, France Italy 26 Oskar Klein Centre, Department of Physics, Stockholm University, 58 Istituto Nazionale di Fisica Nucleare, Sezione di Perugia, 06123 Albanova University Center, 10691 Stockholm, Sweden Perugia, Italy 27 Institut für Astronomie und Astrophysik, Universität Tübingen, 59 Dipartimento di Fisica, Università degli Studi di Perugia, 06123 Sand 1, 72076 Tübingen, Germany Perugia, Italy 60 28 Laboratoire Leprince-Ringuet, École Polytechnique, CNRS/IN2P3, Dipartimento di Fisica e Astronomia “G. Galilei”, Università di 91128 Palaiseau, France Padova, 35131 Padova, Italy 61 29 APC, AstroParticule et Cosmologie, Université Paris Diderot, INAF Istituto di Radioastronomia, 40129 Bologna, Italy 62 CNRS/IN2P3, CEA/Irfu, Observatoire de Paris, Sorbonne Paris Dipartimento di Astronomia, Università di Bologna, 40127 Cité, 10 rue Alice Domon et Léonie Duquet, 75205 Paris Cedex 13, Bologna, Italy 63 France Università Telematica Pegaso, Piazza Trieste e Trento, 48, 80132 30 Univ. Grenoble Alpes, IPAG; CNRS, IPAG, 38000 Grenoble, France Napoli, Italy 64 31 Department of Physics and Astronomy, The University of Leicester, Università di Udine, 33100 Udine, Italy 65 University Road, Leicester, LE1 7RH, UK Istituto Nazionale di Fisica Nucleare, Sezione di Padova, 35131 32 Nicolaus Copernicus Astronomical Center, ul. Bartycka 18, 00-716 Padova, Italy 66 Warsaw, Poland Laboratoire de Physique et Chimie de l’Environnement et de 33 Institut für Physik und Astronomie, Universität Potsdam, Karl- l’Espace – Université d’Orléans/CNRS, 45071 Orléans Cedex 02, Liebknecht-Strasse 24/25, 14476 Potsdam, Germany France 67 34 Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen Cen- Station de radioastronomie de Nançay, Observatoire de Paris, tre for Astroparticle Physics, Erwin-Rommel-Str. 1, 91058 Erlan- CNRS/INSU, 18330 Nançay, France 68 gen, Germany NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA 69 35 DESY, 15738 Zeuthen, Germany NASA Postdoctoral Program Fellow, USA 70 36 Obserwatorium Astronomiczne, Uniwersytet Jagiellonski,´ ul. Orla Science Institute, University of Iceland, 107 Reykjavik, Iceland 71 171, 30-244 Kraków, Poland Department of Physics, Graduate School of Science, University of 37 Centre for Astronomy, Faculty of Physics, Astronomy and Informat- Tokyo, 7-3-1 Hongo, Bunkyo-ku, 113-0033 Tokyo, Japan 72 ics, Nicolaus Copernicus University, Grudziadzka 5, 87-100 Torun, Department of Physical Sciences, Hiroshima University, Higashi- Poland Hiroshima, 739-8526 Hiroshima, Japan 73 38 Department of Physics, University of the Free State, PO Box 339, Department of Physics, KTH Royal Institute of Technology, Al- 9300 Bloemfontein, South Africa baNova, 106 91 Stockholm, Sweden 74 39 Heisenberg Fellow (DFG), ITA Universität Heidelberg, 69029 The Oskar Klein Centre for Cosmoparticle Physics, AlbaNova, Heidelberg, Germany 106 91 Stockholm, Sweden 75 40 GRAPPA, Institute of High-Energy Physics, University of Amster- Institute of Space Sciences (IEEC-CSIC), Campus UAB, 08193 dam, Science Park 904, 1098 XH Amsterdam, The Netherlands Barcelona, Spain 76 41 Department of Physics, Rikkyo University, 3-34-1 Nishi-Ikebukuro, Space Science Division, Naval Research Laboratory, Washington, Toshima-ku, 171-8501 Tokyo, Japan DC 20375-5352, USA 77 42 Now at Santa Cruz Institute for Particle Physics and Department Hiroshima Astrophysical Science Center, Hiroshima University, of Physics, University of California at Santa Cruz, Santa Cruz, Higashi-Hiroshima, 739-8526 Hiroshima, Japan 78 CA 95064, USA Istituto Nazionale di Fisica Nucleare, Sezione di Roma “Tor Ver- 43 Deutsches Elektronen Synchrotron DESY,15738 Zeuthen, Germany gata”, 00133 Roma, Italy 79 44 Department of Physics and Astronomy, Clemson University, Kinard Max-Planck-Institut für Physik, 80805 München, Germany 80 Lab of Physics, Clemson, SC 29634-0978, USA Erlangen Centre for Astroparticle Physics, 91058 Erlangen, 45 Università di Pisa and Istituto Nazionale di Fisica Nucleare, Sezione Germany 81 di Pisa, 56127 Pisa, Italy NYCB Real-Time Computing Inc., Lattingtown, NY 11560-1025, 46 W. W. Hansen Experimental Physics Laboratory, Kavli Institute for USA 82 Particle Astrophysics and Cosmology, Department of Physics and Institució Catalana de Recerca i Estudis Avançats (ICREA), 08010 SLAC National Accelerator Laboratory, Stanford University, Stan- Barcelona, Spain 83 ford, CA 94305, USA INAF-IASF Bologna, 40129 Bologna, Italy 84 47 Istituto Nazionale di Fisica Nucleare, Sezione di Trieste, 34127 Department of Physics and Department of Astronomy, University of Trieste, Italy Maryland, College Park, MD 20742, USA

A89, page 13 of 13

114 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

THE 2012 FLARE OF PG 1553+113 SEEN WITH H.E.S.S. AND FERMI-LAT A. Abramowski1, F. Aharonian2,3,4, F. Ait Benkhali2, A. G. Akhperjanian4,5, E. O. Angüner6, M. Backes7, S. Balenderan8, A. Balzer9, A. Barnacka10,11, Y. Becherini12, J. Becker Tjus13, D. Berge14, S. Bernhard15, K. Bernlöhr2,6, E. Birsin6, J. Biteau16,17, M. Böttcher18, C. Boisson19, J. Bolmont20, P. Bordas21, J. Bregeon22, F. Brun23, P. Brun23, M. Bryan9, T. Bulik24, S. Carrigan2, S. Casanova2,25, P. M. Chadwick8, N. Chakraborty2, R. Chalme-Calvet20, R. C. G. Chaves22, M. Chrétien20, S. Colafrancesco26, G. Cologna27, J. Conrad28,43, C. Couturier20,Y.Cui21, M. Dalton29,44, I. D. Davids7,18, B. Degrange16, C. Deil2, P. deWilt30, A. Djannati-Ataï31, W. Domainko2, A. Donath2,L.O’C. Drury3, G. Dubus32, K. Dutson33, J. Dyks34, M. Dyrda25, T. Edwards2, K. Egberts35, P. Eger2, P. Espigat31, C. Farnier28, S. Fegan16, F. Feinstein22, M. V. Fernandes1, D. Fernandez22, A. Fiaßon36, G. Fontaine16, A. Förster2, M. FüSSling35, S. Gabici31, M. Gajdus6, Y. A. Gallant22, T. Garrigoux20, G. Giavitto37, B. Giebels16, J. F. Glicenstein23, D. Gottschall21, M.-H. Grondin2,27, M. Grudzińska24, D. Hadsch15, S. Häffner38, J. Hahn2, J. Harris8, G. Heinzelmann1, G. Henri32, G. Hermann2, O. Hervet19, A. Hillert2, J. A. Hinton33, W. Hofmann2, P. Hofverberg2, M. Holler35, D. Horns1, A. Ivascenko18, A. Jacholkowska20, C. Jahn38, M. Jamrozy10, M. Janiak34, F. Jankowsky27, I. Jung38, M. A. Kastendieck1, K. Katarzyński39,U.Katz38, S. Kaufmann27, B. Khéliﬁ31, M. Kieffer20, S. Klepser37, D. Klochkov21, W. Kluźniak34, D. Kolitzus15, Nu. Komin26, K. Kosack23, S. Krakau13, F. Krayzel36, P. P. Krüger18, H. Laffon29, G. Lamanna36, J. Lefaucheur31, V. Lefranc23, A. Lemiére31, M. Lemoine-Goumard29, J.-P. Lenain20, T. Lohse6, A. Lopatin38, C.-C. Lu2, V. Marandon2, A. Marcowith22, R. Marx2, G. Maurin36, N. Maxted30, M. Mayer35, T. J. L. McComb8, J. Méhault29,44, P. J. Meintjes40, U. Menzler13, M. Meyer28, A. M. W. Mitchell2, R. Moderski34, M. Mohamed27, K. Morå28, E. Moulin23, T. Murach6, M. de Naurois16, J. Niemiec25, S. J. Nolan8, L. Oakes6, H. Odaka2, S. Ohm37, B. Opitz1, M. Ostrowski10,I.Oya6, M. Panter2, R. D. Parsons2, M. Paz Arribas6, N. W. Pekeur18, G. Pelletier32, J. Perez15, P.-O. Petrucci32, B. Peyaud23, S. Pita31, H. Poon2, G. Pühlhofer21, M. Punch31, A. Quirrenbach27, S. Raab38, I. Reichardt31, A. Reimer15, O. Reimer15, M. Renaud22, R. de los Reyes2, F. Rieger2, L. Rob41, C. Romoli3, S. Rosier-Lees36, G. Rowell30, B. Rudak34, C. B. Rulten19, V. Sahakian4,5, D. Salek42, D. A. Sanchez36, A. Santangelo21, R. Schlickeiser13, F. Schüssler23, A. Schulz37, U. Schwanke6, S. Schwarzburg21, S. Schwemmer27, H. Sol19, F. Spanier18, G. Spengler28, F. Spies1, Ł. Stawarz10, R. Steenkamp7, C. Stegmann35,37, F. Stinzing38, K. Stycz37, I. Sushch6,18, J.-P. Tavernet20, T. Tavernier31, A. M. Taylor3, R. Terrier31, M. Tluczykont1, C. Trichard36, K. Valerius38, C. van Eldik38, B. van Soelen40, G. Vasileiadis22,J.Veh38, C. Venter18, A. Viana2, P. Vincent20, J. Vink9, H. J. Völk2, F. Volpe2, M. Vorster18, T. Vuillaume32, P. Wagner6, R. M. Wagner28, M. Ward8, M. Weidinger13, Q. Weitzel2, R. White33, A. Wierzcholska25, P. Willmann38, A. Wörnlein38, D. Wouters23, R. Yang2, V. Zabalza2,33, D. Zaborov16, M. Zacharias27, A. A. Zdziarski34, A. Zech19, and H.-S. Zechlin1 (H.E.S.S. Collaboration) 1 Universität Hamburg, Institut für Experimentalphysik, Luruper Chaussee 149, D-22761 Hamburg, Germany 2 Max-Planck-Institut für Kernphysik, P.O. Box 103980, D-69029 Heidelberg, Germany 3 Dublin Institute for Advanced Studies, 31 Fitzwilliam Place, Dublin 2, Ireland 4 National Academy of Sciences of the Republic of Armenia, Marshall Baghramian Avenue, 24, 0019 Yerevan, Republic of Armenia 5 Yerevan Physics Institute, 2 Alikhanian Brothers St., 375036 Yerevan, Republic of Armenia 6 Institut für Physik, Humboldt-Universität zu Berlin, Newtonstr. 15, D-12489 Berlin, Germany 7 University of Namibia, Department of Physics, Private Bag 13301, Windhoek, Namibia 8 University of Durham, Department of Physics, South Road, Durham DH1 3LE, UK 9 GRAPPA, Anton Pannekoek Institute for Astronomy, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands 10 Obserwatorium Astronomiczne, Uniwersytet Jagielloński, ul. Orla 171, 30-244 Kraków, Poland 11 now at Harvard-Smithsonian Center for Astrophysics, 60 Garden St, MS-20, Cambridge, MA 02138, USA 12 Department of Physics and Electrical Engineering, Linnaeus University, SE-351 95 Växjö, Sweden 13 Institut für Theoretische Physik, Lehrstuhl IV: Weltraum und Astrophysik, Ruhr-Universität Bochum, D-44780 Bochum, Germany 14 GRAPPA, Anton Pannekoek Institute for Astronomy and Institute of High-Energy Physics, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands 15 Institut für Astro- und Teilchenphysik, Leopold-Franzens-Universität Innsbruck, A-6020 Innsbruck, Austria 16 Laboratoire Leprince-Ringuet, Ecole Polytechnique, CNRS/IN2P3, F-91128 Palaiseau, France 17 now at Santa Cruz Institute for Particle Physics, Department of Physics, University of California at Santa Cruz, Santa Cruz, CA 95064, USA 18 Centre for Space Research, North-West University, Potchefstroom 2520, South Africa 19 LUTH, Observatoire de Paris, CNRS, Université Paris Diderot, 5 Place Jules Janssen, F-92190 Meudon, France 20 LPNHE, Université Pierre et Marie Curie Paris 6, Université Denis Diderot Paris 7, CNRS/IN2P3, 4 Place Jussieu, F-75252, Paris Cedex 5, France; camille. [email protected], [email protected] 21 Institut für Astronomie und Astrophysik, Universität Tübingen, Sand 1, D-72076 Tübingen, Germany 22 Laboratoire Univers et Particules de Montpellier, Université Montpellier 2, CNRS/IN2P3, CC 72, Place Eugène Bataillon, F-34095 Montpellier Cedex 5, France 23 DSM/Irfu, CEA Saclay, F-91191 Gif-Sur-Yvette Cedex, France; [email protected] 24 Astronomical Observatory, The University of Warsaw, Al. Ujazdowskie 4, 00-478 Warsaw, Poland 25 Instytut Fizyki Jadrowej̧ PAN, ul. Radzikowskiego 152, 31-342 Kraków, Poland 26 School of Physics, University of the Witwatersrand, 1 Jan Smuts Avenue, Braamfontein, Johannesburg, 2050, South Africa 27 Landessternwarte, Universität Heidelberg, Königstuhl, D-69117 Heidelberg, Germany 28 Oskar Klein Centre, Department of Physics, Stockholm University, Albanova University Center, SE-10691 Stockholm, Sweden 29 Université Bordeaux 1, CNRS/IN2P3, Centre d’Études Nucléaires de Bordeaux Gradignan, F-33175 Gradignan, France 30 School of Chemistry & Physics, University of Adelaide, Adelaide 5005, Australia

115 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

31 APC, AstroParticule et Cosmologie, Université Paris Diderot, CNRS/IN2P3, CEA/Irfu, Observatoire de Paris, Sorbonne Paris Cité, 10, rue Alice Domon et Léonie Duquet, F-75205 Paris Cedex 13, France; [email protected] 32 Univ. Grenoble Alpes, IPAG, F-38000 Grenoble, France CNRS, IPAG, F-38000 Grenoble, France 33 Department of Physics and Astronomy, The University of Leicester, University Road, Leicester, LE1 7RH, UK 34 Nicolaus Copernicus Astronomical Center, ul. Bartycka 18, 00-716 Warsaw, Poland 35 Institut für Physik und Astronomie, Universität Potsdam, Karl-Liebknecht-Strasse 24/25, D-14476 Potsdam, Germany 36 Laboratoire d’Annecy-le-Vieux de Physique des Particules, Université de Savoie, CNRS/IN2P3, F-74941 Annecy-le-Vieux, France; [email protected] 37 DESY, D-15738 Zeuthen, Germany 38 Universität Erlangen-Nürnberg, Physikalisches Institut, Erwin-Rommel-Str. 1, D-91058 Erlangen, Germany 39 Centre for Astronomy, Faculty of Physics, Astronomy and Informatics, Nicolaus Copernicus University, Grudziadzka 5, 87-100 Torun, Poland 40 Department of Physics, University of the Free State, P.O. Box 339, Bloemfontein 9300, South Africa 41 Charles University, Faculty of Mathematics and Physics, Institute of Particle and Nuclear Physics, V Holešovičkách 2, 180 00 Prague 8, Czech Republic 42 GRAPPA, Institute of High-Energy Physics, University of Amsterdam, Science Park 904, 1098 XH Amsterdam, The Netherlands Received 2014 December 17; accepted 2015 January 11; published 2015 March 24

ABSTRACT Very high energy (VHE, E > 100 GeV) γ-ray flaring activity of the high-frequency peaked BL Lac object PG 1553 +113 has been detected by the H.E.S.S. telescopes. The flux of the source increased by a factor of 3 during the nights of 2012 April 26 and 27 with respect to the archival measurements with a hint of intra-night variability. No counterpart of this event has been detected in the Fermi-Large Area Telescope data. This pattern is consistent with VHE γ-ray flaring being caused by the injection of ultrarelativistic particles, emitting γ-rays at the highest energies. The dataset offers a unique opportunity to constrain the redshift of this source at z = 0.49 ± 0.04 using a novel method based on Bayesian statistics. The indication of intra-night variability is used to introduce a novel method to probe for a possible Lorentz invariance violation (LIV), and to set limits on the energy scale at which Quantum Gravity (QG) effects causing LIV may arise. For the subluminal case, the derived limits are 17 10 EQG,1 > 4.10 × 10 GeV and EQG,2 > 2.10 × 10 GeV for linear and quadratic LIV effects, respectively. Key words: BL Lacertae objects: individual (PG 1553+113) – galaxies: active – galaxies: distances and redshifts – gamma-rays: galaxies

1. INTRODUCTION a photon index of G = 4.0 ± 0.6. At high energies (HE, 100 MeV < E < 300 GeV) the source has been detected by the Blazars are active galactic nuclei (AGNs) with their jets Fermi Large Area Telescope (LAT)(Abdo closely aligned with the line of sight to the Earth (Urry & ) fl et al. 2009b, 2010a) with a very hard photon index of G Padovani 1995 . Among their particularities is ux variability = ± – at all wavelengths on various time scales, from years down to 1.68 0.03, making this object the one with the largest HE ( ) ( VHE spectral break (DG ≈ 2.3) ever measured. No variability in some cases minutes Gaidos et al. 1996; Aharonian ( ) et al. 2007a). Flaring activity of blazars is of great interest for in Fermi-LAT was found by Abdo et al. 2009b, 2010a on probing the source-intrinsic physics of relativistic jets, daily or weekly time scales, but using an extended data set of ć ( ) relativistic particle acceleration and generation of high-energy 17 months, Aleksi et al. 2012 reported variability above fl ∼ radiation, as well as for conducting fundamental physics tests. 1 GeV with ux variations of a factor of 5 on a yearly time On the one hand, exploring possible spectral variability scale. between flaring and stationary states helps to understand the With 5 yr of monitoring data of the MAGIC telescopes, ć ( ) γ electromagnetic emission mechanisms at play in the jet. On the Aleksi et al. 2012 discovered variability in VHE -rays with fl ( other hand, measuring the possible correlation between photon only modest ux variations from 4% to 11% of the Crab energies and arrival times allows one to test for possible Nebula flux). In addition to the high X-ray variability, this Lorentz invariance violation (LIV) leading to photon-energy- behavior can be interpreted as evidence for Klein–Nishina dependent variations in the speed of light in vacuum. effects (Abdo et al. 2010a) in the framework of a synchrotron Located in the Serpens Caput constellation, PG 1553+113 self-Compton model. The source underwent VHE γ-ray flares was discovered by Green et al. (1986), who first classified it as in 2012 March (Cortina 2012a) and April (Cortina 2012b), a BL Lac object. Later the classification was refined to a high- detected by the MAGIC telescopes. During the March flare, the frequency peaked BL Lac object (HBL, Giommi et al. 1995). source was at a flux level of about 15% of that of the Crab PG 1553+113 exhibits a high X-ray to radio flux Nebula, while in April it reached ≈50%. During those VHE γ- fl (log (FF2keV 5GHz )>- 4.5, Osterman et al. 2006), which ray ares, also a brightening in X-ray, UV and optical places it among the most extreme HBLs (Rector et al. 2003). wavelengths has been noticed by the MAGIC collaboration. The object was observed in X-rays by multiple instruments in A detailed study of the MAGIC telescopes and multi- different flux states. Its 2–10 keV energy flux ranges from wavelength data is in press (Aleksić et al. 2014). The latter 0.3´ 10---11 erg cm 2 s 1 (Osterman et al. 2006) to event triggered the H.E.S.S. observations reported in this work. 3.5´ 10---11 erg cm 2 s 1 (Reimer et al. 2008) but no fast Note that the VERITAS collaboration has reported an overall variability (in the sub-hour time scale) has been detected so far. higher flux in 2012 (Aliu et al. 2015) in VHE. PG 1553+113 was discovered at very high energies (VHE, Despite several attempts to measure it, the redshift of E > 100 GeV) by H.E.S.S. (Aharonian et al. 2006a, 2008) with PG 1553+113 still suffers from uncertainties. Different attempts, including optical spectroscopy (Treves et al. 2007; 43 Wallenberg Academy Fellow. Aharonian et al. 2008) or comparisons of the HE and VHE 44 Funded by contract ERC-StG-259391 from the European Community. spectra of PG 1553+113 (Prandini et al. 2009; Sanchez

116 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Table 1 Summary of the Statistics for Both Data Sets (First Column)

ﬁ P cst Data Set ON OFF r Excess Signi cance Eth (GeV) Zenith Angle c2 Pre-ﬂare 2205 13033 0.100 901.7 21.5 217 34° 0.77

Flare 559 1593 0.105 391.2 22.0 240 52° 3.3 × 10−3

Note. The second and third columns give the number of ON and OFF events. The fourth column gives the ratio between ON and OFF exposures (r). The excess and the corresponding signiﬁcance are given, as well as the energy threshold and the mean zenith angle of the source during the observations. The last column presents the probability of the ﬂux to be constant within the observations (see text).

Table 2 Summary of the Fitted Spectral Parameters for the Pre-ﬂare and the Flare Data Sets and the Corresponding Integral Flux I Calculated Above 300 GeV

Data Set (Model) Spectral Parameters I (E > 300 GeV) Edec (10−12 ph cm−2 s−1)(GeV)

Pre-ﬂare (PWL) Γ = 4.8 ± 0.2stat ± 0.2sys 4.4 ± 0.4stat ± 0.9sys 306

Pre-ﬂare (LP) a =5.4 0.4stat  0.1 sys 5.0 ± 0.6stat ± 1.0sys L

b = 4.0 ± 1.4stat ± 0.2sys

Flare (PWL) Γ = 4.9 ± 0.3stat ± 0.2sys 15.1 ± 1.3stat ± 3.0sys 327

Note. The Last Column gives the Decorrelation Energy. et al. 2013), were made. Based on the assumption that the The pre-flare data set is composed of 26.4 live time hours of extragalactic background light (EBL)-corrected VHE spectral good-quality data (Aharonian et al. 2006b). For the flare index is equal to the Fermi-LAT one, Prandini et al. (2009) period, eight runs of ∼28 minutes each were taken during the derived an upper limit (UL) of z < 0.67. Comparing PG 1553 nights of 2012 April 26 and 27, corresponding to 3.5 hr of live +113 statistically with other known VHE emitters and taking time. All the data were taken in wobble mode, for which the into account a possible intrinsic γ-ray spectral break through a source is observed with an offset of 0◦.5 with respect to the simple emission model, Sanchez et al. (2013) constrained the center of the instrumentʼs field of view, yielding an acceptance- redshift to be below 0.64 and Aliu et al. (2015) constrained it at corrected live time of 24.7 and 3.2 hr for the pre-flare and flare z < 0.62 using VHE data only. The best estimate to-date was data sets, respectively. obtained by Danforth et al. (2010) who found the redshift to be Data were analyzed using the Model analysis (de Naurois & between 0.43 and 0.58 using far-ultraviolet spectroscopy. Rolland 2009) with Loose cuts. This method–based on the This paper concentrates on the HE and VHE emission of comparison of detected shower images with a pre-calculated PG 1553+113 and is divided as follows: Sections 2.1 and 2.2 model–achieves a better rejection of hadronic air showers and a present the H.E.S.S. and Fermi-LAT analyses. The discus- better sensitivity at lower energies than analysis methods based sion, in Section 3, includes the determination of the redshift on Hillas parameters. The chosen cuts, best suited for sources using a novel method and the constraints derived on LIV with steep spectra such as PG 1553+113,45 require a minimum using a modified likelihood formulation. Throughout this image charge of 40 photoelectrons, which provides an energy paper a LCDM cosmology with H0 =70.4 1.4 threshold of ~217 GeV for the pre-flare and ~240 GeV for the −1 −1 Ω = ± 46 km s Mpc , m 0.27 0.03, ΩL =0.73 0.03 from flare data set. All the results presented in this paper were WMAP (Komatsu et al. 2011) is assumed. cross-checked with the independent analysis chain described in Becherini et al. (2011). Events in a circular region (ON region) centered on the hms 2. DATA ANALYSIS radio position of the source, aJ2000 = 15 55 43. 04, ( ) 2.1. H.E.S.S. Observations and Analysis dJ2000 =¢11 11 24. 4 Green et al. 1986 , with a maximum squared angular distance of 0.0125 deg2, are used for the fi H.E.S.S. is an array of ve imaging atmospheric Cherenkov analysis. In order to estimate the background in this region, telescopes located in the Khomas highland in Namibia the reflected background method (Berge et al. 2007) is used to ( ° ′ ″ ) 231618¢ S, 16 30 01 E , at an altitude of 1800 m above define the OFF regions. The excess of γ-rays in the ON region ( ) fi sea level Hinton & the H.E.S.S. Collaboration 2004 . The fth is statistically highly significant (Li & Ma 1983):21.5σ for H.E.S.S. telescope was added to the system in 2012 July and is the pre-flare period and 22.0σ for the flare. Statistics are not used in this work, reporting only on observations prior to summarized in Table 1. that time. The differential energy spectrum of the VHE γ-ray emission PG 1553+113 was observed with H.E.S.S. in 2005 and 2006 has been derived using a forward-folding method (Piron ( ) Aharonian et al. 2008 . No variability was found in these et al. 2001). For the observations prior to 2012 April, a power observations, which will be referred to as the “pre-flare” data set in the following. New observations were carried out in 2012 45 PG 1553+113 has one of the steepest spectra measured at VHE. fl 46 April after aring activity at VHE was reported by the MAGIC The difference of energy threshold between the two data set is due to the collaboration (“flare” data set, Cortina 2012b). changing observation conditions, e.g., zenith angle and optical efficiency.

117 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Figure 1. Differential fluxes of PG 1553+113 during the pre-flare (left) and flare (right) periods. Error contours indicate the 68% uncertainty on the spectrum. Uncertainties on the spectral points (in black) are given at 1σ level, and upper limits are computed at the 99% confidence level. The gray squares were obtained by the cross-check analysis chain and are presented to visualize the match between both analyses. The gray error contour on the left panel is the best-fit power law model. The lower panels show the residuals of the fit, i.e., the difference between the measured (nobs) and expected numbers of photons (nmodel), divided by the statistical error on ( ) the measured number of photons snobs . law (PWL) model fitted to the data gives a χ2 of 51.7 for 40 degrees of freedom (dof, corresponding to a χ2 probability of Pc2 = 0.10). The values of the spectral parameters (see Table 2) are compatible with previous analyses by H.E.S.S. covering the same period (Aharonian et al. 2008). A log- parabola (LP) model,47 with a χ2 of 37.5 for 39 dof (Pc2 = 0.54), is found to be preferred over the PWL model at a level of 4.3σ using the log-likelihood ratio test. Note that systematic uncertainties, presented in Table 2, have been evaluated by Aharonian et al. (2006b) for the PWL model and using the jack-knife method for the LP model. The jack-knife method consists in removing one run and redoing the analysis. This process is repeated for all runs. For the flare data set, the LP model does not significantly fi improve the t and the simple PWL model describes the data Figure 2. H.E.S.S. light curve of PG 1553+113 during the two nights of the 2 well, with a χ of 33.0 for 23 dof (Pc2 = 0.08). Table 2 flare period. The continuous line is the measured flux during the flare period contains the integral fluxes above the reference energy of while the dashed one corresponds to the pre-flare period (see Table 2 for the fl ) σ 300 GeV. The flux increased by a factor of ∼3 in the flare data ux values . Gray areas are the 1 errors. set compared to the pre-flare one with no sign of spectral variations (when comparing power law fits for both data sets). χ2 ( -5) The derived spectra and error contours for each data set are of 123.2 with 68 dof Pc2 =´6.6 10 . Restricting the fl fi χ2 presented in Figure 1, where the spectral points obtained from analysis to the pre- are data set only, the t yields a of 51.76 ( ) fl the cross-check analysis are also plotted. with 60 dof Pc2 = 0.77 , indicating again a ux increase To compute the light curves, the integrated flux above detected by H.E.S.S. at the time of the flaring activity reported 300 GeV for each observation run was extracted using the by Cortina (2012b). corresponding (pre-flare or flare) best fit spectral model. A fit Figure 2 shows the light curve during the flare together with with a constant of the run-wise light curve of the entire (pre- the averaged integral fluxes above 300 GeV of both data sets. A flare+flare) data set, weighted by the statistical errors, yields a fit with a constant to the H.E.S.S. light curve during the first 2 -3 night yields a χ of 20.76 for 6 dof (Pc2 =´2.0 10 ), 47 fi --ablog( EE0 ) The log-parabola is de ned by dNdE=F00() EE . indicating intra-night variability. This is also supported by the

118 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Table 3 Results of the Fermi-LAT Data Analysis for the Pre-ﬂare and Flare Periods

MJD Range Energy Range TS Spectral Parameters I(E > 300MeV) (GeV) (10−8 (ph cm−2 s−1)

54682–55987 0.3–300 7793.7 a = 1.49 ± 0.06stat ± 0.01sys 2.82 ± 0.1stat ± 0.2sys b = 3.8 ± 1.1stat ± 0.1sys 56040–56047 0.3–300 43.8 Γ = 1.78 ± 0.24stat ± 0.01sys 3.5 ± 1.3stat ± 0.3sys

56040–56047 0.3–80 44.5 Γ = 1.72 ± 0.26stat ± 0.01sys 3.4 ± 1.3stat ± 0.3sys

Note. For the flare period, the analysis has been performed in two energy ranges (see Section 3.2). The first two columns give the time and energy windows and the third the corresponding Test Statistic (TS) value. The model parameters and the flux above 300 MeV are given in the last two columns. The systematic uncertainties were computed using the IRFs bracketing method (Abdo et al. 2009a). use of a Bayesian block algorithm (Scargle 1998) that finds centered on the H.E.S.S. observation windows and lasts for 7 three blocks for the two nights at a 95% confidence level. days. The best fit model is a power law, the flux being consistent with the one measured during the first 3.5 yr. Data 2.2. Fermi-LAT Analysis points or light curves were computed within a restricted energy + − range or time range using a PWL model with the spectral index The Fermi-LAT is detector converting γ-rays to e e pairs frozen to 1.70. (Atwood et al. 2009). The LAT is sensitive to γ-rays from To precisely probe the variability in HE γ-rays, 7-day time 20 MeV to >300 GeV. In survey mode, in which the bulk of the bins were used to compute the light curve of PG 1553+113 in an observations are performed, each source is seen every 3 hr for extended time window (from 2008 August 4 to 2012 October approximately 30 minutes. 30), to probe any possible delay of a HE flare with respect to the The Fermi-LAT data and software are available from the fl 48 VHE one. While the ux of PG 1553+113 above 300 MeV is Fermi Science Support Center. In this work, the ScienceTools found to be variable in the whole period with a variability index V9R32P5 were used with the Pass 7 reprocessed data of Fvar =0.16 0.04 (Vaughan et al. 2003), there is no sign of (Bregeon et al. 2013), specifically SOURCE class event any flaring activity around the 2012 H.E.S.S. observations. This (Ackermann et al. 2012a), with the associated P7REP_SOUR- result has been confirmed by using the Bayesian block algorithm, CE_V15 instrument response functions. Events with energies which finds no block around the H.E.S.S. exposures in 2012. from 300 MeV to 300 GeV were selected. Additional cuts on Similar results were obtained when considering only photons the zenith angle (<100°) and rocking angle (<52°) were 49 with an energy greater than 1 GeV. No sign of enhancement of applied as recommended by the LAT collaboration to reduce the HE flux associated to the VHE event reported here was the contamination from the Earth atmospheric secondary found. This might be due to the lack of statistics at high energy in radiation. the LAT energy range. The analysis of the LAT data was performed using the Enrico Python package (Sanchez & Deil 2013). The sky 3. DISCUSSION OF THE RESULTS model was defined as a region of interest of 15° radius with γ PG 1553+113 in the center and additional point-like sources 3.1. Variability in -rays from the internal 4 yr source list. Only the sources within a The VHE data do not show any sign of variation of the 3° radius around PG 1553+113 and bright sources (integral flux ( fl fl −7 −2 −1 spectral index when comparing are and pre- are data sets greater than 5 × 10 ph cm s ) had their parameters free to with the same spectral model), and in HE no counterpart of this vary during the likelihood minimization. The template files event can be found. The indication for intra-night variability is isotrop_4years_P7_V15_repro_v2_source. similar to other TeV HBLs (Mrk 421, Mrk 501 or PKS 2155- txt for the isotropic diffuse component, and templa- 304) with, in this case, flux variations of a factor 3. te_4years_P7_v15_repro_v2.fits for the standard As noticed in previous works, PG 1553+113 presents a sharp Galactic model, were included. A binned likelihood analysis break between the HE and VHE ranges (Abdo et al. 2010a) and (Mattox et al. 1996), implemented in the gtlike tool, was the peak position of the γ-ray spectrum in the νf(ν) used to find the best-fit parameters. representation is located around 100 GeV. This is confirmed As for the H.E.S.S. data analysis, two spectral models were by the fact that the LP model better represents the pre-flare used: a simple PWL and a LP. A likelihood ratio test was used period in HE. Nonetheless, the precise location of this peak to decide which model best describes the data. Table 3 gives cannot be determined with the Fermi-LAT data only. Combin- the results for the two time periods considered in this work, and ing both energy ranges and fitting the HE and VHE data points Figure 3 presents the γ-ray spectral energy distributions. The with a power law with an exponential cutoff51 allows us to first one (pre-flare), before the H.E.S.S. exposures in 2012, determine the νf(ν) peak position for both time periods. The includes more than 3.5 yr of data (from 2008 August 4 to 2012 functional form of the model is March 1). The best fit model is found to be the LP (with a Test 50 -G Statistic of 11.3, ≈3.4σ). The second period (flare) is 2dN æ E ö E = N ç ÷ exp()-EEc . dE èç 100 GeV ø÷ 48 http://fermi.gsfc.nasa.gov/ssc/data/analysis/ 49 http://fermi.gsfc.nasa.gov/ssc/data/analysis/documentation/Cicerone/ index.html 50 Here the TS is two times the difference between the log-likelihood of the fit 51 A fit with a LP model has been attempted, but the power law with an with a LP minus the log-likelihood with a PL. exponential cutoff leads to a better description of the data.

119 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al. the results). In the modest redshift range of VHE emitters ( ) detected so far z ⩽ 0.6 , the EBL absorption is negligible below 80 GeV (tgg ~ 0.1 at 80 GeV for z = 0.6). A measure of the EBL energy density was obtained by Ackermann et al. (2012b) and Abramowski et al. (2013b) based on the spectra of sources with a known z. In the case of PG 1553+113, for which the redshift is unknown, the effects of the EBL on the VHE spectrum might be used to derive constraints on its distance. Ideally, this would be done by comparing the observed spectrum with the intrinsic one but the latter is unknown. The Fermi-LAT spectrum, derived below 80 GeV, can be considered as a proxy for the intrinsic spectrum in the VHE regime, or at least, as a solid UL (assuming no hardening of the spectrum). Following the method used by Abramowski et al. (2013a),it Figure 3. Spectral energy distribution of PG 1553+113 in γ-rays as measured has been assumed that the intrinsic spectrum of the source in by the Fermi-LAT and H.E.S.S. Red (blue) points and butterflies have been the H.E.S.S. energy range cannot be harder than the obtained during the flare (pre-flare) period. The Fermi and H.E.S.S. data for the extrapolation of the Fermi-LAT measurement. From this, one fl – pre- are are not contemporaneous. H.E.S.S. data were taken in 2005 2006 can conclude that the optical depth cannot be greater than while the Fermi data were taken between 2008 and 2012. τmax(E), given by:

For this purpose, Fermi-LAT and H.E.S.S. systematic uncer- é ù ê fint ú tainties were taken into account in a similar way as in tmax ()E = lnê ú,(1) (1--Daf ) 1.64 f Abramowski et al. (2014) and added quadratically to the ëê ()obs obs ûú statistical errors. The Fermi-LAT systematic uncertainties were ϕ estimated by Ackermann et al. (2012a) to be 10% of the where int is the extrapolation of the Fermi-LAT measurement fl effective area at 100 MeV, 5% at 316 MeV and 15% at 1 TeV toward the H.E.S.S. energy range. fobsDf obs is the ux and above. For the VHE γ-ray range, they were taken into measured by H.E.S.S. The factor (1 − α)=0.8 accounts for the account by shifting the energy by 10%. This effect translates systematic uncertainties of the H.E.S.S. measurement and the into a systematic uncertainty for a single point of number 1.64 has been calculated to have a confidence level of ( ) s()ffEsys =¶¶ 0.1· where f is the differential flux at energy 95% Abramowski et al. 2013a . The comparison is made at E. the H.E.S.S. decorrelation energy where the flux is best The results of this parameterization are given in Table 4. measured. τ Using the pre-flare period, the peak position is found to be Figure 4 shows the 95% UL on max. The resulting UL on the redshift is z < 0.43. This method does not allow the located at log10 (Emax 1 GeV)= 1.7 0.2 stat  0.4 sys with no evidence of variation during the flare and no spectral variation. statistical and systematic uncertainties of the Fermi- This is consistent with the fact that no variability in HE γ-rays LAT measurement to be taken into account and does not take was found during the H.E.S.S. observations. This is also in advantage of the spectral features of the absorbed spectrum agreement with the fact that HBLs are less variable in HE γ- (see Abramowski et al. 2013b). rays than other BL Lac objects (Abdo et al. 2010b), while A Bayesian approach has been developed with the aim of numerous flares have been reported in the TeV band. taking all the uncertainties into account. It also uses the fact that EBL-absorbed spectra are not strictly power laws. The details of the model are presented in Appendix A and only the main 3.2. Constraints on the Redshift assumptions and results are recalled here. Intrinsic curvature between the HE and VHE ranges that naturally arises due to The EBL is a field of UV to far-infrared photons produced either curvature of the emitting distribution of particles or by the thermal emission from stars and reprocessed starlight by emission effects (e.g., Klein–Nishina effects) is permitted by dust in galaxies (see Hauser & Dwek 2001, for a review) that construction of the prior (Equation (A.1)). A spectral index interacts with very high energy γ-rays from sources at softer than the Fermi-LAT measurement is allowed with a cosmological distances. As a consequence, a source at redshift -t(,)Ez constant probability, in contrast with the previous calculation. It z exhibits an observed spectrum fobs()EEe=´f int () γ τ is assumed that the observed spectrum in VHE -rays cannot be where fint ()E is the intrinsic source spectrum and is the harder than the Fermi-LAT measurement by using a prior that optical depth due to interaction with the EBL. Since the optical γ fl follows a Gaussian for indices harder than the Fermi-LAT one. depth increases with increasing -ray energy, the integral ux The prior on the index is then: is lowered and the spectral index is increased.52 In the following, the model of Franceschini et al. (2008) was used  P ()GµG () GG ,Fermi ,sG (2) to compute the optical depth τ as a function of redshift and energy. In this section, the data taken by both instruments if G < GFermi and during the flare period are used, with the Fermi-LAT analysis restricted to the range 300 MeV < E < 80 GeV (see Table 3 for P ()Gµ 1

52 otherwise. G is the index measured by Fermi-LAT and s is For sake of simplicity it is assumed here that the best-ﬁt model is a power Fermi G law, an assumption which is true for most of the cases due to limited statistics the uncertainty on this measurement that takes all the systematic in the VHE range. and statistical uncertainties into account.

120 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Table 4 Parametrization Results of the Two Time Periods (First Column) Obtained by Combining H.E.S.S. and Fermi-LAT ( = ) Γ Period N E 100 GeV log10 (Ec 1 GeV) log10 (Emax 1 GeV) − (10 11 erg cm--21 s )

Pre-ﬂare 9.6 ± 0.7stat ± 1.7sys 1.59 ± 0.02stat ± 0.03sys 2.03 ± 0.02stat ± 0.04sys 1.7 ± 0.2stat ± 0.4sys Flare 13.0 ± 3.5stat ± 5.7sys 1.56 ± 0.08stat ± 0.11sys 2.16 ± 0.04stat ± 0.09sys 1.8 ± 0.7stat ± 1.3sys

Note. The second column gives the normalization at 100 GeV, while the third and the fourth present the spectral index and cut-off energy of the ﬁtted power law with an exponential cut-off. The last column is the peak energy in a nnf ( ) representation.

Table 5 found in Mattingly (2005) and Liberati (2013). An energy- Calibrated 95% 1-Sided LL and UL (including systematic errors) on the dependent dispersion in vacuum is searched for in the data by Dispersion Parameter tn and Derived 95% one-sided Lower Limits on EQG testing a correlation between arrival times of the photons and −n their energies. For two photons with arrival times t and t and Limits on τn (s TeV ) Lower Limits on EQG (GeV) 1 2 calib + syst energies E1 and E2, the dispersion parameter of order n is n calib+ syst UL s = −1 s = +1 LL ﬁ tt21- Dt ( = ) 17 17 de ned as tn ==nn n . Here only the linear n 1 1 −838.9 576.4 2.83 × 10 4.11 × 10 EE21- D()E 10 10 ( = ) 2 −1570.5 1012.4 1.68 × 10 2.10 × 10 and quadratic n 2 dispersion parameters are calculated. Assuming no intrinsic spectral variability of the source, the dispersion τn can be related to the normalized distance of the source κn corrected for the expansion of the universe and an energy EQG at which QG effects are expected to occur (Jacob & Piran 2008): Dt (1+ n ) tk=  s (3) n n  n n D()E EHQG 2 0

where H0 is the Hubble constant and s± = −1 (resp. +1) in the superluminal (resp. subluminal) case, in which the high-energy photons arrive before (resp. after) low-energy photons. The normalized distance κn is calculated from the redshift of the L source z and the cosmological parameters Ωm, Ω given in the introduction: z (1+¢zdz )n ¢ kn = (4) ò03 Ω(1m +¢z ) + ΩL = Figure 4. Values of τmax as a function of the photon energy. The black line is Using the central value of z 0.49 determined in Section 3.2, the 95% UL obtained with the H.E.S.S. data and the red line is the optical depth the distance κn for n = 1 and 2 is κ1 = 0.541 and κ2 = 0.677. computed with the model of Franceschini et al. (2008) for a redshift of 0.43. First, the dispersion measurement method will be described. The blue line is the decorrelation energy for the H.E.S.S. analysis. The gray fl ( lines are the value of optical depth for different redshifts. It will then be applied to the H.E.S.S. are dataset Monte Carlo (MC) simulations and original dataset), in order to measure the dispersion and provide 95% 1-sided lower and The most probable redshift found with this method is τ τ = ± upper limits on the dispersion parameter n. These limits on n z 0.49 0.04, in good agreement with the independent will lead to lower limits (LLs) on E using Equation (3). measure of Danforth et al. (2010), who constrained the QG distance to be between 0.43 < z < 0.58. Figure 5 gives the 3.3.1. Modified Maximum Likelihood Method posterior probability obtained with the Bayesian method compared with other measurements of z. Lower and upper A maximum likelihood method, following Martinez & limits at a confidence level of 95% can also be derived as Errando (2009), was used to calculate the dispersion parameter 0.41 < z < 0.56. Note that this method allows the systematic τn. Albert et al. (2008) applied this method to a flare of uncertainties of both instruments (Fermi-LAT and H.E.S.S.) to Mkn 501, while Abramowski et al. (2011) applied it to a flare be taken into account. The spectral index obtained when fitting of PKS 2155-304. More recently, it was used by Vasileiou et al. the H.E.S.S. data with an EBL absorbed PWL using a redshift (2013) to analyze Fermi data of four gamma-ray bursts. The of 0.49 is compatible with the Fermi measurement below data from Cherenkov telescopes are contaminated by π0 decay 80 GeV. from proton showers, misidentified electrons, or heavy elements such as helium. In the case of PG 1553+113, and 3.3. Lorentz Invariance Violation contrary to previous analyses, this background is not negligible: the signal-over-background ratio S/B is about 2, As stated in Section 2.1, the H.E.S.S. data of the flare show compared to 300 for the PKS 2155-304 flare event of 2006 July an indication of intra-night variability, which is used here to (Aharonian et al. 2007b). The background was included in the test for a possible LIV. Some Quantum Gravity (QG) models formulation of the probability density function (PDF) used in a predict a change of the speed of light at energies close to the likelihood maximization method. Given the times ti and 19 Planck scale (~10 GeV). A review of such models can be energies Ei of the gamma-like (ON) particles received by the

121 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Figure 5. Posterior probability density as a function of redshift (red). The blue area represents the redshift range estimated by Danforth et al. (2010) while the green dashed line indicates the limit of Sanchez et al. (2013). detector, the unbinned likelihood function of the dispersion parameter τn is:

n ON Figure 6. Time distribution of the excess ON − αOFF in the first six runs LPEt()ttn =  ()iin , . (5) (70971–70976), with energies between 300 and 400 GeV. T = 0 corresponds i=1 to the time of the first detected event in run 70971. The vertical bars correspond to 1σ statistical errors; the horizontal bars correspond to the bin width in time. The PDF PE(,iin t∣t )associated with each ON event is The best fit, in red, was used as the template light curve in the maximum composed of two terms: likelihood method; the ±1σ error envelope is shown in green.

PE()iin,·, ttt= w s PSig () E iin t

+-()()1·wPsiiBkg Et , (6) 3.3.2. Specific Selection Cuts and Timing Model with The flare data set of the H.E.S.S. analysis (see Section 2.1) was used with additional cuts. To perform the dispersion 1 studies, only uninterrupted data have been kept. Thus, the PEtSig()iin, t = AEt eff ()ii, N ()t analysis was conducted on the first seven runs, taken during the n fl n night of April 26. Moreover, the cosmic ray ux increases ´LSig()EFiin Sig () t -t ·(7) Ei substantially for the seventh run, due to a variation of the zenith angle during this night. This fact, along with its large statistical 1 errors, leads us to discard this run from the analysis. The sixth PEtBkg()ii, = AEt eff () ii,(8)L Bkg () EF i Bkg N¢ run shows little to no variability and was therefore also removed from the LIV analysis. Since within the ON data set, nnON- a OFF the signal and the background spectra have different indices ws = .(9)( ) nON GSig = 4.8 for the signal and GBkg = 2.5 for the background , the ratio S/B is expected to decrease with increasing energy. An The PDF PSig includes the emission time distribution of the upper energy cut at Emax = 789 GeV was set, corresponding to photons FSig determined from a parameterization of the the last bin with more than 3σ significance in the reconstructed observed light curve at low energies (discussed in the next photon spectrum (see the differential flux during the flare in n section) and evaluated on tE- tn · to take into account the Figure 1). A lower cut on the energy at Emin = 300 GeV was delay due to a possible LIV effect, the measured signal used in order to avoid large systematic effects arising from high spectrum ΛSig and the effective area Aeff. The PDF PBkg is uncertainties on the H.E.S.S. effective area at lower energies. composed of the uniform time distribution FBkg of the The intrinsic light curve of the flare, needed in the formulation background events, the measured background spectrum ΛBkg of the likelihood, can be obtained from a model of the timed and the effective area Aeff. No delay due to a possible LIV emission or approximated from a subset of the data. To be as effect is expected in the background events of the ON data set. model-independent as possible, it was here derived from a fitof N(τn) and N′ are the normalization factors of PSig and PBkg the measured light curve at low energies (with E < Ecut). The respectively, in the (E, t) range of the likelihood fit. The high-energy events (E > Ecut) were processed in the calcula- coefficient ws corresponds to the relative weight of the signal tion of the likelihood to search for potential dispersion. Here events in the total ON data set, derived from the number of Ecut was set to Ecut = 400 GeV, which is approximately the events in the ON region nON and the number of events in the median energy of the ON event sample. Other cuts on the OFF regions nOFF weighted by the inverse number of OFF energy did not introduce significant effects on the final results. regions α. More details on the derivation of this function are The histogram and the fit (Figure 6) were obtained as follows: given in Appendix B.1. the main idea was to preserve the maximum detected variability

122 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Figure 7. Lower limits on EQG,1 from linear dispersion (left) and on EQG,2 from quadratic dispersion (right) for the subluminal case (s = +1) obtained with AGNs as a function of redshift. The limits are given in terms of EPlanck. The constraints from Mkn 421 have been obtained by Biller et al. (1999), from Mkn 501 by Albert et al. (2008), and from PKS 2155-304 by Abramowski et al. (2011). in the PG 1553+113 flare, together with a significant response intrinsic dispersion was artificially added. Each simulated data in each observed peak: set produces a LL and an UL on τn. The calibrated lower (upper) limit of the confidence interval is obtained from the 1. The binning was chosen so that at least two adjacent bins mean of the distribution of the per-set individual lower (upper) of the distribution yield a minimum of 3σ excess with limits. Both CIs (from the data only and from the simulated respect to the average value. sets) are listed in Table B2. Sources of systematic errors 2. Simple parameterizations have been tested on the whole include uncertainties on the light curve parameterization, the data set (all energies): constant (χ2/dof = 25/12), single background contribution, the calculation of the effective area, Gaussian (χ2/dof = 20/10) and double Gaussian (χ2/ the energy resolution, and the determination of the photon dof = 8.5/7) functions. The latter is preferred, since it index (see Appendix B.4). improves the quality of the fit. This shape was chosen to The resulting limits on the dispersion τ using the quadratic fit the low energy subset of events. Choosing a single n sum of the statistical errors from the simulations and the Gaussian parameterization would result in a decrease of systematic errors determined from data and simulations were the sensitivity to time-lag measurements by a factor computed, leading to limits on the energy scale E (Equa- of two. QG tion (3)). The 95% 1-sided LLs for the subluminal case (s = +1) ∼ 17 10 There is a gap of 2 minutes between each two consecutive are: EQG,1 >4.11×10 GeV and EQG,2 >´2.10 10 GeV for runs. We did not consider the effect of these gaps as it is small linear and quadratic LIV effects, respectively. For the super- ∼ 17 with respect to the bin width of 10 minutes. More luminal case (s = –1) the limits are: EQG,1 >´2.83 10 GeV 10 importantly, their occurrence is not correlated with the binning: and EQG,2 >1.68×10 GeV for linear and quadratic LIV effects, one gap falls in the rising part of the light curve, one is at a respectively. Figure 7 shows a comparison of the different LLs on maximum, two fall in the decreasing parts and none of the gaps EQG,1 and EQG,2 for the subluminal case (s = +1) obtained with is at the minimum. AGNs at different redshifts studied at VHE. All these limits, Table B1 in Appendix B.2 shows the number of ON and including the present results, have been obtained under the OFF events for the different cuts applied to the data. assumption that no intrinsic delays between photons of different energies occur at the source. For the linear/subluminal case, the most constraining limit on EQG with transient astrophysical events 3.3.3. Results: Limits on τn and EQG 19 has been obtained with GRB 090510: EQG,1 >6.3×10 GeV The maximum likelihood method was performed using high- (Vasileiou et al. 2013). The most constraining limits on EQG with ( ) energy events with Ei > Ecut. First, confidence intervals (CIs) AGN so far have been obtained by Abramowski et al. 2011 corresponding to 95% confidence level (1-sided) were with PKS 2155-304 data observed with H.E.S.S.: 18 10 determined from the likelihood curve at the values of τn where EQG,1 >´2.1 10 GeV and EQG,2 >6.4×10 GeV for linear the curve reaches 2.71, which corresponds to the 90% CL and quadratic LIV effects, respectively (95% CL, 1-sided). quantile of a χ2 distribution. However, these CIs are derived Compared to the PKS 2155-304 limits, the limits on the linear from one realization only and do not take into account the dispersion for PG 1553+113 are one order of magnitude less “luckiness” factor of this measurement. To get statistically constraining, but the limits on the quadratic dispersion are of the significant CIs (“calibrated CIs”), several sets were generated same order of magnitude since the source is located at a higher with MC simulations, with the same statistical significance, redshift. This highlights the interest in studying distant AGNs, in light curve model and spectrum as the original data set. No spite of the difficulties due to limited photon statistics.

123 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al. 4. CONCLUSIONS Department of Energy in the United States, the Commissariat àl’Energie Atomique and the Centre National de la Recherche A VHE γ-ray flaring event of PG 1553+113 has been Scientifique/Institut National de Physique Nucléire et de detected with the H.E.S.S. telescopes, with a flux increasing by Physique des Particules in France, the Agenzia Spaziale a factor of 3. No variability of the spectral index has been found Italiana and the Istituto Nazionale di Fisica Nucleare in Italy, in the data set, but indication of intra-night flux variability is the Ministry of Education, Culture, Sports, Science and reported in this work. In HE γ-rays, no counterpart of this event Technology (MEXT), High Energy Accelerator Research can be identified, which may be interpreted as the sign of Organization (KEK) and Japan Aerospace Exploration Agency injection of high energy particles emitting predominantly in (JAXA) in Japan, and the KA Wallenberg Foundation, the VHE γ-rays. Such particles might not be numerous enough to fi fl Swedish Research Council and the Swedish National Space have a signi cant impact on the HE ux during either their Board in Sweden. Additional support for science analysis acceleration or cooling phases. during the operations phase is gratefully acknowledged from The data were used to constrain the redshift of the source the Istituto Nazionale di Astrofisica in Italy and the Centre using a new approach based on the absorption properties of the National d’Etudes Spatiales in France. DS work is supported by EBL imprinted in the spectrum of a distant source. Taking into the LABEX grant enigmass. The authors want to thank F. account all the instrumental systematic uncertainties, the Krauss for her useful comments. redshift of PG 1553+113 is determined as being z = 0.49 ± 0.04. APPENDIX A Flares of variable sources can be used to probe LIV effects, BAYESIAN MODEL USED TO CONSTRAIN THE manifesting themselves as an energy-dependent delay in the REDSHIFT photon arrival time. A likelihood method, adapted to flares with a large amount of background and modest statistics, was A Bayesian approach has been used to compute the redshift presented. To demonstrate the analysis power of this method, it value of PG 1553+113 in Section 3.2. The advantage of such a was applied to the H.E.S.S. data of a flare of PG 1553+113. model is that systematic uncertainties, which are important in This analysis relies on the indication of the intra-night Cherenkov astronomy, can easily be included in the calcula- variability of the flare at VHE. No significant dispersion was tion. In the following, the notation Θ for the model parameters measured, and limits on the EQG scale were derived, in a region and Y for the data set is adopted. All normalization constants of redshift unexplored until now. Limits on the energy scale at are dropped in the development of the model, and the final which QG effects causing LIV may arise, derived in this work, probability is normalized at the end. 17 10 Bayes’ Theorem, based on the conditional probability rule, are EQG,1 >´4.11 10 GeV and EQG,2 > 2.10 × 10 GeV for the subluminal case. Compared with previous limits obtained allows us to write the posterior probability PY(Q∣ ) for the Θ with the PKS 2155-304 flare of 2006 July, the limits for model parameters as the product of the likelihood PY()∣Q PG 1553+113 for a linear dispersion are one order of and the prior probability P(Θ): magnitude less constraining while limits for a quadratic PY()QµQ PPY()() Q. dispersion are of the same order of magnitude. With the new telescope placed at the center of the H.E.S.S. array that The likelihood is the quantity that is maximized during fi ( ) provides an energy threshold of several tens of GeV, a better determination of the best- t spectrum Piron et al. 2001 .Itis fl picture of the variability patterns of AGN flares should be at this step that the H.E.S.S. data, taken during the are, were obtained. The future Cherenkov Telescope Array will increase actually used. The spectrum model here is a simple power law the number of flare detections (Sol et al. 2013) with better corrected for the EBL absorption: sensitivity, allowing for the extraction of even more constrain- -G f =ŃEEe ´-t(,)Ez. ing limits on the LIV effects. ()0 The model parameters are then N, G, and z. The support of the Namibian authorities and of the The prior is the most difficult and most interesting part of University of Namibia in facilitating the construction and the model. To derive it, N and Γ are assumed to be independent operation of H.E.S.S. is gratefully acknowledged, as is the from each other and independent of the redshift. In contrast, support by the German Ministry for Education and Research the prior on the redshift might depend on N and Γ. Then, (BMBF), the Max Planck Society, the French Ministry for the prior can be simplified using the conditional probability Research, the CNRS-IN2P3 and the Astroparticle Interdisci- rule: plinary Programme of the CNRS, the UK Particle Physics and Astronomy Research Council (PPARC), the IPNP of the PPzNPNP()Q=( , G) ()() G Charles University, the South African Department of Science As much as possible, weak assumptions should be made to and Technology and National Research Foundation, and the write a robust prior then often flat priors (i.e., P ∝ const) are University of Namibia. We appreciate the excellent work of the used. Priors should also be based on a physical meaning and technical support staff in Berlin, Durham, Hamburg, Heidel- not contradict the physical and observed properties of the berg, Palaiseau, Paris, Saclay, and Namibia in the construction objects. For the purpose of this model, the prior on N is and operation of the equipment. assumed to be flat and the prior on the spectral index is a The Fermi LAT Collaboration acknowledges generous truncated Gaussian P ()GµG (, GGFermi ,sG ) if G

124 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

LAT measurement. σΓ takes into account the statistical and Table B1 systematic uncertainties on the Fermi-LAT measurement and Selections Applied to the ON and OFF Data Sets also the systematic uncertainty on the H.E.S.S. spectrum (σ = ) #ofON Weighted # of OFF 0.20, see Aharonian et al. 2006b added quadratically and Selection Events Events S/B σ = Γ = Γ 0.33 for a mean value of Fermi 1.72. ( ) ( ) The prior on z is much more difficult to determine. A flat Total sample 461 100% 144.3 100% 2.2 (1)=Time in 500–8500s 358 (77.7%) 95.8 (66.4%) 2.7 prior has no physical motivations since the probability to detect (1) and E in 0.3–0.789 TeV 154 (33.4%) 36.3 (25.1%) 3.2 sources at TeV energy decreases with the redshift. The number ( ) – ( ) ( ) fi 1 and E in 0.3 0.4 TeV 82 17.8% 14.2 9.9% 4.8 of sources detected at TeV energy is not suf cient to use the (Template) corresponding redshift distribution as a prior. (1) and E in 0.4–0.789 TeV 72 (15.6%) 21.9 (15.2%) 2.3 A prior which takes into account the EBL can be derived (LH fit) assuming a population of sources with a constant spatial density. In the small space element 4πz2 dz, the number of such sources scales ∝z2. For any given luminosity, their flux (which scales with the probability to detect them) is scaled by z-2 exp (-t (z )). Lacking a proper knowledge of the intrinsic luminosity function of VHE γ-ray blazars, a reasonable assumption on the detection probability of a blazar at any redshift is a scaling proportional to the flux for a given luminosity, i.e., µ-zz-2 exp (t ( )). Putting everything together, the prior on the redshift reads PzN(,)∣ G = Pz()µ- exp(t () z). Finally, the prior we use for our analysis is:

Pz(Qµ ) exp ( -t ( ))G ( G , 1.72, 0.33) (A.1) if Γ < 1.72 and Pz()Qµ exp( -t ()) otherwise. Putting all the components of the model together and marginalizing over the nuisance parameters N and Γ, the probability on the redshift can be computed numerically. The obtained mean value is z = 0.49 ± 0.04. At a confidence level of 95%, the redshift is between 0.41 < z < 0.56. In this work, only the model of Franceschini et al. (2008) has been used. Other EBL models available in the literature predict slightly different absorption depths. This will lead to a small difference in the redshift. The use of a flat prior for the Figure B1. Means of the reconstructed dispersion vs. the real (injected dispersion) for the linear case n = 1; for a given injected dispersion, errors redshift distribution of the sources or a prior based on estimates bars correspond to the means of the distribution of the upper and lower limits of the HBLs luminosity function (Ajello et al. 2014) leads to (90% 2-sided  95% 1-sided). The blue line is a linear fit to the points. The changes of order of 0.01 on the resulting redshift. red line shows the ideally obtained curve ttrecontructed= injected obtained in the case S/B = ¥. APPENDIX B DEVELOPMENT OF THE LIV METHOD terms (signal and background): B.1 Modified Maximum Likelihood Method PE,,, tsbtt= w · P E , t In previous LIV studies with AGN flares (Albert et al. 2008; ()()ii n sSig iin Abramowski et al. 2011) the signal was clearly dominating +-()()1·wPsiiBkg Et ,(B.2) over the background, whereas in the present study the signal- over-background ratio is about 2. The background has been with included in the formulation of the PDF: in the most general s case, for given numbers of signal and background events s and ws = .(B.3) b in the observation region (“ON” region), for a given sb+ dispersion parameter τn, the unbinned likelihood is: nON is the number of events detected in the source ON region fi æ ö included in the t range [;EEcut max ][;´ t min t max]. nOFF is the ç b ÷ Ln()()ON,,,Pois·Pois n OFF sbtn =+ n ON s bç n OFF ÷ number of events in the OFF regions, in the same (E, t) range; èç a ø α is the inverse number of OFF regions. Pois(nsbON ∣ + ) nON ( ) ·,,,(B.1)PE tsbt Pois(nbOFF ∣ a ) is the Poisson distribution with index nON  ()ii n (n ) s b (b α) i=1 OFF and parameter + / . The likelihood function can be simplified by fixing s and b from a comparison of ON and = − α = α The PDF PE(,ii tsb∣ ,,t n )associated with each gamma-like OFF sets: s nON nOFF and b nOFF. In this case, the particle characterized by its time ti and energy Ei contains two Poisson terms in Equation (B.2) are equal to 1. The

125 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

Table B2 Linear (Top) and Quadratic (Bottom) Dispersion Parameters; from left to right: Best Estimate, LL and UL from Data (Cut on Likelihood Curve), LL and UL from MC Simulations (means of Per-set LL and UL Distributions), Calibrated LL and UL (Combination of Data and MC), Calibrated LL and UL Including Systematic Errors

calib calib data data data MC MC MC calib calib LLn ULn n tn,best LLn ULn tn,best LLn ULn LLn ULn With Systematics 1 −131.7 −806.7 554.7 99.1 −526.3 725.6 −757.1 494.8 −838.9 576.4 2 −287.5 −1449.9 853.6 217.2 −942.0 1395.0 −1446.7 890.3 −1570.5 1012.4

−n Note. Dispersion Parameters τn,best, LLs and ULs are in s TeV .

probabilities PSig and PBkg are defined as: B.2 Selection Cuts 1 Table B1 shows the effect of the selection cuts on the PEtSig()iin, t = ·,REt Sig ()iint (B.4)number of ON and OFF events. Other choices of E and E N ()t min cut n did not introduce significant changes in the final results. 1 PEtBkg()ii, = ·,REt Bkg () ii (B.5) N¢ B.3 Test of the Method, CIs with The method has been tested on MC simulated sets. Each set was composed of n = 72 ON events, as in the real data ¥ ON sample: REtSig (),,,tn = ò DEEAE (true )() eff true t Etrue=0 1. s = 50 signal events with times following the template ´LEFt -t · EdEn Sig() true Sig ()n true true light curve (Figure 6) shifted by a factor tn,inj · Ei; (B.6) energies follow a power law spectrum of photon index GSig = 4.8, degraded by the acceptance and convolved ¥ with the energy resolution. REtBkg (,)= ò DEE() , true = Etrue=0 2. b 22 background events with times following a uniform distribution and energies drawn from a power ´LAEeff()() true,(). t Bkg E true F Bkg tdE true law spectrum of index GBkg = 2.5, degraded by the (B.7) acceptance and convolvted with the energy resolution. ( ) For a given injected dispersion, the maximum likelihood PEtSig (,iin∣t ) is the probability that the event Ei, ti is a photon n method is applied to each MC-simulated set. The initial light emitted at the source and detected on Earth with a delay tn E .It takes into account the emission (time distribution Ft()and curve and energy spectrum were used as templates in the model Sig fi energy spectrum L ()E at the source), the propagation (delay instead of tting them for each set. Sig Figure B1 shows the means of the reconstructed dispersion t n ) n · Ei due to possible LIV effect and the detection of a versus the real (injected) dispersion for n = 1; for a given photon by the detector (H.E.S.S. energy resolution D EE (, true ) injected dispersion, error bars correspond to the rms of the and effective area Aeff (,)Et). PEtBkg (,)iiis the probability that ( ) distribution of the best estimates tˆ1. The blue line shows the the event Ei, ti is a background event; it is not expected to be result of a linear fit. The slope roughly corresponds to the variable with time, thus FBkg(t) is a uniform time distribution: Λ percentage of signal in the total ON data set. It is due to the loss FtBkg()= F Bkg. The background energy distribution Bkg is of sensitivity resulting from the part of the data sets with no (τ )( ′) − measured from OFF regions. N n resp. N is the normal- dispersion. A systematic shift is observed of about 100 s TeV 1, ( ) ization factor of the PDF PSig resp. PBkg in the range well bellow 1σ value—the rms of the best estimate distribution fi − [;EEcut max ][;´ t min t max] where the likelihood t is performed. is 361 s TeV 1. The results in this paper have not been ( ) Also, the energy resolution D E, Etrue is assumed to be corrected for this bias. [ ] 53 fi perfect in the range Ecut; Emax . This leads to simpli ed The coverage is not necessarily proper, i.e. the number of expressions of PEtSig (,iint ) and PEtBkg (,)ii: ∣ sets for which the injected dispersion value τinj lies between ʼ 1 the set s LL and UL does not match the required 95% 1-sided PEtSig()iin, t = ·,AEt eff ()ii confidence level. The common cut used on the likelihood N ()tn curves to get the LLs/ULs has been iteratively adjusted to n ´LSig()EFiin Sig () t -t ·(B.8) Ei ensure a correct statistical coverage: using this new cut, 95% of the realizations provide CIs that include the injected 1 dispersion tn,inj. The initial coverage was about 85% for a cut PEtBkg()ii, = ·,AEt eff ()() iiL Bkg EF i Bkg (B.9)on 2 ln L of 2.71. The new common cut, found iteratively at N¢ 3.5, ensures the desired 90% 2-sided CL (approx. 95% one- ) The best estimate of the dispersion parameter tn is obtained by sided CL .FigureB2 shows the distributions of the best −1 maximizing the likelihood L(τn). estimates, the 95% 1-sided LLs and ULs for t1,inj = 0 sTeV −2 (linear case) and t2,inj = 0 sTeV (quadratic case);the 53 The actual energy resolution is of the order of 10% in this range. means of the lower and upper limit distributions, shown as a

126 A.4. The 2012 ﬂare of PG 1553+113 seen with H.E.S.S. and Fermi-LAT

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al.

−n Figure B2. Distributions of the best estimates, the 95% one-sided lower and upper limits from simulations in case of no injected dispersion (tn,inj = 0 s TeV ), for n = 1 (top) and n = 2 (bottom); dispersion values are in s TeV−n. The blue vertical line on the LL (resp. UL) distribution shows LLMC (resp. ULMC),defined as the mean of the distribution. blue vertical line, are used to construct the “calibrated Table B3 confidence interval.” Summary of all Studied Systematic Contributions. The Main Systematic Errors To get CIs from data, a maximum likelihood method is are due to the Uncertainties on the Light Curve Parametrization data τ τ applied to the original data set and gives a best estimate tbest . Estimated Error 1 2 The cut value determined from the simulations to ensure proper on Input Parameters (s TeV−1)(s TeV−2) data coverage is applied on the original data set to obtain LL and Background contribution L <45 <80 data calib calib UL . The “calibrated” limits LL and UL , combining Acceptance factors 10% <1 <1 data tbest from data together with MC results, are taken as Energy resolution 10% <55 <85 Photon index 5% <55 <50 L LLcalib =-ttdata MC - LLMC Light curve <300 <500 best best parameterization calib data MC MC L ∼ ∼ UL=+ttbest best - UL (B.10) Systematic bias 100 200 2 <330 <555 MC MC MC fi Total: åi systi with tbest , LL and UL de ned as the means of the per-set best-estimate distribution, LL distribution, and UL distribution respectively. Table B2 lists the CIs determined in both ways, i.e., data- by Abramowski et al. (2011) and GRB studies by Vasileiou data calib ( data only and calibrated ones: LLn and LLn resp. ULn and et al. (2013). calib) ULn are compatible within 10%. In this work, calibrated CIs have been used to derive the final LLs on E . They are QG B.4 Estimation of the Systematics preferred over data-only CIs as they provide statistically well defined confidence levels. They also ensure coherent compar- Estimations of the systematic effects on the dispersion ison with previous published results, e.g., with PKS 2155-304 measurement were performed. It was found that the main

127 A. Selected publications

The Astrophysical Journal, 802:65 (14pp), 2015 March 20 Abramowski et al. systematic errors are due to the uncertainties on the light curve Ackermann, M., Ajello, M., Allafort, A., et al. 2012b, Sci, 338, 1190 parameterization. Other sources of systematic errors include the Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006a, A&A, contribution of the background, effect of the change of photon 448, L19 Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2006b, A&A, index, the energy resolution and the effective area determina- 457, 899 tion of the detector. To study the following four contributions, Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2007a, ApJL, new simulated data sets have been built, each one with different 664, L71 input parameters: Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2007b, ApJL, 664, L71 1. background contribution: photons and background events Aharonian, F., Akhperjanian, A. G., Barres de Almeida, U., et al. 2008, A&A, have been reallocated within the ON data set in the ﬁt 477, 481 [ ] σ ﬂ Ajello, M., Romani, R. W., Gasparrini, D., et al. 2014, ApJ, 780, 73 range Ecut; Emax , introducing a 1 uctuation in the Albert, J., Aliu, E., Anderhub, H., et al. 2008, PhLB, 668, 253 number of signal event s in the ON data set; Aleksić, J., Alvarez, E. A., Antonelli, L. A., et al. 2012, ApJ, 748, 46 2. effective area: set to a constant, equal to 120000 m2 for Aleksić, J., Ansoldi, S., Antonelli, L. A., et al. 2014, arXiv:1408.1975 all energies and all times, which corresponds to a Aliu, E., Archer, A., Aune, T., et al. 2015, ApJ, 799, 7 ( Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, ApJ, 697, 1071 maximum shift of 10% the actual effective area increases Becherini, Y., Djannati-Ataï, A., Marandon, V., Punch, M., & Pita, S. 2011, with energy); APh, 34, 858 3. energy resolution: reconstructed energies have been Berge, D., Funk, S., & Hinton, J. 2007, A&A, 466, 1219 replaced by the true energies; this corresponds to a shift Biller, S. D., Breslin, A. C., Buckley, J., et al. 1999, PhRvL, 83, 2108 of about 10% on the reconstructed energy values; Bregeon, J., Charles, E., & Wood, M. for the Fermi-LAT collaboration 2013, arXiv:1304.5456 4. photon index: changed by one standard devia- Cortina, J. 2012a, ATel, 3977, 1 tion (±0.25). Cortina, J. 2012b, ATel, 4069, 1 Danforth, C. W., Keeney, B. A., Stocke, J. T., Shull, J. M., & Yao, Y. 2010, For the determination of systematic errors arising from the ApJ, 720, 976 light curve parameterization, the calibration of the CIs has been de Naurois, M., & Rolland, L. 2009, APh, 32, 231 redone using successively the upper 1σ and the lower 1σ Franceschini, A., Rodighiero, G., & Vaccari, M. 2008, A&A, 487, 837 contours of the template, shown in Figure 6. The change in Gaidos, J. A., Akerlof, C. W., Biller, S. D., et al. 1996, Natur, 383, 319 τ Giommi, P., Ansari, S. G., & Micol, A. 1995, A&AS, 109, 267 mean lower and ULs on the dispersion parameter n gives an Green, R. F., Schmidt, M., & Liebert, J. 1986, ApJS, 61, 305 estimate of the systematic error associated to each contribu- Hauser, M. G., & Dwek, E. 2001, ARA&A, 39, 249 tion.54 An additional systematic contribution comes from the Hinton, J. A. The H.E.S.S. Collaboration 2004, NewAR, 48, 331 shift arising from the method found with simulation (see Jacob, U., & Piran, T. 2008, J. Cosmol. Astropart. Phys., 2008, 031 Appendix B.3). Table B3 summarizes all studied systematic Komatsu, E., Smith, K. M., Dunkley, J., et al. 2011, ApJS, 192, 18 τ Li, T.-P., & Ma, Y.-Q. 1983, ApJ, 272, 317 contributions. The overall estimated systematic error on n is Liberati, S. 2013, CQGra, 30, 133001 −1 −2 330 s TeV for the linear case (n = 1) and 555 s TeV for the Martinez, M., & Errando, M. 2009, APh, 31, 226 quadratic case (n = 2); they were included in the calculation of Mattingly, D. 2005, LRR 8 the limits on E by adding the statistical and the systematic Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, ApJ, 461, 396 QG Osterman, M. A., Miller, H. R., Campbell, A. M., et al. 2006, AJ, 132, 873 errors in quadrature. Piron, F., Djannati-Atai, A., Punch, M., et al. 2001, A&A, 374, 895 Prandini, E., Dorner, D., Mankuzhiyil, N., et al. 2009, arXiv: 0907.0157 REFERENCES Rector, T. A., Gabuzda, D. C., & Stocke, J. T. 2003, AJ, 125, 1060 Reimer, A., Costamante, L., Madejski, G., Reimer, O., & Dorner, D. 2008, Abdo, A. A., Ackermann, M., Ajello, M., et al. 2009a, ApJ, 707, 1310 ApJ, 682, 775 Abdo, A. A., Ackermann, M., Ajello, M., et al. 2009b, ApJS, 183, 46 Sanchez, D. A., & Deil, C. 2013, arXiv:1307.4534 Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010a, ApJ, 708, 1310 Sanchez, D. A., Fegan, S., & Giebels, B. 2013, A&A, 554, A75 Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010b, ApJ, 722, 520 Scargle, J. D. 1998, ApJ, 504, 405 Abramowski, A., Acero, F., Aharonian, F., et al. 2011, APh, 34, 738 Sol, H., Zech, A., Boisson, C., et al. 2013, APh, 43, 215 Abramowski, A., Acero, F., Akhperjanian, A. G., et al. 2013a, A&A, Treves, A., Falomo, R., & Uslenghi, M. 2007, A&A, 473, L17 552, A118 Urry, C. M., & Padovani, P. 1995, PASP, 107, 803 Abramowski, A., Acero, F., Aharonian, F., et al. 2013b, A&A, 550, A4 Vasileiou, V., Jacholkowska, A., Piron, F., et al. 2013, PhRvD, 87, 122001 Abramowski, A., Acero, F., Aharonian, F., et al. 2014, A&A, submitted Vaughan, S., Edelson, R., Warwick, R. S., & Uttley, P. 2003, MNRAS, Ackermann, M., Ajello, M., Albert, A., et al. 2012a, ApJS, 203, 4 345, 1271

54 In particular the errors on the peak positions constitute the most important part of the uncertainty on the template light curve contributing to the likelihood ﬁt—see previous works, e.g., Abramowski et al. (2011). Therefore, the covariance matrix of the ﬁt of the template was studied in detail; the peak positions were varied by values of ±1σ extracted from the covariance matrix. This study led to an increase in overall systematics of the order of 20% for τ1 and 40% for τ2, and a decrease of maximum 7% and 2% of limits on EQG,1 and EQG,2 respectively.

128 A.5. Seyfert 2 galaxies in the GeV band: jets and starburst A&A 524, A72 (2010) Astronomy DOI: 10.1051/0004-6361/201015644 & c ESO 2010 Astrophysics

Seyfert 2 galaxies in the GeV band: jets and starburst

J.-P. Lenain, C. Ricci, M. Türler, D. Dorner, and R. Walter

ISDC Data Centre for Astrophysics, Observatoire de Genève, Université de Genève, Chemin d’Ecogia 16, 1290 Versoix, Switzerland e-mail: [email protected]

Received 26 August 2010 / Accepted 23 September 2010

ABSTRACT

Context. The Fermi/LAT collaboration recently reported the detection of starburt galaxies in the high energy γ-ray domain, as well as radio-loud narrow-line Seyfert 1 objects. Aims. Motivated by the presence of sources close to the location of composite starburst/Seyfert 2 galaxies in the first year Fermi/LAT catalogue, we aim at studying high energy γ-ray emission from such objects, and at disentangling the emission of starburst and Seyfert activity. Methods. We analysed 1.6 years of Fermi/LAT data from NGC 1068 and NGC 4945, which count among the brightest Seyfert 2 galaxies. We search for potential variability of the high energy signal, and derive a spectrum of these sources. We also analyse public INTEGRAL IBIS/ISGRI data over the last seven years to derive their hard X-ray spectrum. Results. Wefindanexcessofhighenergyγ-rays of 8.3σ and 9.2σ for 1FGL J0242.7+0007 and 1FGL J1305.4 4928, which are found to be consistent with the position of the Seyfert 2 galaxies NGC 1068 and NGC 4945, respectively. The− energy spectrum of the sources can be described by a power law with a photon index of Γ=2.31 0.13andafluxofF100 MeV 100 GeV = 12 2 1 ± − (8.60 2.27) 10− erg cm− s− for NGC 1068, while for NGC 4945, we obtain a photon index of Γ=2.31 0.10andafluxof ± × 11 2 1 ± F100 MeV 100 GeV = (1.58 0.32) 10− erg cm− s− . For both sources, we detect no significant variability nor any indication of a curvature− of the spectrum.± While× the high energy emission of NGC 4945 is consistent with starburst activity, that of NGC 1068 is an order of magnitude above expectations, suggesting dominant emission from the active nucleus. We show that a leptonic scenario can account for the multi-wavelength spectral energy distribution of NGC 1068. Conclusions. High energy γ-ray emission is revealed for the first time in a Seyfert 2 galaxy. If this result is confirmed in other objects, new perspectives would be opened up into the GeV band, with the discovery of a new class of high energy γ-ray emitters. Key words. gamma rays: galaxies – galaxies: Seyfert – galaxies: individual: NGC 1068 – galaxies: individual: NGC 4945 – radiation mechanisms: non-thermal

1. Introduction Caproni et al. (2006) studied the morphology of the inner warped disc, likely due to the Bardeen-Petterson effect, using Somewhat unexpectedly, the Fermi/LAT collaboration reported VLBA observations. Cotton et al. (2008) showed that the emis- the discovery of four radio-loud narrow-line Seyfert 1 galaxies sion at 43 GHz from the so-called S1 central radio component is (Abdo et al. 2009a,b), suggesting that these objects, previously more likely dominated by thermal emission from the hot inner undetected at high energies, could constitute a new class of high region of the obscuring torus, although Gallimore et al. (2004) energy emitters. Extending this idea to Seyfert 2 galaxies, we pointed out that the North-East radio component could be dom- show here that two such active galactic nuclei (AGN) among the inated by synchrotron emission. This is corroborated by the re- closest and brightest in the X-ray sky, NGC 1068 and NGC 4945, sults of Hönig et al. (2008), who argued that the radio emission are detected at high energies using Fermi/LAT data. We investi- from the core could be dominated by synchrotron or free-free gate whether this emission is dominated by AGN or starburst emission, while near-IR data are dominated by the thermal emis- activity. sion from the dusty torus. Thanks to near-IR interferometric data NGC 1068 is an archetypal Seyfert 2 galaxy, located at z = using the MIDI instrument at the VLT, Raban et al. (2009) found 0.003786, i.e. 14.4 Mpc away, and harbours a hidden Seyfert 1 the torus to be composed of two thermal components at 300 K core. It was used by Antonucci & Miller (1985) to propose and 800 K. the AGN unification. Given the proximity of this spiral galaxy, its extension is well observable in visible light, and it is the clos- In the high energy domain, Chandra observations of the core est, as well as one of the brightest, Seyfert 2 galaxy. This source of NGC 1068 (Ogle et al. 2003) showed that the X-ray emis- exhibits both AGN and starburst activities in its central region sion is due to photoionisation in the extended, clumpy narrow (e.g. Lester et al. 1987; Jaffe et al. 2004). A dusty torus lies in line region. Matt et al. (2004) showed that the neutral reflector is its central part, dominating the soft X-ray emission by reflection Compton-thick, using XMM-Newton observations. from the central nucleus. A circumnuclear starburst region is lo- Motivated by the presence of a Fermi/LAT source, cated at 1 kpc from the core, dominating the infrared emission 1FGL J0242.7+0007, in the region of NGC 1068 in the of the broadband∼ spectral energy distribution (SED) (Thronson 11-months Fermi/LAT catalogue (1FGL, Abdo et al. 2010a), et al. 1989). with no proposed counterpart in radio nor in γ-rays, we analyse Article published by EDP Sciences Page 1 of 7

129 A. Selected publications A&A 524, A72 (2010)

Fig. 1. Left: TS map of NGC 1068 between 100 MeV and 100 GeV. The green ellipses show the 68% and the 95% position errors from the 1FGL catalogue, the cyan and magenta circles show respectively the position error (at 68% and 95% CL) for the full data set with all the events accounted for, and for front events only. The white contours are taken from an optical image from the Digital Sky Survey, showing the extent of the Seyfert galaxy. The red boxed points denote the position of the two quasars nearby NGC 1068 (see text in Sect. 4 for more details). Right:same as left panel, for NGC 4945. For clarity, we only present here the position error circle for all events. here 1.6 years of data from the Fermi/LAT instrument, in order Using the gtlike tool and assuming a power-law shape for to better constrain the origin and properties of the γ-ray emission the source spectrum, the Test Statistic (TS, Mattox et al. 1996) in this region. of the likelihood analysis is 68.6, corresponding approximately NGC 4945 is also a Seyfert 2 galaxy at z = 0.001908 ex- to a 8.3σ source detection in the 100 MeV 100 GeV range. hibiting starburst activity in its central region (Iwasawa et al. The corresponding TS map is shown in Fig. 1.− The best-fit loca- h m s 1993; Moorwood & Oliva 1994). Its emission extends up to soft tion of the source using the gtfindsrc tool is αJ2000 = 2 42 46 , γ-rays (Petry et al. 2009), as observed by INTEGRAL/SPI. It is δJ2000 = 0◦2′14′′ with an error circle radius of 6.′1 (68% con- one of the brightest hard X-ray AGN (see Itoh et al. 2008,and fidence level, CL), and is fully compatible with∼ the position re- references therein, and Ricci et al., in prep.). NGC 4945 was ported in the 1FGL catalogue. The maximum photon energy de- found to be a Compton thick AGN, based on GINGA (Iwasawa tected from the source is 20.0 GeV, located at 0.′48 from the et al. 1993)andINTEGRAL observations (Beckmann et al. position of NGC 1068. 2009). This source was already reported as a high-energy γ-ray Given the energy dependence of the point spread function emitter by the Fermi/LAT collaboration in the 11-months cat- (PSF) of Fermi/LAT, we performed a second analysis, using only alogue (Abdo et al. 2010a), although the authors did not con- front events for which the PSF is narrower2. Front events are clude whether this high energy emission is due to starburst or those converted in the top layers of the tracker of the LAT in- AGN activity. strument (see Atwood et al. 2009, for more details). In this latter We present here a detailed analysis of 1.6 years of analysis, the TS of the source is 42, still sufficient to derive a Fermi/LAT data of NGC 4945 to compare the results to those position of the Fermi/LAT excess. The best-fit position of the h m s of NGC 1068. source is then α = 2 42 49 , δJ = 0 0 30 , with an er- J2000 2000 − ◦ ′ ′′ ror circle radius of 3.′4 (68% CL), which is only 2.′1awayfrom the nominal position∼ of NGC 1068. This latter result on the posi- 2. Fermi/LAT data analysis tion is only marginally compatible with the position reported in We present in the following our analysis of Fermi/LAT data on the 1FGL catalogue. the sources 1FGL J0242.7+0007 and 1FGL J1305.4 4928 re- Given the angular distance between the Fermi source ported in the 1FGL catalogue. We will show that these− sources 1FGL J0242.7+0007 and NGC 1068, and its optical extension can be associated to the Seyfert 2 objects NGC 1068 and of 6.′5, we propose that this Fermi source is actually associated ∼ NGC 4945, respectively. with the Seyfert 2 galaxy NGC 1068. We analysed 1.6 yr of Fermi/LAT data, spanning from All the sources reported in the Fermi/LAT 11-months cat- August 4, 2008 to∼ March 15, 2010, from a region of interest alogue (Abdo et al. 2010a) within a radius of 15◦ around of 10◦ in radius around NGC 1068, using the publicly available NGC 1068 were included in the likelihood analysis, and mod- Science Tools1, and we followed the unbinned likelihood analy- elled with power-law spectra. We first identify the sources sis scheme presented in Atwood et al. (2009). We always used not contributing significantly to the likelihood and remove the so-called “diffuse class” events, which are the events de- them from the model. The spectral parameters of the remain- tected by the Fermi/LAT with the highest probability to be γ-ray ing sources are left freely varying. For NGC 1068, we obtain 12 2 1 photons, and the P6V3 instrument response. F100 MeV 100 GeV = (8.60 2.27) 10− erg cm− s− and − ± × 1 http://fermi.gsfc.nasa.gov/ssc/data/analysis/ 2 See http://www-glast.slac.stanford.edu/software/IS/ software/ glast_lat_performance.htm

Page 2 of 7

130 A.5. Seyfert 2 galaxies in the GeV band: jets and starburst J.-P. Lenain et al.: Seyfert 2 galaxies in the GeV band: jets and starburst

to a 9.2σ detection. Assuming a power-law shape on the source energy spectrum, a photon index of Γ=2.31 0.10 and a ﬂux of 11 ± 2 1 F100 MeV 100 GeV = (1.58 0.32) 10− erg cm− s− are found. − ± × The highest energy photon detected is 20.7 GeV, located at 0.′22 from NGC 4945. As for NGC 1068, the use of a broken power- law or a log parabola did not improve the likelihood, and no signiﬁcant variability was found in the data (see Fig. 2), which are statistically consistent with a constant for both sources. The analysis of the whole data set gives a position of αJ2000 = 13h05m33s, δ = 49 26 44 with an error circle radius J2000 − ◦ ′ ′′ of 3.′2 (68% CL), only 1.′6 away from the nominal position of NGC 4945, while the analysis for the front events only results in a position of α = 13h05m34s, δ = 49 26 49 with an J2000 J2000 − ◦ ′ ′′ error circle radius of 3.′3 (68% CL). The source position is fully compatible with the results reported in the 1FGL catalogue. It should also be noted that for both sources, the spectral parameters found are fully compatible with the ones reported in the 1FGL catalogue.

3. INTEGRAL data analysis For the extraction of the IBIS/ISGRI spectra, we used all the public data obtained by INTEGRAL as of May 2010, for a total of 564 and 865 pointings (ScWs) on NGC 1068 and NGC 4945, respectively. The typical exposure of each pointing is (1 3) 103 s, and only ScWs with an effective exposure longer− than× 200 s were kept, spanning times between December 30, 2002 (revolution 26) and April 7, 2009 (revolution 791). The total exposure is of 622 ks for NGC 1068, and of 927 ks for NGC 4945. In the large 17 80 keV band, NGC 1068 and NGC 4945 were detected with a− significance of 14.0σ and 138.8σ, respectively. IBIS/ISGRI has a large field of view of 29◦ 29◦ with a spatial resolution of 12 arcmin. Note that because× of the nature of coded mask imaging the whole sky image taken by the instrument has to be considered in the analysis, because all sources in the field of view contribute to the signal (Caroli et al. 1987). The ISGRI data were reduced using the INTEGRAL Offline Scientific Analysis software3 version 9.0, publicly released by the INTEGRAL Science Data Centre (Courvoisier et al. 2003). The analysis of the IBIS/ISGRI data is based on a cross- correlation procedure between the recorded image on the detector plane and a decoding array derived from the mask pattern. Fig. 2. Top: light curve of NGC 1068 in the 100 MeV 100 GeV energy − We created mosaic images of all pointings in 10 energy bins, band. The time bins are 60 days wide, and the arrows represent 95% CL and extracted the spectra using mosaic_spec. upper limits. Bottom: same as top for NGC 4945. In both sources, no We used the latest detector redistribution matrix files (RMF) significant variability is visible. and calculated the ancillary response functions (ARFs) with a weighted average of the 9 available ARFs, based on the number Γ=2.31 0.13. We also tried to fit the data with a broken power- of ScWs within the validity time of a particular ARF. law or a log± parabola, but this did not improve the likelihood. We have also checked the Swift/BAT (Barthelmy et al. 2005) We performed two other analyses, focussing on the spectra of the two sources. The 18 months Swift/BAT spectra 1 100 GeV band, for all events and for front events only. This have been extracted from the BAT archive (Segreto et al. 2010) energy− band benefits from a better PSF compared to lower en- served by the HEAVENS source results archive (Walter et al., ergy events, despite a much smaller count rate due to the soft in prep.). The ISGRI and BAT spectra are in good agreement for spectrum of the source. We obtain consistent results to those re- both sources. ported above, including comparable position error on the source. A simple power-law fit to the combined INTEGRAL/ISGRI We also investigated the potential variability, performing a and Swift/BAT data of NGC 1068 using Xspec results in likelihood analysis in different time intervals, for a time bin of χ2/d.o.f. = 10.21/11. The resulting photon index is Γ= +0.21 +0.15 60 days, letting only the normalisation of the source free to vary. 2.08 0.19, and the flux density is F17 80 keV = 1.86 1.23 11− 2 1 − − × As shown in Fig. 2, no significant variability was detected. 10− erg cm− s− (both 90% CL). We followed the same procedure for the analysis of The hard X-ray spectrum for the combined ISGRI and Fermi/LAT data in the region of NGC 4945, for the source BAT data of NGC 4945 is not well reproduced by a simple 1FGL J1305.4 4928. The likelihood analysis on NGC 4945 results in a TS of− 85.3 in the 100 MeV 100 GeV band, equivalent 3 http://www.isdc.unige.ch/integral/ − Page 3 of 7

131 A. Selected publications A&A 524, A72 (2010)

Table 1. Comparison of the starburst properties of NGC 1068 with NGC 4945, NGC 253, M 82, the LMC and the Milky Way.

a a b c Source d Γ Lγ RSN Mgas LIR L5 GHz Lγ/L5 GHz 39 1 1 9 44 1 38 1 (Mpc) (10 erg s− )(yr− )(10M )(10erg s− )(10erg s− ) NGC 1068 14.4 2.31 0.13 170 32 0.20 0.08 4.4⊙ 2.7 16.6 102 NGC 4945 3.6 2.30 ± 0.10 19.5 ± 2.9 0.1–0.5± 4.21.4 2.289 NGC 253 3.9 1.95± 0.47.2 ±4.70.2 0.12.5 0.61.0 2.0 36 M82 3.6 2.2 ±0.213.0± 5.00.2 ± 0.12.5 ± 0.71.2 3.0 43 LMC 0.049± – 0.041 ±0.007 0.005 ± 0.002 0.67 ± 0.08 0.016 0.017 24 Milky Way – – 3.2 ± 1.60.02 ± 0.01 6.5 ± 2.0 0.5 0.36 89 ± ± ± Notes. (a) Photon indices and luminosities in the γ-ray band from Abdo et al. (2010c) for M 82, NGC 253, the LMC and the Milky Way. The γ-ray luminosities are given in the 100 MeV 5 GeV band, and the photon indices Γ are given above 200 MeV for an easier comparison with the results from Abdo et al. (2010c). − (b) SN rate estimates from Wilson & Ulvestad (1982); Blietz et al. (1994); Mannucci et al. (2003) for NGC 1068, from Lenc & Tingay (2009)for NGC 4945, and from Abdo et al. (2010c, and references therein) for M 82, NGC 253, the LMC and the Milky Way. (c) Gas mass estimates from Sage et al. (1990) for NGC 1068, from Weiß et al. (2008) for NGC 4945, and from Abdo et al. (2010c, and references therein) for M 82, NGC 253, the LMC and the Milky Way. power-law, and a more complex model is required. A power- law with exponential cut-oﬀ yields χ2/d.o.f. = 89.6/9, which is not satisfactory, even though the data clearly show a curvature. Instead, an absorbed power-law yields χ2/d.o.f. = 54.77/9, with a photon index of Γ=2.03 0.04, and an intrinsic absorption ± 24 2 2 column NH = (6.66 0.45) 10 cm− . Even though the χ is not entirely satisfactory,± we× will use the latter spectral model to build the broadband SED of NGC 4945, because it provides a better description of the data. A detailed analysis of the hard X-ray spectral shape is beyond the scope of this paper.

4. Discussion It shall be noted that two quasars are present in the field of view of NGC 1068. The quasar SDSS J024230.65-000029.6 (z = 2.51) lies 2.′4 from NGC 1068, and another quasar, SDSS J024250.98-000031.6∼ (z = 2.18, a.k.a. QSO 0240 0012), is − 2.′6 away from NGC 1068. However, given their high redshift, it∼ is unlikely that one of these objects could be at the origin of the high energy emission, although this can not be completely Fig. 3. Relationship between SN rate, total gas mass and γ-ray lumi- excluded. Indeed according to Abdo et al. (2010b), the highest nosity of NGC 1068, NGC 4945, NGC 253, M 82, the LMC and the redshift source detected by the Fermi/LAT is a flat spectrum ra- Milky Way. dio quasar with a redshift of z = 3.10. Given the similarities of the high energy spectra of 1FGL J0242.7+0007 and NGC 4945, it appears however more likely that 1FGL J0242.7+0007 is as- NGC 1068 and NGC 4945 to the ones of NGC 253, M 82, the sociated with NGC 1068. Large Magellanic Cloud (LMC) and the Milky Way, as well as An obvious test to rule out the starburst origin of the high- their infrared and radio luminosities (see Table 1). These objects energy emission would be to detect significant γ-ray variations are the only extragalactic sources which are not AGN known from one of these objects, as the emission arising from starburst to emit high energy γ-rays. Models attributing the γ-rays to activity is expected to be steady. Due to the lack of statistics, cosmic-ray processes, as expected in a starburst galaxy, depend no conclusions can be drawn about the variability of NGC 1068 on the product MgasRSN. Adding NGC 1068 and NGC 4945 to or NGC 4945 from their light curves (see Fig. 2). this picture (see Fig. 2 in Abdo et al. 2010c), and assuming that Another way to disentangle the starburst or AGN origin of their γ-ray emission is entirely due to starburst activity, con- the high-energy emission from these galaxies is to compare their firms the general trend, but we only find a correlation coeffi- γ-ray luminosity with those of the famous starburst galaxies cient of 0.44 for a linear relationship between MgasRSN and Lγ NGC 253 and M 82, which are also detected in the very high (see Fig. 3), corresponding to a null hypothesis probability as energy domain by H.E.S.S. (Acero et al. 2009)andVERITAS high as 38%, although the number of considered sources is ob- (Acciari et al. 2009), respectively. Both have a luminosity in the viously too small to definitely conclude. 40 1 100 MeV 5 GeV band of the order of 10 erg s− , as detected The γ-ray luminosity and the SN rate of NGC 4945 are with Fermi− (Abdo et al. 2010c). Computing≈ the luminosities of fully consistent with those of NGC 253 and M 82, hence even NGC 1068 and NGC 4945 in the same energy band, for com- though this object is a composite starburst/AGN, its high en- 41 1 40 1 parison, we obtain 1.7 10 erg s− and 2.0 10 erg s− , ergy emission detected using Fermi could be explained only in respectively. × × terms of its starburst activity. Concerning NGC 1068, the situa- Following Abdo et al. (2010c), we compare the supernova tion is more complex. Its SN rate is comparable to those of M 82 rate RSN, the total gas mass Mgas and the γ-ray luminosities of and NGC 253, but its radio and γ-ray luminosities are higher

Page 4 of 7

132 A.5. Seyfert 2 galaxies in the GeV band: jets and starburst J.-P. Lenain et al.: Seyfert 2 galaxies in the GeV band: jets and starburst

Table 2. Model parameters for NGC 1068 and NGC 4945.

a 1 3 Source Component δb B (G) rb (cm) T (K) τLnuc (erg s− ) R (cm) K (cm− ) n1 n2 γbreak γmax 4 19 42 20 4 6 NGC 1068 1 1.2 10− 2.0 10 130–520 1.5 10 2.2 10 12.5 2.2 3.3 10 10 NGC 1068 2 1.2 1 10×15 104 3.0 × 1041 5 ×1014 106 2.0 – – 300 NGC 4945 2 1.2 1 2.6 1015 104 5.0 × 1041 5 × 1015 106 1.5 – – 160 × × × Notes. (a) The component 1 refers to the model for the large outﬂow, while component 2 points to the model for the Seyfert emission in hard X-rays from the accretion disc.

by a factor 10. This would suggest that its high energy γ-ray emission is∼ more likely dominated by the central AGN activity. Indeed, removing NGC 1068 from the sample of sources, the correlation coefficient for the relationship between MgasRSN and the γ-ray luminosity becomes 0.95, with a null hypothesis probability of only 1%. This also tends to prove the peculiar role of NGC 1068 in∼ this study. This is also strengthened by the fact that radio maps of NGC 1068 clearly show a structured jet, on parsec- and kiloparsec-scales, modelled by the outflow from the central AGN (see e.g. Gallimore et al. 2004, 2006). On the contrary, the radio morphology of NGC 4945 shows an extended emission consistent with the optical morphology tracing the edge-on galaxy (see e.g. Jones & McAdam 1992), indicating a starburst emission. Assuming that the high energy emission of NGC 1068 is indeed due to the AGN activity, and taking a scenario similar to what was proposed in Lenain et al. (2008) where the jet is misaligned with respect to the line of sight, we consider here the Fig. 4. Spectral energy distribution of NGC 1068, including the Fermi/LAT spectrum. The black and red points are archival data from possibility that the outflow in NGC 1068 could be a high energy the NED, the red ones denote data taken from the central region of emitter. NGC 1068. For clarity, we only show the INTEGRAL IBIS/ISGRI data A large, mildly-relativistic zone of the wind-like outflow, in blue in the hard X-rays. The EIC model for the outflow is shown at a few tenth of parsecs from the core, could emit high-energy in blue, and the corresponding SSC emission is shown in thin red and γ-rays through external inverse Compton process (EIC) (see e.g. magenta lines for first and second order components, respectively. The Begelman & Sikora 1987). At such distances, the infrared pho- thick red line shows the sum of the different emission components from ton energy density is still high enough to ensure a significant the large outflow. The EIC component from the accretion disc is shown emission while being not too important to prevent high optical in green. opacity from pair production. At about 100 parsecs from the core, the magnetic field strength is also expected to be low, of 4 be found in Katarzynski´ et al. (2001), Lenain et al. (2008), and the order of 10− G according to the estimate for the equiparti- tion magnetic field of the radio component A, located at a pro- Lenain (2009). We use γmin = 1 in the following. jected distance of 350 pc from the core of NGC 1068 (Pedlar In Fig. 4, we present such a model with synchrotron emission et al. 1983). No significant∼ short-term variability is expected responsible for the radio emission, while the Fermi/LAT data are from such a large emitting zone. interpreted as EIC emission with the infrared emission providing The stationary emission of a blob of plasma is modelled here the seed photons, from a multi-temperature blackbody. We also through leptonic processes. The macrophysics of the blob is de- show the contribution from the SSC process, which is negligible compared to the EIC emission at the highest energies. We note scribed by its bulk Doppler factor δb,radiusrb, and strength of the tangled magnetic field B filling it up. The energy distribution that the contribution of second order SSC is negligible in the of the leptons is described by a broken power-law with the den- case of our interpretation of the SED of NGC 1068. The model parameters are summarised in Table 2. sity normalisation K, the indices n1 and n2, and the Lorentz factors of the individual particles γmin, γbreak and γmax, as follows: The hard X-ray spectrum observed with INTEGRAL does not seem to fit in this picture and we propose that this emis- n1 Kγ− if γmin γ γbreak 3 sion originates from EIC processes on another population of lep- N (γ) = n n [cm ]. (1) e 2 1 n2 − tons, within hot plasma located in the vicinity of the accretion Kγbreak− γ− if γbreak γ γmax disc. The seed photons would then originate from the accretion These leptons emit through synchrotron self-Compton (SSC, disc itself as is usually invoked to explain the X-ray emission see e.g. Ginzburg & Syrovatskii 1965) and EIC processes. The of Seyfert galaxies. Evidence for an accretion disc comes from latter is described by the fraction of the core luminosity τLnuc re- the fact that the soft X-ray spectrum of NGC 1068, e.g. as ob- processed by the emitting source (see Inoue & Takahara 1996), served with XMM-Newton, is dominated by thermal reflection the temperature T of the seed radiation field described as a black- emission (see e.g. Kinkhabwala et al. 2002; Pounds & Vaughan body, and the distance R between the external radiation field and 2006). We checked that the synchrotron and the SSC emission the re-emitting blob. Second order SSC emission (Rees 1967)is from this component is negligible compared to the one from the also accounted for in this model. More details on the model can large outflow described above.

Page 5 of 7

133 A. Selected publications A&A 524, A72 (2010)

5. Conclusion We reported the detection of the Seyfert 2 galaxy NGC 1068 in the high energy (100 MeV 100 GeV) domain with the Fermi/LAT, and compared the analysis− of this source with the one of NGC 4945 obtained by extending the observing period to 1.6 years. The object 1FGL J0242.7+0007 was reported as a γ-ray source without association in the 1FGL catalogue. We can now associate it quite firmly to NGC 1068 with an overall detection significance of 8.4σ. The data do not show any significant variability over the whole period. The high energy γ-ray spectrum of NGC 1068 is consistent with a power-law of index Γ=2.31 0.13andafluxdensityof 12 2 1 ± (8.60 2.27) 10− erg cm− s− . Compared to the γ-ray luminosities± of M× 82 and NGC 253, whose high-energy emission is dominated by starburst activity, we find a too high γ-ray luminosity of NGC 1068 to be explained only by starburst activity. Fig. 5. Spectral energy distribution of NGC 4945, including the We thus propose a leptonic scenario to interpret the high energy Fermi/LAT spectrum (black points). For clarity, we only show the emission in terms of external inverse Compton process from an INTEGRAL IBIS/ISGRI data in blue in the hard X-rays. The model for outflowing relativistic wind launched by the central AGN. the EIC component from the accretion disc is shown in green. We show NGC 4945 is detected at a 9.2σ level in Fermi/LAT data, in red the data of NGC 253 as taken from the NED, with the Fermi/LAT and its γ-ray spectrum is best described by a power-law spectrum published in Abdo et al. (2010c) as well as the H.E.S.S. flux measurement from Acero et al. (2009), for comparison. The two ob- of index Γ=2.31 0.10andafluxdensityof(1.58 0.32) 10 11 erg cm 2±s 1. This object has a very similar multi-± jects have clearly very similar SEDs. The luminosity axis on the right is × − − − given for NGC 4945. wavelength SED compared to NGC 253, and based on its radio, infrared and γ-ray luminosities, its high energy γ-ray emission is most likely due to starburst activity. If high energy γ-ray emission due to AGN activity is confirmed in other Seyfert 2 galaxies, this would mark the discov- It should be noted that imaging atmospheric Cerenkovˇ tele- ery of yet a new class of high-energy γ-ray emitters. New data scopes have the ability to strongly constrain the highest energy from Fermi/LAT in the coming years will be extremely valuable part of the particle energy distribution. If the maximum energy in this regard. 6 of the leptons is γmax > 5 10 , a significant signal could be detected from NGC 1068∼ with× H.E.S.S., VERITAS, MAGIC, or the future Cerenkovˇ Telescope Array (CTA) (CTA consortium 2010). Acknowledgements. We thank our colleague Dr. Christian Farnier for valuable discussions. This research has made use of NASA’s Astrophysics Data System An alternative solution from our model would consist in in- (ADS), of the SIMBAD database, operated at CDS, Strasbourg, France, and of terpreting the hard X-ray emission from INTEGRAL and the high the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet energy γ-ray emission from Fermi/LAT as originating from the Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. J.-P. L. would like to dedi- same spectral component. In this case, the overall high energy cate this work to the memory of his missed friend Jean-Claude Rouffignat. emission would be due to EIC process from the same lepton population. The spectral shapes of both INTEGRAL and Fermi data then strongly constrain the parameters on the lepton energy distribution, for which we obtain n1 = 2.0, n2 = 3.6, γbreak = 700 3 and K = 450 cm− . However, such a particle energy distribu- References tion can not account correctly for the SED observed in the radio domain. Abdo, A. A., Ackermann, M., Ajello, M., et al. (Fermi/LAT collaboration) In Fig. 5, we show the SED of NGC 4945, including the 2009a, ApJ, 699, 976 Abdo, A. A., Ackermann, M., Ajello, M., et al. (Fermi/LAT collaboration) Fermi/LAT and the INTEGRAL spectra we derived in Sects. 2 2009b, ApJ, 707, L142 and 3, respectively. For comparison, we also show the data of the Abdo, A. A., Ackermann, M., Ajello, M., et al. (Fermi/LAT collaboration) starburst galaxy NGC 253, as taken from the NED, along with 2010a, ApJS, 188, 405 the Fermi/LAT spectrum as published in Abdo et al. (2010c). Abdo, A. A., Ackermann, M., Ajello, M., et al. (Fermi/LAT collaboration) 2010b, ApJ, 715, 429 Clearly, the two objects have very similar broadband SEDs, Abdo, A. A., Ackermann, M., Ajello, M., et al. (Fermi/LAT collaboration) which match almost perfectly, strengthening the idea that the 2010c, ApJ, 709, L152 high energy γ-ray emission from NGC 4945 could be well ex- Acciari, V. A., Aliu, E., Arlen, T., et al. (VERITAS collaboration) 2009, Nature, plained only in terms of starburst activity. NGC 4945 should 462, 770 then be detectable in the very high energy domain, and espe- Acero, F., Aharonian, F., Akhperjanian, A. G., et al. (H.E.S.S. collaboration) 2009, Science, 326, 1080 cially with H.E.S.S. given its coordinates, at a flux level similar Antonucci, R. R. J., & Miller, J. S. 1985, ApJ, 297, 621 to NGC 253. However, one difference between NGC 4945 and Atwood, W. B., Abdo, A. A., Ackermann, M., et al. (Fermi/LAT collaboration) NGC 253 resides in their hard X-ray emission. While NGC 253 2009, ApJ, 697, 1071 is not thought to harbour a central AGN, NGC 4945 exhibits Barthelmy, S. D., Barbier, L. M., Cummings, J. R., et al. 2005, Space Sci. Rev., 120, 143 a characteristic Seyfert emission, which is modelled as above Beckmann, V., Soldi, S., Ricci, C., et al. 2009, A&A, 505, 417 for NGC 1068. The corresponding parameters are also included Begelman, M. C., & Sikora, M. 1987, ApJ, 322, 650 in Table 2. Blietz, M., Cameron, M., Drapatz, S., et al. 1994, ApJ, 421, 92

Page 6 of 7

134 A.5. Seyfert 2 galaxies in the GeV band: jets and starburst J.-P. Lenain et al.: Seyfert 2 galaxies in the GeV band: jets and starburst

Caproni, A., Abraham, Z., & Mosquera Cuesta, H. J. 2006, ApJ, 638, 120 Lester, D. F., Joy, M., Harvey, P. M., Ellis, Jr., H. B., & Parmar, P. S. 1987, ApJ, Caroli, E., Stephen, J. B., Di Cocco, G., Natalucci, L., & Spizzichino, A. 1987, 321, 755 Space Sci. Rev., 45, 349 Mannucci, F., Maiolino, R., Cresci, G., et al. 2003, A&A, 401, 519 Cotton, W. D., Jaffe, W., Perrin, G., & Woillez, J. 2008, A&A, 477, 517 Matt, G., Bianchi, S., Guainazzi, M., & Molendi, S. 2004, A&A, 414, 155 Courvoisier, T., Walter, R., Beckmann, V., et al. 2003, A&A, 411, L53 Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, ApJ, 461, 396 CTA consortium 2010 [arXiv:1008.3703] Moorwood, A. F. M., & Oliva, E. 1994, ApJ, 429, 602 Gallimore, J. F., Baum, S. A., & O’Dea, C. P. 2004, ApJ, 613, 794 Ogle, P. M., Brookings, T., Canizares, C. R., Lee, J. C., & Marshall, H. L. 2003, Gallimore, J. F., Axon, D. J., O’Dea, C. P., Baum, S. A., & Pedlar, A. 2006, AJ, A&A, 402, 849 132, 546 Pedlar, A., Booler, R. V., Spencer, R. E., & Stewart, O. J. 1983, MNRAS, 202, Ginzburg, V. L., & Syrovatskii, S. I. 1965, ARA&A, 3, 297 647 Hönig, S. F., Prieto, M. A., & Beckert, T. 2008, A&A, 485, 33 Petry, D., Beckmann, V., Halloin, H., & Strong, A. 2009, A&A, 507, 549 Inoue, S., & Takahara, F. 1996, ApJ, 463, 555 Pounds, K., & Vaughan, S. 2006, MNRAS, 368, 707 Itoh, T., Done, C., Makishima, K., et al. 2008, PASJ, 60, 251 Raban, D., Jaffe, W., Röttgering, H., Meisenheimer, K., & Tristram, K. R. W. Iwasawa, K., Koyama, K., Awaki, H., et al. 1993, ApJ, 409, 155 2009, MNRAS, 394, 1325 Jaffe, W., Meisenheimer, K., Röttgering, H. J. A., et al. 2004, Nature, 429, 47 Rees, M. J. 1967, MNRAS, 137, 429 Jones, P. A., & McAdam, W. B. 1992, ApJS, 80, 137 Sage, L. J., Shore, S. N., & Solomon, P. M. 1990, ApJ, 351, 422 Katarzynski,´ K., Sol, H., & Kus, A. 2001, A&A, 367, 809 Segreto, A., Cusumano, G., Ferrigno, C., et al. 2010, A&A, 510, A47 Kinkhabwala, A., Sako, M., Behar, E., et al. 2002, ApJ, 575, 732 Thronson, Jr., H. A., Hereld, M., Majewski, S., et al. 1989, ApJ, 343, 158 Lenain, J.-P. 2009, Ph.D. Thesis, Observatoire de Paris Weiß, A., Kovács, A., Güsten, R., et al. 2008, A&A, 490, 77 Lenain, J.-P., Boisson, C., Sol, H., & Katarzynski,´ K. 2008, A&A, 478, 111 Wilson, A. S., & Ulvestad, J. S. 1982, ApJ, 263, 576 Lenc, E., & Tingay, S. J. 2009, AJ, 137, 537

Page 7 of 7

135

Curriculum Vitæ

137 B. Curriculum Vitæ

Jean-Philippe Lenain PhD

Personal data Date of birth November 30, 1983 Citizenship French Address 8 rue René Mas, F-91350 Grigny Family status Civil union, 2 children

e-mail [email protected] Web site http://lpnhe.in2p3.fr/hess/∼jlenain

Professional experience 2012 Permanent scientist (Chargé de Recherche), LPNHE, CNRS/IN2P3, Paris, France. Staﬀ resarcher within the H.E.S.S., and CTA projects and work on Fermi/LAT. 2011 Post-doctoral position, Landessternwarte, University of Heidelberg, Heidelberg, Germany. 2012 High energy astrophysics of active galactic nuclei, within the CTA, H.E.S.S., and ATOM projects and work on Fermi/LAT, under the supervision of Stefan WAGNER. 2009 Post-doctoral position, ISDC, University of Geneva, Geneva, Switzerland. 2011 Extragalactic high energy astrophysics, within the CTA consortium and work on Fermi/LAT, under the supervision of Roland WALTER. 2006 PhD in Astrophysics, LUTh, Observatoire de Paris-Meudon, Meudon, France, Upper class honors, 2009 European label. "Gamma radiation from Active Galactic Nuclei observed at Very High Energies with H.E.S.S.: multiwavelength studies and radiative modeling", within the H.E.S.S. collaboration, under the supervision of Catherine BOISSON and Hélène SOL, defended on October 1, 2009.

Collective interest activities Refereeing Ad-hoc referee for Science, ApJ and MNRAS.

Sept. 2016 Academic mentor of 3 PhD students at LPNHE.

July 2016 CTA group leader at LPNHE.

Oct. 2014 Elected representative at the Council of Laboratory of the LPNHE.

Oct. 2014–Sept. 2016 Convener of the Extragalactic working group of the H.E.S.S. collaboration.

Sept.–Dec. 2014 Editorial board for the LPNHE activity report for 2012–2014.

April 2013–Oct. 2014 Deputy convener of the Extragalactic working group of the H.E.S.S. collaboration.

2013 Responsible in France for the Monte Carlo simulations of the H.E.S.S. collaboration. Sept. 2013–Dec. 2014 Co-organiser of the seminars at LPNHE.

2012 Contact person for the astroparticle community for the France Grilles GIS.

Æ +33 6 51 67 50 41 Q [email protected] lpnhe.in2p3.fr/hess/ jlenain • •

138 Oct. 2012–April 2013 Co-convener of the former Active Galactic Nuclei working group of the H.E.S.S. collaboration.

2006 Elected representative of the young researchers at the Council of Laboratory of the LUTh, Paris 2009 Observatory. 2007 Elected representative of the graduate students at the Graduate School “Astronomy & Astrophysics” 2008 of the region Ile-de-France (ED 127).

Workshop organisation

June 2016 • SOC, Workshop MACROS 2016 (Multi-messenger Approaches to Cosmic Rays: Origins and Space Frontiers), June 20–22, 2016, Pennsylvania State University, USA.

Nov. 2015 • SOC, Days of France Grilles and Groupe Calcul SUCCES 2015 (“Rencontres Scientiﬁques des Utilisateurs de Calcul intensif, de Cloud Et de Stockage”), Nov. 5–6, 2015, IPGP, Paris, France.

Feb. 2015 • SOC/LOC, Exploratory workshop on GRAND (Giant Radio Array for Neutrino Detection), Feb. 9–11, 2015, LPNHE, Paris, France.

Nov. 2014 • SOC/LOC, Thematic day “Methods: Solutions and Challenges” of the ILP (Institut Lagrange de Paris) laboratory of excellence, Nov. 14, 2014, LPNHE, Paris, France.

June 2014 • SOC, Workshop “Astrophysique des très hautes énergies et perspectives pour CTA”, June 30 – July 1, 2014, OBSPM, Paris, France.

Nov. 2013 • SOC/LOC, Days of France Grilles and Groupe Calcul SUCCES 2013 (“Rencontres Scientiﬁques des Utilisateurs de Calcul intensif, de Cloud Et de Stockage”), Nov. 13–14, 2013, IPGP, Paris, France.

Nov. 2013 • SOC/LOC, Workshop MACROS 2013 (Multi-messenger Approaches to Cosmic Rays: Origins and Space Frontiers), Nov. 27–29, 2013, IAP, Paris, France. Supervision 2017 Co-supervision of Gabriel EMERY’s PhD thesis, with Julien BOLMONT, on the temporal properties of active galactic nuclei at very high energies with H.E.S.S. and prospects for CTA. Nov. 2015 Supervision of the post-doctoral work of Matteo CERRUTI on extragalactic transient sources with H.E.S.S. 2014 Co-supervision of Daniel KERSZBERG’s PhD thesis, with Pascal VINCENT, on the study of diﬀuse 2017 Galactic emission from electrons and positrons, and study of the performances of the second phase of the H.E.S.S. experiment. 2012 Partial co-supervision of Camille COUTURIER’s PhD thesis (supervised by A. JACHOLKOWSKA 2014 and J. BOLMONT) on searches for Lorentz Invariance Violation with H.E.S.S. and Fermi/LAT. 2013 Supervision of several B.Sc. (2) and M.Sc. (4) internships.

Languages French Mother tongue English Written, read and spoken ﬂuently TOEIC passed in 2005 (score: 835/990) Spanish Good knowledge and practice German Good knowledge and practice

Computer skills Operating systems GNU/Linux (Ubuntu, Debian, Redhat, Fedora, ...), Unix (Sun Solaris), Mac OS, Windows.

Æ +33 6 51 67 50 41 Q [email protected] lpnhe.in2p3.fr/hess/ jlenain • •

139 B. Curriculum Vitæ

Languages C, C++, Python, ROOT, Shell (BASH, CSH), Fortran, (X)HTML, PHP, MySQL, IDL/GDL, basic MPI/OpenMP and NVIDIA-CUDA. Skills Numerical computating on national facilities (CCIN2P3) and grids (EGEE-III, EGI), system administration (Linux, MySQL, apache, ...).

Teaching and internships June 2015 & June 2017 Lecturer at the STEP’UP (Earth, Environment, and Universe Physics Sciences) graduate school (ED 560), Paris, France. High energy astrophysics (10 h/year). 2006 Physics lecturer, University Pierre and Marie Curie, Paris, France. 2009 Practical and lab works. June–August 2008 Marie Curie Early Stage Training fellow, Astronomy group, Department of Physics, University of Durham, Durham, U.K. 3 months on the study of the multi-wavelength radiation of jets in blazars, in collaboration with Martin J. WARD and Paula M. CHADWICK. April–June 2006 Trainee for the 2nd year of M.Sc., LUTh, Observatoire de Paris-Meudon, Meudon, France. 3-month training period on the study of γ-rays in Active Galactic Nuclei within the framework of the H.E.S.S. project, under the supervision of Catherine BOISSON and Hélène SOL. March 2006 • Observational expedition, Savalou, Benin. Total solar eclipse on March 29, organized by the Master of Astrophysics, Observatoire de Paris-Meudon: observations and talks in high schools. May–July 2005 Trainee for the 1st year of M.Sc., High Energy Astrophysics Division, Harvard-Smithsonian Center for Astrophysics (CfA), Cambridge, MA, USA. 13-week training period on the study of the gas–radio plasma interaction in the galaxy NGC 4261 observed in X-ray with the Chandra space telescope, under the supervision of William R. FORMAN, Ralph KRAFT and Christine JONES. June–July 2004 Trainee for the 3rd year of B.Sc., IAS (Institut d’Astrophysique Spatiale), University of Paris XI, Orsay, France. 6-week training period on spectro-imaging of Uranus and its rings, under the supervision of François POULET.

Education 2004 Master of Science, University of Paris XI and Observatory of Paris, Paris, France, Upper second 2006 class honors. + Master 2 in Astronomy & Astrophysics, Astrophysics course. + Master 1 in Fundamental Physics. 2003 Magistère de Physique, University of Paris XI, Orsay, France, Upper second class honors. 2006 Fundamental physics. A “magistère” is a three-year program in which the student is required to maintain a B average at least. 2003 Bachelor of Science, University of Paris XI, Orsay, France, Lower second class honors. 2004 Fundamental physics, Specialization in astrophysics. 2001 Preparatory classes for engineering schools, Kléber secondary school, Strasbourg, France. 2003 Physics-Chemistry. 2001 Scientiﬁc “Baccalauréat”, Gymnase Jean Sturm secondary school, Strasbourg, France, A-. • French equivalent of A-Levels.

Æ +33 6 51 67 50 41 Q [email protected] lpnhe.in2p3.fr/hess/ jlenain • •

140 C

Publication list

141 C. Publication list PhD thesis

“Gamma radiation from Active Galactic Nuclei observed at Very High Energies with H.E.S.S.: multiwavelength studies and radiative modeling”, under the supervision of Catherine Boisson and Hélène Sol, LUTH (Laboratory Universe & Theories, Observatoire de Paris, CNRS UMR 8102), within the H.E.S.S. collaboration, European label, defended on 01/10/2009 with upper class honors. Complete text (in French) available online: http://tel.archives-ouvertes.fr/tel-00431288/fr/.

Articles

1. J.-P. Lenain. “FLaapLUC: A pipeline for the generation of prompt alerts on transient Fermi-LAT γ-ray sources.” Astronomy and Computing 22 (Jan. 2018), pp. 9–15. doi: 10.1016/j.ascom.2017.11.002. arXiv: 1709. 04065 [astro-ph.IM]

2. M. Zacharias, M. Böttcher, F. Jankowsky, et al. “Cloud Ablation by a Relativistic Jet and the Extended Flare in CTA 102 in 2016 and 2017.” ApJ 851, 72 (Dec. 2017), p. 72. doi: 10.3847/1538-4357/aa9bee. arXiv: 1711.06117 [astro-ph.HE]

3. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155-304 and PG 1553+113.” A&A 600, A89 (Apr. 2017), A89. doi: 10.1051/0004- 6361/201629427. arXiv: 1612.01843 [astro-ph.HE] (corresponding author)

4. P. Goldoni, S. Pita, C. Boisson, et al. “Optical-NIR spectroscopy of the puzzling γ-ray source 3FGL 1603.9- 4903/PMN J1603-4904 with X-Shooter.” A&A 586, L2 (Feb. 2016), p. L2. doi: 10 . 1051 / 0004 - 6361 / 201527582. arXiv: 1510.06234 [astro-ph.HE]

5. A. Abramowski, F. Aharonian, F. Ait Benkhali, et al. “The 2012 Flare of PG 1553+113 Seen with H.E.S.S. and Fermi-LAT.” ApJ 802, 65 (Mar. 2015), p. 65. doi: 10.1088/0004- 637X/802/1/65. arXiv: 1501.05087 [astro-ph.HE] (corresponding author)

6. S. Pita, P. Goldoni, C. Boisson, et al. “Spectroscopy of high-energy BL Lacertae objects with X-shooter on the VLT.” A&A 565, A12 (May 2014), A12. doi: 10.1051/0004- 6361/201323071. arXiv: 1311.3809 [astro-ph.HE]

7. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of very high energy γ-ray emission from the BL Lacertae object PKS 0301-243 with H.E.S.S..” A&A 559, A136 (Nov. 2013), A136. doi: 10.1051/0004- 6361/201321639. arXiv: 1309.6174 [astro-ph.HE] (corresponding author)

8. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “H.E.S.S. discovery of VHE γ-rays from the quasar PKS 1510-089.” A&A 554, A107 (June 2013), A107. doi: 10.1051/0004-6361/201321135. arXiv: 1304.8071 [astro-ph.HE] (major contribution to the Fermi-LAT data analysis)

9. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of TeV γ-ray emission from PKS 0447-439 and derivation of an upper limit on its redshift.” A&A 552, A118 (Apr. 2013), A118. doi: 10.1051/0004- 6361/201321108. arXiv: 1303.1628 [astro-ph.HE] (major contribution to the Fermi-LAT data analysis)

10. H. Sol, A. Zech, C. Boisson, et al. “Active Galactic Nuclei under the scrutiny of CTA.” Astroparticle Physics 43 (Mar. 2013), pp. 215–240. doi: 10.1016/j.astropartphys.2012.12.005

11. K. Bernlöhr, A. Barnacka, Y. Becherini, et al. “Monte Carlo design studies for the Cherenkov Telescope Array.” Astroparticle Physics 43 (Mar. 2013), pp. 171–188. doi: 10.1016/j.astropartphys.2012.10.002. arXiv: 1210.3503 [astro-ph.IM]

12. B. S. Acharya, M. Actis, T. Aghajani, et al. “Introducing the CTA concept.” Astroparticle Physics 43 (Mar. 2013), pp. 3–18. doi: 10.1016/j.astropartphys.2013.01.007

13. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of VHE γ-ray emission and multi-wavelength observations of the BL Lacertae object 1RXS J101015.9-311909.” A&A 542, A94 (June 2012), A94. doi: 10.1051/0004-6361/201218910 (corresponding author)

142 14. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “A multiwavelength view of the ﬂaring state of PKS 2155- 304 in 2006.” A&A 539, A149 (Mar. 2012), A149. doi: 10.1051/0004-6361/201117509. arXiv: 1201.4135 [astro-ph.HE] (major contribution to the modelling part)

15. C. Farnier and J.-P. Lenain. Performance studies for an altitude of 3700 m and 2000 m with Moonlight. CTA internal note. CTA consortium, Sept. 2011

16. M. Actis, G. Agnetta, F. Aharonian, et al. “Design concepts for the Cherenkov Telescope Array CTA: an advanced facility for ground-based high-energy gamma-ray astronomy.” Experimental Astronomy 32 (Dec. 2011), pp. 193– 316. doi: 10.1007/s10686-011-9247-0

17. J.-P. Lenain and R. Walter. “Search for high-energy γ-ray emission from galaxies of the Local Group with Fermi/LAT.” A&A 535, A19 (Nov. 2011), A19. doi: 10.1051/0004-6361/201117523. arXiv: 1110.3905 [astro-ph.HE]

18. J.-P. Lenain, C. Ricci, M. Türler, D. Dorner, and R. Walter. “Seyfert 2 galaxies in the GeV band: jets and starburst.” A&A 524 (Dec. 2010), A72+. doi: 10 . 1051 / 0004 - 6361 / 201015644. arXiv: 1008 . 5164 [astro-ph.CO]

19. V. A. Acciari, E. Aliu, T. Arlen, et al. “Radio Imaging of the Very-High-Energy γ-Ray Emission Region in the Central Engine of a Radio Galaxy.” Science 325 (July 2009), pp. 444–. doi: 10.1126/science.1175406. arXiv: 0908.0511 (corresponding author for the “Supporting Online Material”)

20. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Discovery of Very High Energy γ-Ray Emission from Centaurus A with H.E.S.S..” ApJ 695 (Apr. 2009), pp. L40–L44. doi: 10.1088/0004-637X/695/1/L40. arXiv: 0903.1582 (corresponding author)

21. K. Katarzyński, J.-P. Lenain, A. Zech, C. Boisson, and H. Sol. “Modelling rapid TeV variability of PKS2155-304.” MNRAS 390 (Oct. 2008), pp. 371–376. doi: 10.1111/j.1365-2966.2008.13753.x. arXiv: 0807.4533

22. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Discovery of VHE γ-rays from the high-frequency- peaked BL Lacertae object RGB J0152+017.” A&A 481 (Apr. 2008), pp. L103–L107. doi: 10.1051/0004- 6361:200809603. eprint: arXiv:0802.4021 (corresponding author)

23. J.-P. Lenain, C. Boisson, H. Sol, and K. Katarzyński. “A synchrotron self-Compton scenario for the very high energy γ-ray emission of the radiogalaxy M 87. Unifying the TeV emission of blazars and other AGNs?” A&A 478 (Jan. 2008), pp. 111–120. doi: 10.1051/0004-6361:20077995

24. F. Aharonian, A. G. Akhperjanian, A. R. Bazer-Bachi, et al. “Fast variability of tera-electron volt rays from the radio galaxy M87.” Science 314 (2006), pp. 1424–1427. eprint: arXiv:astro-ph/0612016 (exceptional pre-doctoral co-authorship following the work accomplished during my Master 2 internship)

In the framework of my participation in several collaborations, especially H.E.S.S. and CTA, I also contributed to the following publications:

25. H. Abdalla, A. Abramowski, F. Aharonian, et al. “TeV Gamma-Ray Observations of the Binary Neutron Star Merger GW170817 with H.E.S.S..” ApJ 850, L22 (Dec. 2017), p. L22. doi: 10.3847/2041-8213/aa97d2. arXiv: 1710.05862 [astro-ph.HE]

26. H. E. S. S. Collaboration, : H. Abdalla, et al. “HESS J1741-302: a hidden accelerator in the Galactic plane.” ArXiv e-prints (Nov. 2017). arXiv: 1711.01350 [astro-ph.HE] (accepted for publication in A&A)

27. B. P. Abbott, R. Abbott, T. D. Abbott, et al. “Multi-messenger Observations of a Binary Neutron Star Merger.” ApJ 848, L12 (Oct. 2017), p. L12. doi: 10.3847/2041-8213/aa91c9. arXiv: 1710.05833 [astro-ph.HE]

28. H. E. S. S. Collaboration, H. Abdalla, A. Abramowski, et al. “Measurement of the EBL spectral energy distribution using the VHE γ-ray spectra of H.E.S.S. blazars.” A&A 606, A59 (Oct. 2017), A59. doi: 10.1051/0004- 6361/201731200. arXiv: 1707.06090 [astro-ph.HE]

143 C. Publication list

29. CTA Consortium, B. S. Acharya, I. Agudo, et al. “Science with the Cherenkov Telescope Array.” ArXiv e-prints (Sept. 2017). arXiv: 1709.07997 [astro-ph.IM]

30. E. Petroﬀ, S. Burke-Spolaor, E. F. Keane, et al. “A polarized fast radio burst at low Galactic latitude.” MNRAS 469 (Aug. 2017), pp. 4465–4482. doi: 10.1093/mnras/stx1098. arXiv: 1705.02911 [astro-ph.HE]

31. MAGIC Collaboration, M. L. Ahnen, S. Ansoldi, et al. “Constraints on particle acceleration in SS433/W50 from MAGIC and H.E.S.S. observations.” ArXiv e-prints (July 2017). arXiv: 1707.03658 [astro-ph.HE] (accepted for publication in A&A)

32. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Characterising the VHE diﬀuse emission in the central 200 parsecs of our Galaxy with H.E.S.S.” ArXiv e-prints (June 2017). arXiv: 1706.04535 [astro-ph.HE] (accepted for publication in A&A)

33. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Systematic search for very-high-energy gamma-ray emission from bow shocks of runaway stars.” ArXiv e-prints (May 2017). arXiv: 1705.02263 [astro-ph.HE] (accepted for publication in A&A)

34. F. Acero, R. Aloisio, J. Amans, et al. “Prospects for Cherenkov Telescope Array Observations of the Young Supernova Remnant RX J1713.7-3946.” ApJ 840, 74 (May 2017), p. 74. doi: 10.3847/1538-4357/aa6d67. arXiv: 1704.04136 [astro-ph.HE]

35. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “The population of TeV pulsar wind nebulae in the H.E.S.S. Galactic Plane Survey.” ArXiv e-prints (Feb. 2017). arXiv: 1702.08280 [astro-ph.HE] (accepted for publication in A&A)

36. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Characterizing the γ-ray long-term variability of PKS 2155-304 with H.E.S.S. and Fermi-LAT.” A&A 598, A39 (Feb. 2017), A39. doi: 10.1051/0004-6361/ 201629419. arXiv: 1610.03311 [astro-ph.HE]

37. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “First limits on the very-high energy gamma-ray afterglow emission of a fast radio burst. H.E.S.S. observations of FRB 150418.” A&A 597, A115 (Jan. 2017), A115. doi: 10.1051/0004-6361/201629117. arXiv: 1611.09209 [astro-ph.HE]

38. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Deeper H.E.S.S. Observations of Vela Junior (RX J0852.0-4622): Morphology Studies and Resolved Spectroscopy.” ArXiv e-prints (Nov. 2016). arXiv: 1611.01863 [astro-ph.HE] (accepted for publication in A&A)

39. H. Abdalla, A. Abramowski, F. Aharonian, et al. “H.E.S.S. Limits on Linelike Dark Matter Signatures in the 100 GeV to 2 TeV Energy Range Close to the Galactic Center.” Physical Review Letters 117.15, 151302 (Oct. 2016), p. 151302. doi: 10.1103/PhysRevLett.117.151302. arXiv: 1609.08091 [astro-ph.HE]

40. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Search for Dark Matter Annihilations towards the Inner Galactic Halo from 10 Years of Observations with H.E.S.S..” Physical Review Letters 117.11, 111301 (Sept. 2016), p. 111301. doi: 10.1103/PhysRevLett.117.111301. arXiv: 1607.08142 [astro-ph.HE]

41. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “H.E.S.S. observations of RX J1713.7-3946 with improved angular and spectral resolution; evidence for gamma-ray emission extending beyond the X-ray emitting shell.” ArXiv e-prints (Sept. 2016). arXiv: 1609.08671 [astro-ph.HE] (accepted for publication in A&A)

42. H. E. S. S. Collaboration, H. Abdalla, A. Abramowski, et al. “The supernova remnant W49B as seen with H.E.S.S. and Fermi-LAT.” ArXiv e-prints (Sept. 2016). arXiv: 1609.00600 [astro-ph.HE] (accepted for publication in A&A)

43. H. E. S. S. Collaboration, H. Abdalla, A. Abramowski, et al. “A search for very high-energy ﬂares from the micro- quasars GRS 1915+105, Circinus X-1, and V4641 Sgr using contemporaneous H.E.S.S. and RXTE observations.” ArXiv e-prints (July 2016). arXiv: 1607.04613 [astro-ph.HE] (accepted for publication in A&A)

44. H.E.S.S. Collaboration, H. Abdalla, A. Abramowski, et al. “Extended VHE gamma-ray emission towards SGR1806- 20, LBV1806-20, and stellar cluster Cl*1806-20.” ArXiv e-prints (June 2016). arXiv: 1606.05404 [astro-ph.HE] (accepted for publication in A&A)

144 45. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “Acceleration of petaelectronvolt protons in the Galactic Centre.” Nature 531 (Mar. 2016), pp. 476–479. doi: 10.1038/nature17147. arXiv: 1603.07730 [astro-ph.HE]

46. H. E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “Detailed spectral and morphological analysis of the shell type SNR RCW 86.” ArXiv e-prints (Jan. 2016). arXiv: 1601.04461 [astro-ph.HE] (accepted for publication in A&A)

47. H. E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “Discovery of variable VHE γ-ray emission from the binary system 1FGL J1018.6-5856.” A&A 577, A131 (May 2015), A131. doi: 10.1051/0004-6361/201525699. arXiv: 1503.02711 [astro-ph.HE]

48. B. S. Acharya, C. Aramo, A. Babic, et al. “The Cherenkov Telescope Array potential for the study of young supernova remnants.” Astroparticle Physics 62 (Mar. 2015), pp. 152–164. doi: 10.1016/j.astropartphys. 2014.08.005

49. A. Abramowski, F. Aharonian, F. Ait Benkhali, et al. “Constraints on an Annihilation Signal from a Core of Constant Dark Matter Density around the Milky Way Center with H.E.S.S..” Physical Review Letters 114.8, 081301 (Feb. 2015), p. 081301. doi: 10.1103/PhysRevLett.114.081301. arXiv: 1502.03244 [astro-ph.HE]

50. H. E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “H.E.S.S. reveals a lack of TeV emission from the supernova remnant Puppis A.” A&A 575, A81 (Feb. 2015), A81. doi: 10.1051/0004-6361/201424805. arXiv: 1412.6997 [astro-ph.HE]

51. H. E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “Probing the gamma-ray emission from HESS J1834-087 using H.E.S.S. and Fermi LAT observations.” A&A 574, A27 (Feb. 2015), A27. doi: 10.1051/0004- 6361/201322694. arXiv: 1407.0862 [astro-ph.HE]

52. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “The exceptionally powerful TeV γ-ray emitters in the Large Magellanic Cloud.” Science 347 (Jan. 2015), pp. 406–412. doi: 10.1126/science.1261313. arXiv: 1501.06578 [astro-ph.HE]

53. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of the VHE gamma-ray source HESS J1832- 093 in the vicinity of SNR G22.7-0.2.” MNRAS 446 (Jan. 2015), pp. 1163–1169. doi: 10.1093/mnras/stu2148. arXiv: 1411.0572 [astro-ph.HE]

54. H. E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “H.E.S.S. detection of TeV emission from the interaction region between the supernova remnant G349.7+0.2 and a molecular cloud.” A&A 574, A100 (Jan. 2015), A100. doi: 10.1051/0004-6361/201425070. arXiv: 1412.2251 [astro-ph.HE]. Erratum: H. E. S. S. Col- laboration, A. Abramowski, F. Aharonian, et al. “H.E.S.S. detection of TeV emission from the interaction region between the supernova remnant G349.7+0.2 and a molecular cloud (Corrigendum).” A&A 580, C1 (Aug. 2015), p. C1. doi: 10.1051/0004-6361/201425070e

55. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “The high-energy γ-ray emission of AP Librae.” A&A 573, A31 (Jan. 2015), A31. doi: 10.1051/0004-6361/201321436. arXiv: 1410.5897 [astro-ph.HE]

56. A. Abramowski, F. Aharonian, F. Ait Benkhali, et al. “Diﬀuse Galactic gamma-ray emission with H.E.S.S..” Phys. Rev. D 90.12, 122007 (Dec. 2014), p. 122007. doi: 10.1103/PhysRevD.90.122007. arXiv: 1411.7568 [astro-ph.HE]

57. A. Abramowski, F. Aharonian, F. Ait Benkhali, et al. “Search for dark matter annihilation signatures in H.E.S.S. observations of dwarf spheroidal galaxies.” Phys. Rev. D 90.11, 112012 (Dec. 2014), p. 112012. doi: 10.1103/ PhysRevD.90.112012. arXiv: 1410.2589 [astro-ph.HE]

58. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “Long-term monitoring of PKS 2155-304 with ATOM and H.E.S.S.: investigation of optical/γ-ray correlations in diﬀerent spectral states.” A&A 571, A39 (Nov. 2014), A39. doi: 10.1051/0004-6361/201424142. arXiv: 1409.0253 [astro-ph.HE]

59. A. Abramowski, F. Aharonian, F. Ait Benkhali, et al. “Discovery of the Hard Spectrum VHE γ-Ray Source HESS J1641-463.” ApJ 794, L1 (Oct. 2014), p. L1. doi: 10.1088/2041- 8205/794/1/L1. arXiv: 1408.5280 [astro-ph.HE]

145 C. Publication list

60. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “TeV γ-ray observations of the young synchrotron- dominated SNRs G1.9+0.3 and G330.2+1.0 with H.E.S.S..” MNRAS 441 (June 2014), pp. 790–799. doi: 10.1093/mnras/stu459. arXiv: 1404.1613 [astro-ph.HE]

61. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “Search for TeV Gamma-ray Emission from GRB 100621A, an extremely bright GRB in X-rays, with H.E.S.S..” A&A 565, A16 (May 2014), A16. doi: 10.1051/ 0004-6361/201322984. arXiv: 1405.0488 [astro-ph.HE]

62. A. Abramowski, F. Aharonian, F. A. Benkhali, et al. “HESS J1640-465 - an exceptionally luminous TeV γ- ray supernova remnant.” MNRAS 439 (Apr. 2014), pp. 2828–2836. doi: 10.1093/mnras/stu139. arXiv: 1401.4388 [astro-ph.HE]. Erratum: H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “Erratum: HESS J1640-465 - an exceptionally luminous TeV γ-ray supernova remnant.” MNRAS 441 (July 2014), pp. 3640– 3642. doi: 10.1093/mnras/stu826

63. H. E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “Flux upper limits for 47 AGN observed with H.E.S.S. in 2004-2011.” A&A 564, A9 (Mar. 2014), A9. doi: 10 . 1051 / 0004 - 6361 / 201322897. arXiv: 1402.2332

64. H E. S. S. Collaboration, A. Abramowski, F. Aharonian, et al. “H.E.S.S. observations of the Crab during its March 2013 GeV gamma-ray ﬂare.” A&A 562, L4 (Feb. 2014), p. L4. doi: 10.1051/0004-6361/201323013. arXiv: 1311.3187 [astro-ph.HE]

65. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “Search for extended γ-ray emission around AGN with H.E.S.S. and Fermi-LAT.” A&A 562, A145 (Feb. 2014), A145. doi: 10.1051/0004-6361/201322510. arXiv: 1401.2915 [astro-ph.HE]

66. H.E.S.S. Collaboration, A. Abramowski, F. Aharonian, et al. “HESS J1818-154, a new composite supernova remnant discovered in TeV gamma rays and X-rays.” A&A 562, A40 (Feb. 2014), A40. doi: 10.1051/0004- 6361/201322914. arXiv: 1310.6956 [astro-ph.HE]

67. E. Aliu, S. Archambault, T. Aune, et al. “Long-term TeV and X-Ray Observations of the Gamma-Ray Binary HESS J0632+057.” ApJ 780, 168 (Jan. 2014), p. 168. doi: 10.1088/0004-637X/780/2/168. arXiv: 1311.6083 [astro-ph.HE]

68. A. Abramowski, F. Acero, F. Aharonian, et al. “Constraints on axionlike particles with H.E.S.S. from the irregularity of the PKS 2155-304 energy spectrum.” Phys. Rev. D 88.10, 102003 (Nov. 2013), p. 102003. doi: 10.1103/ PhysRevD.88.102003. arXiv: 1311.3148 [astro-ph.HE]

69. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “HESS and Fermi-LAT discovery of γ-rays from the blazar 1ES 1312-423.” MNRAS 434 (Sept. 2013), pp. 1889–1901. doi: 10.1093/mnras/stt1081

70. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of high and very high-energy emission from the BL Lacertae object SHBL J001355.9-185406.” A&A 554, A72 (June 2013), A72. doi: 10.1051/0004- 6361/201220996. arXiv: 1304.4023 [astro-ph.HE]

71. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “H.E.S.S. observations of the binary system PSR B1259- 63/LS 2883 around the 2010/2011 periastron passage.” A&A 551, A94 (Mar. 2013), A94. doi: 10.1051/0004- 6361/201220612. arXiv: 1301.3930 [astro-ph.HE]

72. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Search for very-high-energy γ-ray emission from Galactic globular clusters with H.E.S.S..” A&A 551, A26 (Mar. 2013), A26. doi: 10.1051/0004-6361/201220719. arXiv: 1301.1678 [astro-ph.HE]

73. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Measurement of the extragalactic background light imprint on the spectra of the brightest blazars observed with H.E.S.S..” A&A 550, A4 (Feb. 2013), A4. doi: 10.1051/0004-6361/201220355. arXiv: 1212.3409 [astro-ph.HE]

74. A. Abramowski, F. Acero, F. Aharonian, et al. “Search for Photon-Linelike Signatures from Dark Matter An- nihilations with H.E.S.S..” Physical Review Letters 110.4, 041301 (Jan. 2013), p. 041301. doi: 10.1103/ PhysRevLett.110.041301. arXiv: 1301.1173 [astro-ph.HE]

146 75. T. Krähenbühl, H. Anderhub, M. Backes, et al. “G-APDs in Cherenkov astronomy: The FACT camera.” Nuclear Instruments and Methods in Physics Research A 695 (Dec. 2012), pp. 96–99. doi: 10.1016/j.nima.2011.11. 068

76. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Identiﬁcation of HESS J1303-631 as a pulsar wind nebula through γ-ray, X-ray, and radio observations.” A&A 548, A46 (Dec. 2012), A46. doi: 10.1051/0004- 6361/201219814

77. A. Abramowski, F. Acero, F. Aharonian, et al. “Probing the extent of the non-thermal emission from the Vela X region at TeV energies with H.E.S.S..” A&A 548, A38 (Dec. 2012), A38. doi: 10.1051/0004-6361/201219919. arXiv: 1210.1359 [astro-ph.HE]

78. A. Abramowski, F. Acero, F. Aharonian, et al. “Spectral Analysis and Interpretation of the γ-Ray Emission from the Starburst Galaxy NGC 253.” ApJ 757, 158 (Oct. 2012), p. 158. doi: 10.1088/0004-637X/757/2/158. arXiv: 1205.5485 [astro-ph.HE]

79. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of gamma-ray emission from the extragalactic pulsar wind nebula N 157B with H.E.S.S..” A&A 545, L2 (Sept. 2012), p. L2. doi: 10.1051/0004-6361/ 201219906. arXiv: 1208.1636 [astro-ph.HE]

80. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Constraints on the gamma-ray emission from the cluster- scale AGN outburst in the Hydra A galaxy cluster.” A&A 545, A103 (Sept. 2012), A103. doi: 10.1051/0004- 6361/201219655. arXiv: 1208.1370 [astro-ph.CO]

81. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “HESS observations of the Carina nebula and its enigmatic colliding wind binary Eta Carinae.” MNRAS 424 (July 2012), pp. 128–135. doi: 10.1111/j.1365-2966.2012. 21180.x. arXiv: 1204.5690 [astro-ph.HE]

82. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Discovery of VHE emission towards the Carina arm region with the H.E.S.S. telescope array: HESS J1018-589.” A&A 541, A5 (May 2012), A5. doi: 10.1051/0004- 6361/201218843. arXiv: 1203.3215 [astro-ph.HE]

83. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Simultaneous multi-wavelength campaign on PKS 2005- 489 in a high state.” A&A 533 (Sept. 2011), A110+. doi: 10.1051/0004-6361/201016170

84. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Search for Lorentz Invariance breaking with a likelihood ﬁt of the PKS 2155-304 ﬂare data taken on MJD 53944.” Astroparticle Physics 34 (Apr. 2011), pp. 738–747. doi: 10.1016/j.astropartphys.2011.01.007

85. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Detection of very-high-energy γ-ray emission from the vicinity of PSR B1706-44 and G 343.1-2.3 with H.E.S.S..” A&A 528, A143 (Apr. 2011), A143. doi: 10.1051/ 0004-6361/201015381. arXiv: 1102.0773 [astro-ph.HE]

86. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Revisiting the Westerlund 2 ﬁeld with the HESS telescope array.” A&A 525 (Jan. 2011), A46+. doi: 10 . 1051 / 0004 - 6361 / 201015290. arXiv: 1009 . 3012 [astro-ph.HE]

87. H.E.S.S. Collaboration, F. Acero, F. Aharonian, et al. “Discovery and follow-up studies of the extended, oﬀ-plane, VHE gamma-ray source HESS J1507-622.” A&A 525 (Jan. 2011), A45+. doi: 10.1051/0004-6361/201015187. arXiv: 1010.4907 [astro-ph.HE]

88. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Discovery of VHE γ-rays from the BL Lacertae object PKS 0548-322.” A&A 521 (Oct. 2010), A69+. doi: 10.1051/0004-6361/200912363

89. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “VHE γ-ray emission of PKS 2155-304: spectral and temporal variability.” A&A 520 (Sept. 2010), A83. doi: 10.1051/0004-6361/201014484

90. F. Acero, F. Aharonian, A. G. Akhperjanian, et al. “First detection of VHE γ-rays from SN 1006 by HESS.” A&A 516 (June 2010), A62+. doi: 10.1051/0004-6361/200913916

91. H.E.S.S. Collaboration, A. Abramowski, F. Acero, et al. “Multi-wavelength observations of H 2356-309.” A&A 516 (June 2010), A56+. doi: 10.1051/0004-6361/201014321

147 C. Publication list

92. F. Acero, F. Aharonian, A. G. Akhperjanian, et al. “Localizing the VHE γ-ray source at the Galactic Centre.” MNRAS 402 (Mar. 2010), pp. 1877–1882. doi: 10.1111/j.1365-2966.2009.16014.x

93. H.E.S.S. Collaboration, F. Acero, F. Aharonian, et al. “PKS 2005-489 at VHE: four years of monitoring with HESS and simultaneous multi-wavelength observations.” A&A 511 (Feb. 2010), A52+. doi: 10.1051/0004- 6361/200913073

94. H.E.S.S. Collaboration, F. Acero, F. Aharonian, et al. “HESS upper limits on very high energy gamma-ray emission from the microquasar GRS 1915+105.” A&A 508 (Dec. 2009), pp. 1135–1140. doi: 10.1051/0004- 6361/200913389

95. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Probing the ATIC peak in the cosmic-ray electron spectrum with H.E.S.S..” A&A 508 (Dec. 2009), pp. 561–564. doi: 10.1051/0004-6361/200913323

96. F. Acero, F. Aharonian, A. G. Akhperjanian, and et al. “Detection of gamma rays from a starburst galaxy.” Science 326 (2009), pp. 1080–1082

97. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Very high energy γ-ray observations of the binary PSR B1259-63/SS2883 around the 2007 Periastron.” A&A 507 (Nov. 2009), pp. 389–396. doi: 10.1051/0004- 6361/200912339

98. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Spectrum and variability of the Galactic center VHE γ-ray source HESS J1745-290.” A&A 503 (Sept. 2009), pp. 817–825. doi: 10.1051/0004-6361/200811569. arXiv: 0906.1247

99. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Simultaneous multiwavelength observations of the second exceptional γ-ray ﬂare of PKS 2155-304 in July 2006.” A&A 502 (Aug. 2009), pp. 749–770. doi: 10.1051/0004- 6361/200912128. arXiv: 0906.2002

100. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Constraints on the multi-TeV particle population in the Coma galaxy cluster with HESS observations.” A&A 502 (Aug. 2009), pp. 437–443. doi: 10.1051/0004- 6361/200912086. arXiv: 0907.0727

101. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Detection of very high energy radiation from HESS J1908+063 conﬁrms the Milagro unidentiﬁed source MGRO J1908+06.” A&A 499 (June 2009), pp. 723–728. doi: 10.1051/ 0004-6361/200811357. arXiv: 0904.3409

102. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Simultaneous Observations of PKS 2155 304 with HESS, − Fermi, RXTE, and Atom: Spectral Energy Distributions and Variability in a Low State.” ApJ 696 (May 2009), pp. L150–L155. doi: 10.1088/0004-637X/696/2/L150

103. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “HESS upper limit on the very high energy γ-ray emission from the globular cluster 47 Tucanae.” A&A 499 (May 2009), pp. 273–277. doi: 10.1051/0004-6361/200811564

104. F. Aharonian, A. G. Akhperjanian, U. B. de Almeida, et al. “Discovery of Gamma-Ray Emission From the Shell- Type Supernova Remnant RCW 86 With H.E.S.S..” ApJ 692 (Feb. 2009), pp. 1500–1505. doi: 10.1088/0004- 637X/692/2/1500

105. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “HESS observations of γ-ray bursts in 2003-2007.” A&A 495 (Feb. 2009), pp. 505–512. doi: 10.1051/0004-6361:200811072. arXiv: 0901.2187

106. F. Aharonian, A. G. Akhperjanian, G. Anton, et al. “Very high energy gamma-ray observations of the galaxy clusters Abell 496 and Abell 85 with HESS.” A&A 495 (Feb. 2009), pp. 27–35. doi: 10.1051/0004-6361: 200811372

107. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “A Search for a Dark Matter Annihilation Signal Toward the Canis Major Overdensity with H.E.S.S..” ApJ 691 (Jan. 2009), pp. 175–181. doi: 10.1088/0004- 637X/691/1/175

108. F. Aharonian, A. G. Akhperjanian, U. Barres DeAlmeida, et al. “H.E.S.S. Observations of the Prompt and Afterglow Phases of GRB 060602B.” ApJ 690 (Jan. 2009), pp. 1068–1073. doi: 10.1088/0004-637X/690/2/ 1068

148 109. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Energy Spectrum of Cosmic-Ray Electrons at TeV Energies.” Physical Review Letters 101.26 (Dec. 2008), p. 261104. doi: 10.1103/PhysRevLett.101.261104

110. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Simultaneous HESS and Chandra observations of Sagitarius A? during an X-ray ﬂare.” A&A 492 (Dec. 2008), pp. L25–L28. doi: 10.1051/0004-6361:200810912

111. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Discovery of a VHE gamma-ray source coincident with the supernova remnant CTB 37A.” A&A 490 (Nov. 2008), pp. 685–693. doi: 10.1051/0004- 6361: 200809722. arXiv: 0803.0702

112. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Limits on an Energy Dependence of the Speed of Light from a Flare of the Active Galaxy PKS 2155-304.” Physical Review Letters 101.17 (Oct. 2008), p. 170402. doi: 10.1103/PhysRevLett.101.170402

113. F. Aharonian, A. G. Akhperjanian, U. B. de Almeida, et al. “Search for gamma rays from dark matter annihilations around intermediate mass black holes with the HESS experiment.” Phys. Rev. D 78.7 (Oct. 2008), p. 072008. doi: 10.1103/PhysRevD.78.072008

114. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “HESS upper limits for Kepler’s supernova remnant.” A&A 488 (Sept. 2008), pp. 219–223. doi: 10.1051/0004-6361:200809401

115. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Chandra and HESS observations of the supernova remnant CTB 37B.” A&A 486 (Aug. 2008), pp. 829–836. doi: 10.1051/0004-6361:200809655

116. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Discovery of very-high-energy γ-ray emission from the vicinity of PSR J1913+1011 with HESS.” A&A 484 (June 2008), pp. 435–440. doi: 10.1051/0004- 6361:20078715

117. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Exploring a SNR/molecular cloud association within HESS J1745-303.” A&A 483 (May 2008), pp. 509–517. doi: 10.1051/0004-6361:20079230. eprint: arXiv:0803.2844

118. F. Aharonian, A. G. Akhperjanian, A. R. Bazer-Bachi, et al. “Discovery of very high energy gamma-ray emission coincident with molecular clouds in the W 28 (G6.4-0.1) ﬁeld.” A&A 481 (Apr. 2008), pp. 401–410. doi: 10.1051/0004-6361:20077765

119. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Upper limits from HESS active galactic nuclei observations in 2005-2007.” A&A 478 (Feb. 2008), pp. 387–393. doi: 10.1051/0004-6361:20078604. eprint: arXiv:0711.3196

120. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “HESS observations and VLT spectroscopy of PG 1553+113.” A&A 477 (Jan. 2008), pp. 481–489. doi: 10.1051/0004-6361:20078603. eprint: arXiv: 0710.5740

121. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “HESS very-high-energy gamma-ray sources without identiﬁed counterparts.” A&A 477 (Jan. 2008), pp. 353–363. doi: 10.1051/0004-6361:20078516. eprint: arXiv:0712.1173

122. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “New constraints on the mid-IR EBL from the HESS discovery of VHE γ-rays from 1ES 0229+200.” A&A 475 (Nov. 2007), pp. L9–L13. doi: 10.1051/0004- 6361:20078462

123. F. Aharonian, A. G. Akhperjanian, U. Barres de Almeida, et al. “Discovery of VHE γ-rays from the distant BL Lacertae 1ES 0347-121.” A&A 473 (Oct. 2007), pp. L25–L28. doi: 10.1051/0004-6361:20078412

124. F. Aharonian, A. G. Akhperjanian, A. R. Bazer-Bachi, et al. “An Exceptional Very High Energy Gamma-Ray Flare of PKS 2155-304.” ApJ 664 (Aug. 2007), pp. L71–L74. doi: 10.1086/520635. eprint: arXiv:0706.0797

149 C. Publication list Conference proceedings

1. M. Holler, J. Chevalier, J.-P. Lenain, M. de Naurois, and D. Sanchez. “Run-Wise Simulations for Imaging Atmospheric Cherenkov Telescope Arrays.” ArXiv e-prints (Nov. 2017). arXiv: 1711.01118 [astro-ph.HE] (ICRC 2017)

2. J. Lefaucheur, C. Boisson, Z. Bosnjak, et al. “Gammapy: high level data analysis for extragalactic science cases with the Cherenkov Telescope Array.” ArXiv e-prints (Sept. 2017). arXiv: 1709.10169 [astro-ph.HE] (ICRC 2017)

3. H. E. S. S. Collaboration, : H. Abdalla, et al. “Contributions of the High Energy Stereoscopic System (H.E.S.S.) to the 35th International Cosmic Ray Conference (ICRC), Busan, Korea.” ArXiv e-prints (Sept. 2017). arXiv: 1709.06442 [astro-ph.HE] (ICRC 2017)

4. F. Acero, B. S. Acharya, V. Acín Portella, et al. “Cherenkov Telescope Array Contributions to the 35th Interna- tional Cosmic Ray Conference (ICRC2017).” ArXiv e-prints (Sept. 2017). arXiv: 1709.03483 [astro-ph.HE] (ICRC 2017)

5. C. Deil, R. Zanin, J. Lefaucheur, et al. “Gammapy - A prototype for the CTA science tools.” ArXiv e-prints (Sept. 2017). arXiv: 1709.01751 [astro-ph.IM] (ICRC 2017)

6. S. Bonnefoy, T. Ashton, M. Backes, et al. “Performance of the upgraded H.E.S.S. cameras.” ArXiv e-prints (Aug. 2017). arXiv: 1708.04550 [astro-ph.IM] (ICRC 2017)

7. F. Schüssler, M. Seglar-Arroyo, M. Arrieta, et al. “Target of Opportunity Observations of Blazars with H.E.S.S.” ArXiv e-prints (Aug. 2017). arXiv: 1708.01083 [astro-ph.HE] (ICRC 2017)

8. C. Romoli, M. Zacharias, M. Meyer, et al. “Observation of the extremely bright ﬂare of the FSRQ 3C279 with H.E.S.S. II.” ArXiv e-prints (Aug. 2017). arXiv: 1708.00882 [astro-ph.HE] (ICRC 2017)

9. M. Cerruti, J.-P. Lenain, H. Prokoph, and for the H.E.S.S. Collaboration. “H.E.S.S. discovery of very-high-energy emission from the blazar PKS 0736+017: on the location of the γ-ray emitting region in FSRQs.” ArXiv e-prints (Aug. 2017). arXiv: 1708.00658 [astro-ph.HE] (ICRC 2017)

10. F. Schüssler, M. Backes, A. Balzer, et al. “H.E.S.S. observations following multi-messenger alerts in real-time.” ArXiv e-prints (Aug. 2017). arXiv: 1708.00466 [astro-ph.HE] (ICRC 2017)

11. S. Klepser, T. Ashton, M. Backes, et al. “Hardware and software architecture of the upgraded H.E.S.S. cameras.” ArXiv e-prints (July 2017). arXiv: 1707.04415 [astro-ph.IM] (ICRC 2017)

12. F. Schüssler, M. Backes, A. Balzer, et al. “The H.E.S.S. multi-messenger program: Searches for TeV gamma- ray emission associated with high-energy neutrinos.” 6th International Symposium on High Energy Gamma- Ray Astronomy. Vol. 1792. American Institute of Physics Conference Series. Jan. 2017, p. 060006. doi: 10.1063/1.4968989 (Gamma 2016)

13. M. Cerruti, M. Böttcher, N. Chakraborty, et al. “Target of opportunity observations of blazars with H.E.S.S..” 6th International Symposium on High Energy Gamma-Ray Astronomy. Vol. 1792. American Institute of Physics Conference Series. Jan. 2017, p. 050029. doi: 10.1063/1.4968975. arXiv: 1610.05523 [astro-ph.HE] (Gamma 2016)

14. S. Pita, P. Goldoni, C. Boisson, et al. “Redshift measurement of Fermi blazars for the Cherenkov telescope array.” American Institute of Physics Conference Series. Vol. 1792. American Institute of Physics Conference Series. Jan. 2017. doi: 10.1063/1.4968971 (Gamma 2016)

15. M. Zacharias, M. Böttcher, N. Chakraborty, et al. “The complex VHE and multiwavelength ﬂaring activity of the FSRQ PKS 1510-089 in May 2015.” 6th International Symposium on High Energy Gamma-Ray Astronomy. Vol. 1792. American Institute of Physics Conference Series. Jan. 2017, p. 050023. doi: 10.1063/1.4968969. arXiv: 1611.02098 [astro-ph.HE] (Gamma 2016)

150 16. D. Zaborov, A. M. Taylor, D. A. Sanchez, et al. “Gamma-ray blazar spectra with H.E.S.S. II mono analysis: The case of PKS 2155-304 and PG 1553+113.” 6th International Symposium on High Energy Gamma-Ray Astronomy. Vol. 1792. American Institute of Physics Conference Series. Jan. 2017, p. 050017. doi: 10.1063/1.4968963. arXiv: 1612.05111 [astro-ph.HE] (Gamma 2016)

17. T. CTA Consortium, : A. Abchiche, et al. “Contributions of the Cherenkov Telescope Array (CTA) to the 6th International Symposium on High-Energy Gamma-Ray Astronomy (Gamma 2016).” ArXiv e-prints (Oct. 2016). arXiv: 1610.05151 [astro-ph.HE] (Gamma 2016)

18. A. Bulgarelli, K. Kosack, J. Hinton, et al. “The Cherenkov Telescope Array Observatory: top level use cases.” Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series. Vol. 9913. Proc. SPIE. Aug. 2016, p. 991331. doi: 10.1117/12.2232224

19. P. Goldoni, S. Pita, C. Boisson, et al. “X-shooter spectroscopy of the puzzling gamma-ray source 3FGL1603.9- 4903/PMN J1603-4904.” Active Galactic Nuclei: What’s in a Name? Aug. 2016, p. 88. doi: 10.5281/zenodo. 60615

20. T. CTA Consortium, : A. Abchiche, et al. “CTA Contributions to the 34th International Cosmic Ray Conference (ICRC2015).” ArXiv e-prints (Aug. 2015). arXiv: 1508.05894 [astro-ph.HE] (ICRC 2015)

21. J.-F. Glicenstein, O. Abril, J.-A. Barrio, et al. “NectarCAM : a camera for the medium size telescopes of the Cherenkov Telescope Array.” 34th International Cosmic Ray Conference (ICRC2015). Vol. 34. International Cosmic Ray Conference. July 2015, p. 937. arXiv: 1508.06555 [astro-ph.IM] (ICRC 2015)

22. H. Prokoph, Y. Becherini, M. Böttcher, et al. “H.E.S.S. discovery of very-high-energy gamma-ray emission of PKS 1440-389.” 34th International Cosmic Ray Conference (ICRC2015). Vol. 34. International Cosmic Ray Conference. July 2015, p. 862 (ICRC 2015)

23. C. van Eldik, M. Holler, D. Berge, et al. “Observations of the Crab Nebula with H.E.S.S. phase II.” 34th Interna- tional Cosmic Ray Conference (ICRC2015). Ed. by A. S. Borisov, V. G. Denisova, Z. M. Guseva, et al. Vol. 34. International Cosmic Ray Conference. July 2015, p. 847 (ICRC 2015)

24. J. Chevalier, M. A. Kastendieck, F. M. Rieger, et al. “Long term variability of the blazar PKS 2155-304.” 34th International Cosmic Ray Conference (ICRC2015). Vol. 34. International Cosmic Ray Conference. July 2015, p. 829 (ICRC 2015)

25. D. Zaborov, C. Romoli, A. Taylor, et al. “AGN observations with a 100 GeV threshold using H.E.S.S. II.” 34th International Cosmic Ray Conference (ICRC2015). Ed. by A. S. Borisov, V. G. Denisova, Z. M. Guseva, et al. Vol. 34. International Cosmic Ray Conference. July 2015, p. 808. arXiv: 1509.06509 [astro-ph.HE] (ICRC 2015)

26. D. A. Sanchez, F. Brun, C Couturier, et al. “Probe of Lorentz Invariance Violation eﬀects and determination of the distance of PG 1553+113.” ArXiv e-prints (Feb. 2015). arXiv: 1502.02920 [astro-ph.HE]

27. J.-P. Lenain. “The H.E.S.S. Extragalactic Sky.” International Journal of Modern Physics Conference Series 28, 1460163 (Mar. 2014), p. 60163. doi: 10.1142/S201019451460163X. arXiv: 1310.7736 [astro-ph.HE] (invited oral contribution)

28. K. Bernlöhr, A. Barnacka, Y. Becherini, et al. “Progress in Monte Carlo design and optimization of the Cherenkov Telescope Array.” ArXiv e-prints (July 2013). arXiv: 1307.2773 [astro-ph.IM] (ICRC 2013)

29. T. CTA Consortium, : O. Abril, et al. “CTA contributions to the 33rd International Cosmic Ray Conference (ICRC2013).” ArXiv e-prints (July 2013). arXiv: 1307.2232 [astro-ph.HE] (ICRC 2013)

30. J.-P. Lenain and M. de Naurois. “Utilisation de la grille EGI pour l’astrophysique de très hautes énergies avec H.E.S.S..” Journées SUCCES 2013. Paris, France, Nov. 2013. url: https://hal.archives-ouvertes.fr/ hal-00927503

31. L. Arrabito, C. Barbier, R. Graciani Diaz, et al. “Application of the DIRAC framework to CTA: ﬁrst evaluation.” Journal of Physics Conference Series 396.3 (Dec. 2012), p. 032007. doi: 10.1088/1742-6596/396/3/032007

151 C. Publication list

32. V. Stamatescu, Y. Becherini, K. Bernlöhr, et al. “Towards an optimized design for the Cherenkov Telescope Array.” American Institute of Physics Conference Series. Ed. by F. A. Aharonian, W. Hofmann, and F. M. Rieger. Vol. 1505. American Institute of Physics Conference Series. Dec. 2012, pp. 758–761. doi: 10.1063/1.4772370. arXiv: 1211.3856 [astro-ph.IM] (Gamma 2012)

33. S. Pita, P. Goldoni, C. Boisson, et al. “High energy blazars spectroscopy with X-shooter on the VLT.” American Institute of Physics Conference Series. Ed. by F. A. Aharonian, W. Hofmann, and F. M. Rieger. Vol. 1505. American Institute of Physics Conference Series. Dec. 2012, pp. 566–569. doi: 10.1063/1.4772323. arXiv: 1208.1785 [astro-ph.HE] (Gamma 2012)

34. D. Wouters, J.-P. Lenain, Y. Becherini, et al. “H.E.S.S. observations of the distant BL Lac PKS 0301-243.” American Institute of Physics Conference Series. Ed. by F. A. Aharonian, W. Hofmann, and F. M. Rieger. Vol. 1505. American Institute of Physics Conference Series. Dec. 2012, pp. 498–501. doi: 10.1063/1.4772306 (Gamma 2012)

35. T. Bretz, H. Anderhub, M. Backes, et al. “A status report:. FACT - a fact!” Astroparticle, Particle, Space Physics and Detectors For Physics Applications - Proceedings of the 13th ICATPP Conference. Edited by Giani Simone et al. Published by World Scientiﬁc Publishing Co. Pte. Ltd., 2012. ISBN #9789814405072, pp. 29-33. Ed. by S. Giani and et al. Aug. 2012, pp. 29–33. doi: 10.1142/9789814405072_0005

36. C. Barbier, S. Elles, N. Komin, et al. “CTACG - CTA Computing Grid.” Rencontres Scientiﬁques France Grilles 2011. Lyon, France, Sept. 2011. url: https://hal.archives-ouvertes.fr/hal-00653017

37. H. Anderhub, M. Backes, A. Biland, et al. “Electronics for the camera of the First G-APD Cherenkov Telescope (FACT) for ground based gamma-ray astronomy.” Journal of Instrumentation 7 (Jan. 2012), p. C1073. doi: 10.1088/1748-0221/7/01/C01073

38. F. di Pierro. “Performance studies of the CTA observatory.” International Cosmic Ray Conference 9 (2011), p. 54. doi: 10.7529/ICRC2011/V09/0684 (ICRC 2011)

39. A. Biland. “First Results from the First G-APD Cherenkov Telescope.” International Cosmic Ray Conference 9 (2011), p. 195. doi: 10.7529/ICRC2011/V09/1125 (ICRC 2011)

40. T. Bretz. “Status of the First G-APD Cherenkov Telescope (FACT).” International Cosmic Ray Conference 9 (2011), p. 203. doi: 10.7529/ICRC2011/V09/1132 (ICRC 2011)

41. B. Huber. “Solid light concentrators for small-sized photosensors used in Cherenkov telescopes.” International Cosmic Ray Conference 9 (2011), p. 2. doi: 10.7529/ICRC2011/V09/0136 (ICRC 2011)

42. P. Vogler. “Trigger and Data Acquisition electronics for the Geiger-mode avalanche photodiode Cherenkov Tele- scope Camera of FACT mount for the high energy section of the Cherenkov Telescope Array.” International Cosmic Ray Conference 9 (2011), p. 18. doi: 10.7529/ICRC2011/V09/0389 (ICRC 2011)

43. T. Kraehenbuehl. “Calibrating the camera for the First G-APD Cherenkov Telescope (FACT).” International Cosmic Ray Conference 9 (2011), p. 30. doi: 10.7529/ICRC2011/V09/0529 (ICRC 2011)

44. J.-P. Lenain and R. Walter. “Search for high-energy gamma-ray emission from galaxies of the Local Group with Fermi/LAT.” ArXiv e-prints (Oct. 2011). arXiv: 1110.3905 [astro-ph.HE]

45. J.-P. Lenain, M. K. Daniel, C. Boisson, et al. “A Tour of VHE Emitting AGN: Prospects with CTA.” AGN Physics in the CTA Era (AGN 2011). 2011

46. M. Raue, J.-P. Lenain, F. A. Aharonian, et al. “Discovery of VHE γ-rays from Centaurus A.” Accretion and Ejection in AGN: a Global View. Ed. by L. Maraschi, G. Ghisellini, R. Della Ceca, and F. Tavecchio. Vol. 427. Astronomical Society of the Paciﬁc Conference Series. Oct. 2010, p. 302. arXiv: 0904.2654 [astro-ph.CO]

47. A. Zech, B. Behera, Y. Becherini, et al. “Discovery of VHE emission from PKS 0447-439 with H.E.S.S. and MWL studies.” 25th Texas Symposium on Relativistic Astrophysics. 2010, p. 200

48. Y. Becherini, B. Behera, J. Biteau, et al. “New AGNs discovered at VHE by H.E.S.S..” 25th Texas Symposium on Relativistic Astrophysics. 2010, p. 167. arXiv: 1105.5243 [astro-ph.HE]

152 49. J.-P. Lenain, C. Ricci, M. Turler, D. Dorner, and R. Walter. “Seyfert 2 galaxies in the GeV band: jets and starburst.” 25th Texas Symposium on Relativistic Astrophysics. 2010, p. 124. arXiv: 1102.5231 [astro-ph.HE]

50. J.-P. Lenain. “TeV Active Galactic Nuclei: multifrequency modeling.” Memorie della Societa Astronomica Italiana 81 (2010), p. 362. arXiv: 0907.1832 (invited oral contribution)

51. M. M. Reynoso, M. C. Medina, G. E. Romero, et al. “Model for the High Energy Emission from CEN a.” International Journal of Modern Physics D 19 (2010), pp. 949–955. doi: 10.1142/S0218271810016750

52. J.-P. Lenain, M. C. Medina, C. Boisson, et al. “A synchrotron self-Compton model for the VHE gamma-ray emission from Cen A.” ArXiv e-prints (July 2009). arXiv: 0907.2258 (ICRC 2009)

53. J.-P. Lenain, C. Boisson, H. Sol, et al. “Very High Energy Active Galactic Nuclei Synchrotron Self-Compton Modeling Tour.” International Journal of Modern Physics D 18 (2009), pp. 1535–1539. doi: 10 . 1142 / S0218271809015503. arXiv: 0904.1660 [astro-ph.CO]

54. S. Kaufmann, L. Gerard, B. Giebels, et al. “Multiwavelength analysis of the TeV Blazar RGB J0152+017.” American Institute of Physics Conference Series. Vol. 1085. American Institute of Physics Conference Series. Dec. 2008, pp. 549–552. doi: 10.1063/1.3076731 (Gamma 2008)

55. J.-P. Lenain, W. Benbow, C. Boisson, et al. “PKS 2155-304 in July 2006: H.E.S.S. results and simultaneous multi-wavelength observations.” American Institute of Physics Conference Series. Vol. 1085. American Institute of Physics Conference Series. Dec. 2008, pp. 415–418. doi: 10.1063/1.3076696 (Gamma 2008)

56. J.-P. Lenain, C. Boisson, and H. Sol. “SSC scenario for VHE emission from 2 radiogalaxies: M 87 and Cen A.” ArXiv e-prints 807 (July 2008). eprint: 0807.2733

57. J.-P. Lenain. “Multiwavelength modeling of TeV AGN observed by H.E.S.S..” Revista Mexicana de Astronomia y Astroﬁsica Conference Series. Vol. 32. Revista Mexicana de Astronomia y Astroﬁsica Conference Series. Apr. 2008, pp. 28–30

58. J.-P. Lenain, C. Boisson, and H. Sol. “SSC Scenario for Tev Emission from Non-Blazar AGNs.” International Journal of Modern Physics D 17 (2008), pp. 1577–1584. doi: 10.1142/S0218271808013170. arXiv: 0712.3075

59. D. Nedbal, J.-P. Lenain, W. Benbow, et al. “Discovery and Multi-wavelength Study of a BL Lac RGB J0152+017.” 37th COSPAR Scientiﬁc Assembly. Held 13-20 July 2008, in Montréal, Canada., p.2192. Vol. 37. COSPAR, Plenary Meeting. 2008, p. 2192

60. J.-P. Lenain, D. Nedbal, M. Raue, et al. “discovery of VHE gamma rays from RGB J0152 017.” Blazar Variability across the Electromagnetic Spectrum. 2008

61. J.-P. Lenain. “Predictions of very high energy gamma-ray ﬂuxes for three Active Galactic Nuclei.” SF2A-2007: Proceedings of the Annual meeting of the French Society of Astronomy and Astrophysics held in Grenoble, France, July 2-6, 2007, Eds.: J. Bouvier, A. Chalabaev, and C. Charbonnel, p.200. Ed. by J. Bouvier, A. Chalabaev, and C. Charbonnel. July 2007, p. 200

62. J.-P. Lenain. “Modeling of the multiwavelength emission of M87 with H.E.S.S. observations.” SF2A-2007: Proceedings of the Annual meeting of the French Society of Astronomy and Astrophysics held in Grenoble, France, July 2-6, 2007, Eds.: J. Bouvier, A. Chalabaev, and C. Charbonnel, p.196. Ed. by J. Bouvier, A. Chalabaev, and C. Charbonnel. July 2007, p. 196

Alerts to the community

1. M. de Naurois. “Increased VHE activity from PKS 1510-089 detected with H.E.S.S..” The Astronomer’s Telegram 9102 (May 2016) (author)

2. C. Stegmann. “Increased VHE activity from Mrk 501 detected with H.E.S.S..” The Astronomer’s Telegram 6268 (June 2014) (author)

3. J.-P. Lenain, G. Cologna, S. Kaufmann, et al. “Optical activity of the ﬂaring gamma-ray quasar PKS 2326-502.” The Astronomer’s Telegram 4232 (July 2012), p. 1

153 C. Publication list

4. S. Schwemmer, G. Cologna, J.-P. Lenain, M. Mohamed, and S. J. Wagner. “Optical activity of the radio- and gamma-ray quasar CTA 102 (PKS 2230+114).” The Astronomer’s Telegram 4141 (June 2012), p. 1

5. G. Cologna, S. Kaufmann, J.-P. Lenain, et al. “Optical ﬂaring of the radio- and gamma-ray source PMN J2250- 2806.” The Astronomer’s Telegram 3787 (Nov. 2011), p. 1

6. J.-P. Lenain, G. Cologna, S. Kaufmann, S. Schwemmer, and S. Wagn. “Increased optical and gamma-ray emission of the quasar PMN J2345-1555.” The Astronomer’s Telegram 3734 (Nov. 2011), p. 1

7. M. Hauser, J.-P. Lenain, S. Wagner, and H. Hagen. “New ﬂaring activity in PKS 1510-089.” The Astronomer’s Telegram 3509 (July 2011), p. 1

Softwares

1. J.-P. Lenain. FLaapLUC: Fermi-LAT automatic aperture photometry light curve. Astrophysics Source Code Library. Sept. 2017. doi: 10.5281/zenodo.906991, accessible on GitHub (https://github.com/jlenain/ﬂaapluc).

Public outreach

1. J.-P. Lenain, M. Hauser, and S. J. Wagner. “PKS 1510-89: A Robotised Telescope Look at a Quasar.” The Astronomer 48 (Dec. 2011), pp. 211–212

2. J.-P. Lenain. “H.E.S.S. découvre une radiogalaxie émettant dans le domaine gamma.” L’Astronomie 123 (May 2009), p. 6

3. “H.E.S.S. découvre une radio-galaxie émettant dans le domaine gamma” (communiqué de presse, Avril 2009): CNRS/INSU, MPIK, IDW – Informationsdienst Wissenschaft, Futura Sciences

154

Contributions à l’astronomie γ des hautes énergies Noyaux actifs de galaxies et leptons cosmiques De H.E.S.S.à CTA

Résumé

Les noyaux actifs de galaxies, sources parmi les plus énergétiques dans l’Uni- vers, sont le siège de processus extrêmes rapidement variables. L’astrophysique des hautes énergies nous permet de les étudier. Ce mémoire d’habilitation à diriger des recherches présente quelques contributions à ce domaine, eﬀectuées ces huit dernières années. Les noyaux actifs de galaxies sont introduits en première partie, ainsi que l’étude de leurs éruptions aux hautes et très hautes énergies avec Fermi-LAT et H.E.S.S. La réponse instrumentale et les simulations liées à leur détermination pour H.E.S.S. II, ainsi qu’une étude de système de déclenchement topologique pour H.E.S.S. 1U sont présentées en seconde partie, qui comporte également deux études sur la réponse CTA en présence de lumière due à la Lune, et pour un site à haute altitude. La troisième partie porte sur l’étude d’une mise à jour du spectre d’électrons/positrons cosmiques mesuré avec H.E.S.S.

Contributions to high energy γ-ray astronomy Active galactic nuclei and leptonic cosmic rays From H.E.S.S. to CTA

Abstract

Active galactic nuclei, amongst the most energetic sources in the Universe, are the seat of highly variable, extreme processes. High energy astrophysics enables their study. This habilitation thesis presents some contributions in this field, carried out these last eight years. Active galactic nuclei are introduced in the first part, along with the study of their flares at high and very high energies with Fermi-LAT and H.E.S.S. The instrument response of H.E.S.S. II and related simulations, as well as a study on a topological trigger for H.E.S.S. 1U are presented in the second part, which also includes two studies one the response of CTA under Moon light, or for a high-altitude site. The third part focuses on an updated determination of the electron/positron cosmic-ray spectrum as measured with H.E.S.S.

Sorbonne Université, Univ Paris Diderot, Sorbonne Paris Cité, CNRS/IN2P3, Laboratoire de Physique Nucléaire et des Hautes Energies, LPNHE, 4 Place Jussieu, F-75005 Paris, France