Search for 750 GeV Resonance in MonoZ channel at CMS Experiment
by
Kamal Lamichhane, B.S.
AThesis
In
Physics
Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of
MASTER OF SCIENCE
Approved
Dr. Shuichi Kunori Chair of Committee
Dr. Nural Akchurin
Dr. Luis Grave de Peralta
Mark A. Sheridan Dean of the Graduate School
December, 2016 c 2016, Kamal Lamichhane Texas Tech University, Kamal Lamichhane, December 2016
ACKNOWLEDGEMENTS
First and foremost, I would like to express my kudos to my mentor and supervisor Shuichi Kunori. I can’t express in words how much he means to me for all the pain he has gone through while grooming my knowledge and understanding of underlying fundamental principles. A big thank you for his e↵orts to tag the skills of physics analysis on me. This analysis is quite unusual in a sense that it had a very short life span (less than a year) content, and he picked the channel that no one in the collaboration chose. In addition, this work had a strict time constraint as there was a very high competition with it both inside and outside of collaboration. Dr. Kunori’s desire to pursue this challenge was the major reason for the existence of this thesis, and I extend to him a big thank you for the trust he put in me in this regard and for the freedom he provided, even to test his own ideas. Of course, in the end I was wrong every time; but this freedom helped me to learn tremendously. His trust, support, and guidance has been very inspiring to me. It was an honor to work for such a first-rate individual, and I hope he is pleased with this work. I am very thankful to Nural Akchurin for his teaching, support, guidance, and encouragement both in the professional and personal sector. It was a great blessing to be a student of a person with such great idealism, knowledge, inspiration, clear-thinking, and calming influence. I am blessed to have him as a committee member of this thesis. IwouldliketoexpressmysinceregratitudetoLuisGravedePeraltaforkindly accepting my request that he become a thesis committee member. I have not yet had a chance to work with him directly; however, many of his students shared with me how enlightening he is at teaching and mentoring in both professional and personal matters. I would also like to thank Sung-won Lee, Igor Volobouev, Richard Wigmans, and the entire HEP group at TTU. Thank you Chris, Jordon, and Federico for your help. Thanks to my colleagues Tyler and Tielige for their friendship and all those hilarious moments of jokes and chats, including Manchester City and Arsenal’s football game. I would also like to thank Phil Harris from CERN, Adish Vartak
ii Texas Tech University, Kamal Lamichhane, December 2016 from UCSD, Zeynep Demiragli from MIT, Thiago Tomei from SPRACE, and Yalcin Guler from Cukurova University for their help and support. Thank you to the entire TTU physics department for all the support and to my undergraduate students, whom I had the honor to help and teach. I make a special mention to Mahdi Sanati: without your help, I am not sure if I would still be in the graduate school. Thank you for taking the pains to overcome my weakness and helping me to learn physics in an enthusiastic way, which helped me to pass the Ph.D prelim exam. A big thanks to Joel Walker (SHSU), and David Toback (TAMU); without you I don’t think I would be in the field of particle physics. I am indebted to your contribution in my career and life. Thank you for all the support you have provided. Finally, I am thankful to my family for their every support and understanding. Thanks to my wife Usha for all the support, love, encouragement, and especially for asking - “Is HEP just to run the codes?” - you are always inspiring me to push the extent of the knowledge I possess. Bless you for coming into my life.
iii Texas Tech University, Kamal Lamichhane, December 2016
TABLE OF CONTENTS
Acknowledgements ...... ii Abstract...... vi ListofFigures...... vii ListofAbbreviations...... ix 1. Introduction and Motivation ...... 1
2. Theoretical Overview ...... 5
2.1 Historicalperspective ...... 5 2.2 The Standard Model ...... 6 2.3 Comments on Constants ...... 9 3.ExperimentalApparatus ...... 11
3.1 TheLargeHadronCollider ...... 11 3.2 The CMS Detector ...... 13 3.2.1 The Coordinate System ...... 15 3.2.2 Tracking System ...... 15 3.2.3 Electromagnetic Calorimeter ...... 16 3.2.4 HadronicCalorimeter...... 17 3.2.5 Muon System ...... 18 3.2.6 TriggerSystemandDataAcquisition ...... 19 4. Experimental Tools and Objet Reconstruction ...... 22
4.1 JetsandJetReconstruction ...... 22 4.2 JetEnergyCorrection...... 23 4.3 BoostedObjectandJetSubstructure ...... 24 4.3.1 JetSubstructure ...... 26 4.4 Missing Transverse Energy and Transverse Mass ...... 30 5. Data/Monte Carlo Samples, and Event Selection/Rejection ...... 31
5.1 Data and Monte Carlo Sample ...... 31 5.1.1 Private RS Graviton Sample ...... 32 5.2 Event Selection ...... 32
iv Texas Tech University, Kamal Lamichhane, December 2016
5.3 Various Kinematic Distribution from Signal Selection . . . . . 36 6. Background Estimation, Results, and Systematics ...... 44
6.1 Background Estimation ...... 44 6.1.1 Error on Background Estimation, and Excess events . . . . 47 6.2 Limit Calculation on Production Cross Section ...... 47 6.3 Results ...... 49 6.4 Systematics ...... 51 7. Summary ...... 52
Bibliography ...... 53
v Texas Tech University, Kamal Lamichhane, December 2016
ABSTRACT
A search for the 750 GeV resonance on ZZ channel is presented with the 2016 1 ICHEP dataset of 12.9fb from pp collision at CMS experiment at LHC with the center of mass energy of 13 TeV. One Z decays to hadrons (jets), and the other decays to neutrinos (Missing energy). This work was motivated by the observation of the hint of resonance of 750 GeV mass on diphoton channel with 2015 data from both CMS and ATLAS experiment. Since the Z decays to a merged jet, jet substructure techniques are utilized for Z-tagging, which results in better signal selection.
vi Texas Tech University, Kamal Lamichhane, December 2016
LIST OF FIGURES
1.1 Diphoton excess results from CMS on 2015 ...... 3 1.2 DiphotonexcessresultsfromATLASon2015 ...... 3 1.3 BranchingratioofHiggsfromref[32] ...... 4 2.1 Radioactivity Experiment from ref [40] ...... 6 2.2 Standard Model of Particle Physics ...... 7 3.1 DiagramoftheCERNAcceleratorComplex ...... 12 3.2 Diagram of the CMS Detector ...... 14 3.3 CMS Tracker in rz plane ...... 16 3.4 GeometricviewofCMSECAL ...... 17 3.5 GeometricviewofCMSHCAL ...... 18 3.6 Geometric view of CMS Muon System ...... 19 3.7 Overview of CMS Trigger and DAQ system ...... 20 4.1 OverviewoffactorizedapproachofJECinCMS ...... 24 4.2 Overview of Boosted and resolved scenario from ref [needed] . . . . . 25 4.3 Schematic Representation of Boosted scenario ...... 26
4.4 Analytics of ⌧2 vs ⌧1 ...... 29 5.1 E ciency for ak8 vs ak4 jets for boosted object ...... 35 5.2 E ciencyforjetReconstructionforZcandidate ...... 35 5.3 NumberofVertex...... 36 5.4 NumberofAK8Jets ...... 37
5.5 Leading jet PT @(N-1)cut ...... 37
5.6 Leading Jet PT @Ncuts...... 38 5.7 PrunedMass@(N-1)cut ...... 38 5.8 Pruned Mass @ N cuts ...... 39
5.9 L1 Jet ⌧21 @(N-1)cut ...... 39
5.10 ⌧21 vs MPruned @(N-1)cutforsignal ...... 40
5.11 ⌧21 vs MPruned @(N-1)cutforBG ...... 40 5.12MET@(N-1)cut...... 41 5.13 MET @ N cut ...... 41 5.14MT@(N-1)cut...... 42
vii Texas Tech University, Kamal Lamichhane, December 2016
5.15 MToutsideofZselection@(N-1)cut ...... 42 5.16 MT @ N cut ...... 43 6.1 Zbar before and after correction ...... 45 6.2 Comparision of shape for Data vs BG ...... 45 6.3 Investigating (S-B), and S/B ...... 46 6.4 Exclusion of various production cross-section of the RS graviton . . . 49 6.5 Excess on channelin2015 ...... 50 6.6 Disappearance of excess in 2016 ...... 50
viii Texas Tech University, Kamal Lamichhane, December 2016
LIST OF ABBREVIATIONS
ALICE: A Large Ion Collider Experiment ATLAS: A Toroidal LHC ApparatuS CERN: European Organization for Nuclear Research CMS: Compact Muon Solenoid CSC: Cathode Strip Chamber DAQ: Data AcQuisition system DT: Drift Tube EM: ElectroMagnetic ECAL: Electromagnetic CALorimeter EWK: Electromagnetic and WeaK interaction ES: Endcap preShower PF: Particle Flow HB: Hadronic Barrel HCAL: Hadronic CALorimeter HE: Hadronic Endcap HF: Hadronic Forward HLT: High Level Trigge HO: Hadronic Outer ICHEP: International Conference on High Energy Physics JEC: Jet Energy Correction JES: Jet Energy Scale JER: Jet Energy Resolution L1: Level-1 Trigger LO: Leading Order LHC: Largh Hadron Collider LHCb: Large Hadron Collider Beauty experiment LINAC: Linear Acceleartor MC: Monte Carlo MET: Missing Transverse Energy NLO: Next-to-Leading Order
ix Texas Tech University, Kamal Lamichhane, December 2016
NNLL: Next-to-Next-to-Leading-Log NNLO: Next-to-Next-to-Leading Order PAT: Physics Analysis Tool PDF: Parton Distribution Function PF: Particle Flow POG: Particle Object Group PS: Proton Synchrotron PU: PileUp QCD: Quantum ChromoDynamics QED: Quantum ElectroDynamics QFT: Quantum Field Theory RECO: o✏ine RECOnstruction RPC: Resistive Plate Chamber SM: Standard Model SUSY: SUperSYmmetry TEC: The two End Cap detector system TIB: The Inner Barrel detector system TID: The Inner Discs detector system TOB: The Outter Barrel detector system
x Texas Tech University, Kamal Lamichhane, December 2016
CHAPTER 1
INTRODUCTION AND MOTIVATION
From the cosmological perspective, the composition of universe is categorized into three types, Dark energy ( 68%), Dark Matter ( 27%), and the ordinary matter ⇠ ⇠ ( 5%) [1]. The ordinary matter or the known part are ultimately made up of the ⇠ fundamental elementary particles which are experimentally discovered. These elementary particles are very well accommodated in the most promising and successful model of particle physics so far i.e. Standard Model. Standard model was built from the unification of QCD, flavor structure of the fermions, and the Electroweak theory proposed by Salam, Glashow, and Weinberg [2-9]. One after another discovery of the particles incorporated has made it inevitable. The last constituent of it was Higgs boson which was discovered on 2012 at the giant Large Hadron Collider experiment at CERN near Geneva[ 10-11], Switzerland. This discovery bagged the Nobel prize a year later[12], which made this model more elegant and particle physics subject a public fascination. Most of the particles of standard model are discovered at collider experiment. The current biggest accelerator facility is LHC with giant detectors [34]. The necessity of such a giant accelerator and detector to detect such tiny particles is explained on experimental apparatus section. LHC comprise four di↵erent detector with di↵erent purpose: the two biggest detectors, ATLAS [35] and CMS [36] are designed for Higgs boson discovery and dark matter, ALICE [37] is for primordial cosmic plasma at early universe, and LHCb [38] for matter-antimatter di↵erence. Newtonian physics talks about mass in terms of weight, and Einstein’s theory of relativity talks about mass in terms of energy; however, neither of these historic physics explain the origin of mass. Since, particles have mass, and we need the explanation not just to answer the existence of those particles, but ourselves also [13]. Robert Brout, Francois Englert, and Peter Higgs independently presented the phenomena of spontaneously symmetry breaking (SSB) as an answer for the source of the mass of particles[2, 14-16]. SSB requires the existence of another scalar particle which is now widely known as Higgs boson or god particle to non-physicist, the candidate for mass to the elementary particles. Higgs mass was an open window
1 Texas Tech University, Kamal Lamichhane, December 2016
(free parameter), which theorists initially predicted that it should be about 1000 GeV, but the ambitious experimentalists with the improved experimental conditions studied its coupling to other particles as a function of mass and concluded that the Higgs mass should be less than 200 GeV [13]. In 2012 this mystery was solved as the Higgs was discovered with mass of 125 GeV. This observed Higgs mass ⇠ m m + ,where is for correction, and it turned out that the h,observed ⇡ h,bare mh mh observed one is the di↵erence of two 30 decimal numbers resulting m 125 h,observed ⇡ GeV. This correction is known as hierarchy problem [16] or naturalness or fine tuning [2]. Standard model (SM) is simply overwhelming; however, it is not enough to understand the universe, nature, and the phenomena associated. SM could not provide answer to the following [13]:
dark matter, dark energy, matter-antimatter asymmetry, and gravity. • It now answers the origin of mass, but doesn’t answer the origin of matter, • We know from the neutrino oscillation that they have mass; however, SM • incorporates neutrinos as massless particle.
the cosmological observation of the enormous size of the universe, and some • more.
There is no any discovery after the Higgs boson to support any of the theoretical models available to explain these issues. Last December, ATLAS and CMS experiment [17-19] both reported the observation of some excess (local significance of 3.6 ,and2.6 respectively, and we need at least of 5 excess to claim a discovery) of 750 GeV signal on diphoton ( ) channel as shown in figure 1.1 from ref [33] and 1.2 from ref [17]. Since, both experiments have observed this excess, it had brought some excitement in the community, and many people worked on it if we can get 5 excess. I recalled a joke made by G. Isidori on 21 April 2016 at his presentation during CMS Week: ‘Some very negative conclusions have been derived in the last 2-3 years due to the absences of New Physics signals: PHD (= Post-Higgs
2 Texas Tech University, Kamal Lamichhane, December 2016
Figure 1.1. Diphoton excess results from Figure 1.2. Diphoton excess results from ATLAS CMS on 2015 on 2015
Discovery) depression syndrome ..but the situation is changing rapidly, thanks to a small DOSE (Diphoton Over- excitement for a Small Excess) of Run-II data’. We know photon is a spin 1 particle, so the 750 GeV resonance can’t be spin 1 particle as Landau-Yang theorem forbids the spin 1 particle to decay to two other spin 1 particle [20-22]. Hence, the options for 750 GeV object to exist is either a spin 2, which is graviton or spin 0, which is Higgs boson. 750 GeV became very enthusiastic because of the following reasons:
If this is spin 0 i.e. Higgs like particle then this could help to answer the • hierarchy problem mentioned above, as the theoretically expected Higgs mass was about 1000 GeV.
It may open an additional window on dark matter hunt, as some studies were • done to extend this to dark matter side [23-28].
It may also help on SUSY manifestation [29-30] or extra-dimensions [31]. • Final inevitable case is, it may even require something completely new • phenomena and physics to justify its existence.
3 Texas Tech University, Kamal Lamichhane, December 2016
If 750 GeV would be real spin 0 i.e. Higgs like particle, then from the known the menu of probability of higgs decaying to other particle (branching ratio), at 750 GeV mass from figure 1.1, H ZZ decay should be more than 103 times than the ! H decay [2, 32]. This was the motivation of my work, search for 750 GeV ! resonance on ZZ channel to test if it is higgs like particle. Furthermore, probability of Z boson decaying to hadrons (jets) is 70%, and to neutrinos (⌫)is 20% [2]. ⇡ ⇡ Hence, we require one Z to be invisible i.e. decay to ⌫⌫ or Missing Transverse Energy, and another Z to jets. So, we are directly measuring only one Z i.e. monoZ.
Figure 1.3. Branching ratio of Higgs from ref [32]
In this document, historical perspective of particle physics, and the theoretical overview of particle physics is explained on chapter 2, LHC experiment, and CMS detector are explained on chapter 3, experimental tools and objects of interest such as jets production, and substructure techniques are overviewed on chapter 4, chapter 5comprisedofdataandMontecarlosamplesused,Montecarlotechniquesforprivate production, and signal selection. Background estimation, results, limit calculation, and systematics are discussed on chapter 6, and a summary on chapter 7.
4 Texas Tech University, Kamal Lamichhane, December 2016
CHAPTER 2
THEORETICAL OVERVIEW
In this chapter, first the historical perspective of particle physics and then the current framework of particle physics i.e. Standard Model will be discussed.
2.1 Historical perspective A general question that we want to always know in detail is“What particle physics is about?”. It is about particles, and a very ancient question ( 2500 years ⇠ old): Is nature discrete i.e. does matter come with discrete (indivisible) unit which Greeks called “atom”? If so, everything in universe is made up out of something discrete called particles (perhaps a proper name for something discrete). On the other hand, the counter of this question is also ancient and equally legible: Is nature continuous i.e. Is matter uniformly smear out or continuously distributed in nature, which we called field. Field is basically a function in a given space, for example: density of a material, electric field, magnetic field in a given region of space. In some sense (quantum mechanics), neither and both turned out to be right [39]. The first evidence of the matter is discrete came from Chemistry not Physics, which was John Dalton’s idea of mass of a given molecule(or a mole of a given material) comes in discrete multiple of a mass of a mole of hydrogen. Almost everything is made up of a unit of hydrogen in terms of mass. This was one motivation of the periodic table in Chemistry. Evolution of particle physics began in 1896 with the radioactivity experiment of Henri Becquerel as shown in figure 2.1. While observing the radiation from a radioactive source in presence of magnetic field, he noticed the three distinct pattern, one completely undeflected (electrically neutral gamma radiation), two deflected asymmetrically on each side of the neutral one which are positively charged (alpha radiation), and negatively charged (beta radiation). Also, these are the first beam of particles in particle physics. This experiment also served to see the discreteness of particle. On the other hand, the wave nature of the particle was strongly demonstrated by the interference experiment with the interference pattern observed. In the recent era, we know atoms are divisible, and they are made up of protons,
5 Texas Tech University, Kamal Lamichhane, December 2016
Figure 2.1. Radioactivity Experiment from ref [40] electrons, and neutrons. Proton and neutron are made up of quarks. Hence, the quarks, electron (leptons) along with force carrier particles (gauge bosons) are the fundamental elementary particle which explains the latest understanding of universe (Standard Model).
2.2 The Standard Model Like the periodic table of elements in Chemistry, we have a categorical collection of elementary particles, governed by some rules which is called Standard Model of Particle Physics. As seen in figure 2.2 [41], SM comprise of two main categories of particles: fermions, and bosons. Fermions are half integer spin particles which are described with Fermi-Dirac statistics, and comes in three generations. On the other hand, bosons are integer spin particles and follow Bose-Einstein statistics. In addition, one can see from the spin perspective that a combination of two fermions give a boson, two boson give another boson, but a fermion with a boson give a fermion. Fermions are the fundamental constituents of matter whereas bosons are the force carriers and responsible for the interactions. Fermions are further categorized to quarks and leptons of six types each. There are three “up type” quarks which are up (u), charm
6 Texas Tech University, Kamal Lamichhane, December 2016
2 (c), and top (t) each with electric charge of + 3 ,andthree“downtype”whichare down (d), strange (s), and bottom (b) each with electric charge of 1 .Similarly, 3 among the six leptons: electron (e), muon (µ), and tau (⌧), are negatively charged with charge of 1, and their corresponding neutrinos i.e. electron neutrino (⌫ ), e muon neutrino (⌫µ), and tau neutrino (⌫⌧ ) are electrically neutral. On the boson side, there are gluons (g) for strong interaction, photon ( )forelectromagnetic interaction, W ±,andZ for short ranged weak interaction, and lastly the scalar (spin 0) Higgs (H) for the source of mass to other fundamental particles. As seen in the figure, graviton; the force carrier of gravitational force, resides outside of the standard model as gravity is very weak, and we don’t know why.
Figure 2.2. Standard Model of Particle Physics
Generally, the model building in particle physics is done with the input of some conditions, which are symmetries (expressed in terms of mathematical group
7 Texas Tech University, Kamal Lamichhane, December 2016 representation) and fields to a framework(Lagrangian) and the resulting Lagrangian is the formalism of the model. So, using the fields and SU(3) SU(2) U(1) C ⇥ L ⇥ Y symmetries, we get the resulting Lagrangian for SM up to dimension 4 as: = + + ,where is the Lagrangian for the gauge L LKinetic LHiggs LYukawa LKinetic interaction, for the Higgs field/boson, spontaneous symmetry breaking, and LHiggs giving mass to other particles, and is for the mixing of quarks, and for mass LYukawa of fermions via Yukawa interaction in Higgs condensate [2, 16]. One exception is about the mass of neutrinos which standard model assumes massless. This is L basically the concise physical description of Standard Model. So far there is no evidence of the single isolated quark, and the combination or mixing of quark are generally expressed in terms of a 3 3unitarymatrixcalled ⇥ “Cabibbo-Kobayashi-Maskawa (CKM) matrix”. The eigenvalues of CKM matrix sets the mass of the individual quarks [2, 4]. Discussing about the symmetry which is the main ingredient of standard model; each symmetry corresponds to respective interactions such as SU(3) for strong interactions/force mediated by color (red, blue, and green) charged bosons called gluons. Since, gluons are color charged particles, so they interact with each other. Eight 3 3unitaryGell-Mannmatrices,whicharethefundamentalsofthequark ⇥ model are the generators of SU(3) group. In principle, although there are 9 possible combinations; however, the requirement of determinant of SU(special unitary) matrices to be 1 kills the one possibility, also the trace of that one killed matrix is not 0, meaning no any physical significance. This follows the number of generators of a SU(n) to be n2 1; leaving 8 independent Gell-Mann matrices representing the di↵erent color combinations of gluons [2, 4, 42, 43]. As the name represents “Quantum Chromodynamics (QCD)” is the study of the color particles i.e. quarks, gluons and their strong interactions. Furthermore, SU(2) U(1) is for the unified L ⇥ Y electromagnetic and weak interaction called “Electroweak (EWK) interaction” which was introduced by Glashow-Weinberg-Salam [6-8]. The interaction of photon with electron is termed as “Quantum Electrodynamics (QED)” represented by U(1) symmetry [44]. Photons have no charge, hence doesn’t interact with each other. Neutral bosons (Z, g, & ) when coupling with fermions do not change the flavor which is explained mathematically from the resulting diagonalized matrix whereas
8 Texas Tech University, Kamal Lamichhane, December 2016
the charged boson W ± contributes on flavor change and the phenomena is explained as “Flavor Changing Charged Current”. In addition, particles are also categorized as left, and right handed. W boson only couples with left handed fermions, this is why standard model is often called a left handed one, but some theorists object to call the standard model a left handed model defending it as just a cause of vector interaction. Bosons in gauge theory are required to be massless as photon is for electromagnetic interaction on U(1) gauge symmetry; however, the W ±,andZ are quite heavy. This is due to SU(2) or Electroweak symmetry breaking mediated by the Higgs field, and this phenomena is widely known as Higgs mechanism. The broken symmetry, mass of W ±, Z, and the existence of Higgs boson are sourced from the massless scalar Nambu-Goldstone boson. Out of four components of the Nambu-Goldstone boson created from Higgs field, three are linked to the Z, W +, and W giving them mass, and the last one is represented as the Higgs boson with spin 0 [2, 16]. One final thing I would like to mention is about a concept in gauge theory i.e. “coupling constant” or the strength of an interaction which is analogous to charge. In microscopic quantum mechanics, charge is a dimensionless number measured in the unit of charge of an electron, and often called as an amplitude of an interaction. The square of amplitude is nothing but a probability. For e.g. e2 is the | | probability for a charge emitting at cathode and absorbed at anode, this is how charge e is explained. In QED, the interaction strength ↵,alsocalled‘Sommerfeld 2 constant’ or fine structure constant is given by ↵ = e ,whichis 1 .Thisis 4⇡ ⇠ 137 simply the probability of emitting one electron when 137 photons hit a cathode tube. This also hint that electromagnetic interaction or force is a weak force [39]. Also, the range of the force is inversely related to the mass of the force carriers; the photon is massless so the range of electromagnetic force is infinite and the W, Z bosons are heavier so the range of the Weak force is very small.
2.3 Comments on Constants In particle physics equations, couple of constants such as speed of light (c) and the planck’s constant (~)aresetto1.Thereasonforthisare:nothingcanmovefaster than light, so c is the upper limit constraint and from the uncertainty principle, ~ is
9 Texas Tech University, Kamal Lamichhane, December 2016 the maximum precision with which we can measure any parameter precisely [39].
10 Texas Tech University, Kamal Lamichhane, December 2016
CHAPTER 3
EXPERIMENTAL APPARATUS
2 h In particle physics, E = ~! = mc ,andp = are of very high importance, as they help to answer why particle physics is the way it is, and also why we need to build giant experiments to see tiny particles. Roughly speaking, size of accelerator is inversely related to the size of the particles we want to see. As particles have very small wavelength ( ), they have very high momentum (p). Hence, they require very high energy to get detected, which can be achieved by accelerating particles close to speed of light in those gigantic accelerators such as LHC at CERN near Geneva, Switzerland.
3.1 The Large Hadron Collider LHC is a gigantic accelerator to accelerate the protons to very high energy and collide them as it happened right after big bang, the starting point of our universe. The schematic of CERN accelerator complex is shown in figure 3.1 from ref [45]. All this starts with a highly compressed hydrogen in a cylinder which are injected in a precisely controlled rate to the source chamber of linear accelerator (LINAC2), where electrons are stripped o↵from the hydrogen leaving only proton. These protons are now accelerated with the electric field. By the time protons leave 1 LINAC2, they are now traveling with the speed of 3 rd of speed of light (c) [34, 46, 47]. From LINAC2, protons goes to booster, and in order to maximize the intensity of beam, protons are split into 4 booster rings. They are accelerated in booster by repeatedly circulating in the presence of the magnetic field (from powerful electromagnet to keep them in circular track) and pulsating electric field at a certain point (to accelerate) analogous to pushing a child in a swing every time he reaches to a certain point. At booster protons are accelerated up to 91.6% of c,and squeezes them closer together. After recombining the packet together they are now sent to proton synchrotron (PS). Let’s consider just two of such proton packets, they are circulated at PS for 1.2 seconds and gain the speed 99.9% of c.Thisspeedis ⇡ very close to the ultimate speed i.e. ‘c’sotheyreachedthepointoftransition,and the energy added from pulsating electric field now can not raise the speed. Instead
11 Texas Tech University, Kamal Lamichhane, December 2016 the increased energy contributes to increase the mass of protons (E = mc2).
Figure 3.1. Diagram of the CERN Accelerator Complex
Here, protons reach the energy of 25 GeV i.e. protons are now 25 times heavier than they are at rest. They are now sent to Super Proton Synchrotron (SPS) ring where protons are more energized to 450 GeV, and finally launched to the orbit (ring) of LHC which lies at 100 meters underground, and has 27 kilometers of circumference. At LHC ring, two vacuum pipes containing protons are traveling in opposite direction. The incoming protons are synchronized with those already circulating by the kickers magnet near injection point at ring. So, one vacuum pipe has clockwise circulating protons, and another has anti-clockwise circulating protons. These counter rotating beam cross over in four places at detector caverns (ATLAS, CMS, ALICE, LHCb), where they can be made to collide, and debris are tracked in detector. SPS injects protons for about half an hour and finally we will have 2808 packets of protons. At LHC ring, some extra energy is added to each
12 Texas Tech University, Kamal Lamichhane, December 2016 protons, whose speed is now so close to c, that it goes round the 27 Km ring over 11,000 times each second, getting a boost of energy from the pulsating electric field. Finally each proton has an energy of 6.5 TeV (although designed one is 7 TeV). Magnets are used to keep the protons in circular ring of LHC, and in order to run this set up properly, the magnets needs to be colder than outer space i.e. magnets become superconducting [48]. The steady magnet finally brings to the collision mode giving the center of mass energy (ps) of 13 TeV. Furthermore, the tubes where particles go around the accelerator are circulating in a vacuum which is similar to interplanetary space. The pressure on beam pipe is about ten times lower than the pressure on moon. We need that low pressure to make sure that the protons go around and collide to other proton rather than hitting some molecules of gas as they go around. Also, our cooling system is cooler than outer space. Our magnet cooling system is only 1.9 degrees above absolute zero temperature whereas the outer space temperature as measured in cosmic microwave background (CMB) radiation is 2.7 degree above the absolute zero degree. So, it is 0.8 degree cooler than outer space (cosmology) with the exception at collision point where the e↵ective temperature is very high.
The rate of events (Ne)forphysicsprocesswithcrosssection is related to instantaneous luminosity(L) of the accelerator by: kN2f Ne = L ;andL = ;wherek=numberofbunches,N=numberofproton 4⇡ x⇤ y⇤ 11 per bunches (1.15 10 ), f = revolution frequency (11.25 kHZ), and ⇤, ⇤ is beam ⇥ x y sizes at collision point (horizontal, vertical) = 16 µm.ThedesignedLis1 1034 ⇥ 2 1 cm s , but recently LHC is running so smoothly that it slightly exceeded it’s design luminosity.
3.2 The CMS Detector CMS as seen in figure 3.2 from ref [50] is one of the biggest detectors at LHC. It was primarily designed for Higgs boson discovery, which is now already achieved and the next target is to discover the dark matter candidate [36, 49, 50]. CMS detector is 21 m long, 15 m wide, and 15 m high, which is like a filter where, each layer stops, track or measure varieties of particles resulting from collision. Measuring the energy and momentum of those particles helps for the identification
13 Texas Tech University, Kamal Lamichhane, December 2016
Figure 3.2. Diagram of the CMS Detector of particle. The goals of CMS detector are: a high performance system to detect and measure muons as the name stands ‘Compact Muon Solenoid’, high resolution method to detect and measure electrons and photons which is why we have electromagnetic calorimeter, great quality of central tracking system for the precise momentum measurement, and a “hermetic” (air tight) hadron calorimeter to prevent the particles from escaping out. These motivates the necessity of a very powerful magnet, as the curve on the trajectory of a charged particle goes down with higher momentum. Hence, in order to have a precise measurement of even the very high momentum particles like muons, the necessity of the strong magnet is inevitable. This also facilitates on muon momentum measurement both inside the coil by tracking device, and outside by muon chambers. As the name suggests, CMS has the largest solenoidal magnet ever constructed, and the solenoid is a cylindrical coil of superconducting cable cooled to 268.50 C
14 Texas Tech University, Kamal Lamichhane, December 2016 which creates the magnetic field with the supply of the electricity. This solenoid has the dimension of 13 meters in length and 7 meters in diameter; and it produces the magnetic field strength of 4 Tesla which is 100000 Earth’s magnetic field ⇠ ⇥ strength. The beauty of this magnet is, the calorimeters and tracker can be placed inside the coil resulting the overall detector to be “compact”, this is how the name “Compact Muon Solenoid” stands. Beside the solenoidal magnet, CMS detector comprise of tracker, calorimeters (electromagnetic and hadronic), and muon system, and the data from these detector parts are handled through trigger and data acquisition system which are summarized in the respective sections below.
3.2.1 The Coordinate System CMS has the right handed coordinate system with origin at the collision point, x-axis points towards the center of LHC ring, y-axis points up perpendicular to the LHC plane, and the z-axis along the counter-clockwise beam direction [49]. In terms of polar coordinates, radial distance (r) is measured from the origin, the polar angle (✓)ismeasuredfromthepositivez-axis,andtheazimuthalangle( )ismeasuredin the x-y plane. Pseudorapidity (⌘)isthemostcommonlyusedcoordinateandis defined as ⌘ = ln(tan( ✓ )). Generally used coordinates are (⌘, ). 2
3.2.2 Tracking System As mentioned earlier, a good performance of tracker is necessary for the precise momentum measurement. In addition, with the increasing luminosity and pile up, the precise measurement of the primary interaction point (vertex) makes the role of tracking system crucial. Also, the tracker is helpful to reconstruct the trajectory of not only muons, electrons, and hadrons but also the tracks from decay of short lived particles such as ‘b quarks” which is the major candidate for the B-Physics, to study the matter-antimatter di↵erence in universe. CMS tracker [49, 51, 52] as shown in figure below is of cylindrical shape with dimension of 5.8 meters in length and 2.5 meters in diameter, and sits inside the solenoidal magnet covering the range up to ⌘ =2.5. It is all silicon detector with | | sensitive area of 200m2,andcompriseofpixeldetectoratcoresurroundedbystrip detector.
15 Texas Tech University, Kamal Lamichhane, December 2016
Figure 3.3. CMS Tracker in rz plane
The pixel has three cylindrical barrel layers at a distance of 4 centimeters (cm), 7 cm, and 11 cm respectively with two endcaps at either end. Each layer is split to segments each of 100µm 150µm.Ithas1440moduleswithabout65million ⇥ channels. The strip detector surrounds the pixel, and has 10 layers. It has about 10 millions strips in 15,200 module. Strip detector comprise of 4 inner barrel (TIB) layers configured in shells with 2 inner endcaps (TID) each with 3 small discs. Outer barrel (TOB) comprise of six concentric layers, and lastly the two endcaps (TEC) close the tracker.
3.2.3 Electromagnetic Calorimeter The purpose of ECAL [53] is to measure the momentum and energy of the electromagnetic particles like electrons and photons ( )primarilytargetedtodetect the ‘Higgs boson’ (H ). It sits in between the tracker and HCAL. ECAL is ! made up of lead tungstate (PbWO4), and it produces the scintillation light in proportion to the energy of the particle that hits it. ECAL as shown in figure 3.4 above, it comprises of barrel and two endcaps between tracker and HCAL. In addition, it also contains preshower (ES) sitting infront of endcap for additional spatial precision measurement. Barrel comprise of 61, 200 trapezoidal crystals, and on each side (+/ ) consists of 18 ‘supermodules’
16 Texas Tech University, Kamal Lamichhane, December 2016
Figure 3.4. Geometric view of CMS ECAL located at 200 interval on and each consisting of 1700 crystals. Barrel covers ⌘ = | | 1.48 with granularity of ⌘ =0.0175 0.0175 , and endcap covers 1.4 < ⌘ ⇥ ⇥ | | < 3withgranularityof ⌘ =0.05 0.05. ⇥ ⇥
3.2.4 Hadronic Calorimeter The purpose of HCAL [49, 54, 55] is to absorb/detect the hadron (things made up of quarks and gluons). It also gives the information about the direction of the neutrinos by measuring the missing transverse energy as described in section 4.4. It has 4 parts; barrel (HB) at ⌘ =1.3, endcap (HE) at 1.3 < ⌘ < 3, forward at | | | | 3 < ⌘ < 5.2, and finally outer (HO) which sits outside the magnet as a supplement | | to HB. The purpose of HO is to catch the left over shower that escapes out of HB; that is why it is also called the tail catcher. HB, and HE are the sampling calorimeters (i.e. finds the energy, position, and arrival time of particle using alternative layers of absorber and scintillator material which produce the light pulse as the particle hits and pass through it) made up of brass absorber plates and scintillator sheets as the active medium. Hybrid photo diodes (HPD) is used as the readout instrument to read signal (light) from scintillator. HF has the high radiation damage risk as it receives the most particle flux close to the beam line, so
17 Texas Tech University, Kamal Lamichhane, December 2016
Figure 3.5. Geometric view of CMS HCAL it uses steel as absorber and quartz fiber as an active material benefitting from their radiation hardness. HCAL has granularity of ⌘ =0.087 0.087 for ⌘ < 1.6, ⇥ ⇥ | | and ⌘ =0.17 0.17 for ⌘ 1.6. ⇥ ⇥ | |
3.2.5 Muon System CMS muon system [49, 56] are located outside of the calorimeters and magnet as shown in figure 3.6. Since muons have long life, and also the hadrons and electromagnetic particles are already absorbed by calorimeters, so having it outside of magnet helps to get the precise detection of muons and trajectory. Muon system covers up to ⌘ =2.4andconsistsoffollowingthreeparts: | | Drift Tube (DT): DT is located in barrel region up to ⌘ =1.3, and has 4 • | | stations (MB1-MB4).
18 Texas Tech University, Kamal Lamichhane, December 2016
Figure 3.6. Geometric view of CMS Muon System
Cathode Strip Chamber (CSC): CSC is located in endcap region up to ⌘ = • | | 2.4, and has 4 stations (ME1-ME4).
Resistive Plate Chamber (RPC): RPC are in both barrel and endcap region. • RPCs are gaseous plate chambers, and are very beneficial as they give good spatial and time resolution, as well as faster timing for signal.
3.2.6 Trigger System and Data Acquisition The purpose of the experiment is to get data to study physics processes. As the data rate and size of the events is very high hence, it is very challenging to store every single data. At CMS the interaction rate is of 1 GHz, the bunch crossing rate is 40 MHz, and average event size of 1MB. CMS has two level trigger system [57-59] which accept and output the data of interest at certain rate after some filtration or
19 Texas Tech University, Kamal Lamichhane, December 2016 selection cuts are applied to those events. Those events are categorized in terms of physics objects such as energy, momentum of electrons, photons, jets, taus, muons, missing energy, etc.
Figure 3.7. Overview of CMS Trigger and DAQ system
Level 1 (L1) trigger system is based on hardware systems i.e. local pattern recognition and energy evaluation on prompt macro-granular information form calorimeters and muon detectors, and these are linked to global trigger. L1 info is sent to detector frontends and read outs by TTC system, and the L1 output rate is 100kHz. The high level trigger (HLT) system which is a part of DAQ is software based. It gets the input from L1 and selects the events with higher purity for o✏ine storage. The daily data production rate is about 1TB. Data Acquisition system (DAQ) manages the overall flow of data. It comprise of detector frontend, readout modules, event builder, and monitoring systems. It
20 Texas Tech University, Kamal Lamichhane, December 2016 assembles the data fragments from the detector fragments with the trigger information and ultimately sends to the data storage center such as tier-0 site at CERN.
21 Texas Tech University, Kamal Lamichhane, December 2016
CHAPTER 4
EXPERIMENTAL TOOLS AND OBJET RECONSTRUCTION
4.1 Jets and Jet Reconstruction Proton consitutes of quarks and gluons which are also known as partons, and the partons together make hadrons. Since, partons have color charge so they cannot exist freely, and hence after the process of fragmentation or hadronisation they combine with other partons to form color neutral hadrons like ⇡,K giving the narrow conical like shape of spray of the hadrons called Jets. The definition of jet is not unique. Since at LHC, we have proton-proton collision, so jet is a de-facto object in physics analysis at LHC [60]. Jets are reconstructed with algorithms, and the most commonly used algorithm at LHC is anti-kt algorithm (one of the sequential recombination algorithm) which is bottom up (from small separation to higher separation of particles) reconstruction approach [61-63]. The main criteria for the algorithm is safety test i.e. algorithm should be infrared safe, meaning if a soft emission or a collinear splitting of a parton (usually gluons) occurs within a jet then the hard jets still remain unchanged. Failing the safety test allows the soft emission to cause the di↵erent (could even be infinite) sets of final jets which lacks the cancellation of infrared divergences from loop integrals by phase space integrals and voids the Kinoshita-Lee-Nauenberg (KLN) theorem [64]. Sequential Clustering Algorithm is based on the distance measurements such as distance between the two (i, and j) particles (dij), distance between a particle i and 2k 2k R2 2k the beam (diB), and is formulated as: dij = min(pti ,ptj ) R2 ;anddiB = pti where R2 =(⌘2 ⌘2)+( 2 2), p is the transverse momentum of the particle, and R i j i j t is the radius of the cone of jet. For low pt case we generally consider R =0.4which is called AK4 jet, and for high pt case we use R =0.8 which is AK8 jet and often called merged or fat jet. The parameter k sub-classify the jet algorithm as k =1is k algorithm, k = 0 is Cambridge-Aachen (CA) algorithm, and k = 1isanti k t t algorithm. Since anti k algorithm is the one which is predominantly used at LHC t physics, so I will focus on it. In anti k algorithm, the recombination scheme for N particles in final state is as t
22 Texas Tech University, Kamal Lamichhane, December 2016 follows:
first we calculate the distances d ,andd as mentioned above, and compare. • iB ij Find the minimum distance: If d is minimum than d ,declaretheith • iB ij particle as one jet and then repeat the same first step for other N-i particle.
If d is the minimum to d then combine those two particles, and repeat the • ij iB step 1 again. If we have more than one pair satisfying this condition, then the
one with higher pt gets the priority for combination.
These steps are repeated till all the particles are clustered into jet of desired R. • This algorithm does not restrict the number of jets and overlapping of jets as the distance between two jets dR has the least separation of dR2 = R2.Inorderto justify the safety test let’s say there is an emission of new soft particle i.e. p 0, t ! and since the low pt particles are clustered at the end so either it is reconstructed as asingle0pt jet which don’t have any physical importance or even if combined to the last jet which is the lowest pt jet. Hence, in any case it doesn’t e↵ect the hard jet which is reconstructed first. If we have collinear particles, then they have R2 =((⌘2 ⌘2)+( 2 2)) 0, and will be clustered first as suggested from the i j i j ! recombination scheme above without making any e↵ect on jets. These two justification proves that anti k algorithm is always infrared safe. Furthermore, the t recombination scheme has the another important aspect which is the addition of the four vectors of the particles in jet gives mass to the jet, even if the particles of jet are massless. This is significantly noticed in the boosted jet case, where pt is very large, hence the mass of jet seems to be su ciently large as well.
4.2 Jet Energy Correction Jet is reconstructed following the procedure as mentioned above; however, the jet energy we get from reconstruction does not necessarily reflect the precise energy of the particles in jets. This is because of various factors such as; pile up (additional semi-hard interactions from di↵erent proton-proton interactions in the same bunch crossing), underlying events (additional semi-hard interactions in the same proton-proton interaction), nonlinearity nature of detector, and some detector noise.
23 Texas Tech University, Kamal Lamichhane, December 2016
Hence, CMS has a factorized correction scheme with each sub correction for di↵erent detector and physics cause, and each level of correction are responsible for the remedy of di↵erent e↵ect [65-67]. Each level of correction corresponds to the
Figure 4.1. Overview of factorized approach of JEC in CMS implementation of a scale factor i.e. correction to the four momentum of the jet, and the scale factor depends on the di↵erent quantities such as energy o↵set, pt, ⌘, etc. The level of correction follows sequential order as seen in figure 4.1, where the output of one step is the input to the next. The first correction applied on the reconstructed jet is the L1 correction which corresponds to the pile up and detector noise removal, and are done on an event-by-event or jet-by-jet basis. The second correction applied is combination of L2L3 (often individually called as L2 relative, and L3 absolute), is the correction on measured pt the jet response ( actual )tomakeituniformasafunctionof⌘ (L2 relative), and pt pt (L3 absolute). In this step, jet energy is corrected to particle level with most probable value in barrel region ( ⌘ < 1.3), so that the corrected jet p is equal on | | t average to the pt of the jet from simulation (GenJet). L2 corrections are obtained from pt balance in monte-carlo (MC) QCD back to back dijet sample, and L3 is determined from /Z + jet and multijets samples. The final mandatory JEC at CMS is L2L3 Residuals which is used to correct the jet response di↵erence in data and MC.
4.3 Boosted Object and Jet Substructure Heavy objects (mostly beyond standard model prediction such as Extra-Dimenisons of TeV scale) are favored to decay to the standard model electro-weak bosons. These bosons are often boosted (produced with high pt). The
24 Texas Tech University, Kamal Lamichhane, December 2016 comparison of boosted or non-boosted (at rest) is as follows: for non-boosted scenario it is quite simple to resolve jets (from decaying bosons) and calculate the invariant mass but signals are predominantly swamped under background; whereas, for boosted scenario although the cross-section is reduced but the acceptance is very good. Moreover, the implementation of techniques (jet substructure) helps to reduce the background, and QCD contamination quiet well giving much better signal S background ratio ( B ) for boosted objects. Lets put the light on this with the following example; a heavy object (X of mass MX )isproducedfromppcollision, and then decays to a pair of bosons (Y of mass MY ), which further decay to hadrons (jet) or leptons (z) as shown in fig 4.2 [64].
Figure 4.2. Overview of Boosted and resolved scenario from ref [needed]
MX Final states depend on the ratio of mass rM = .IftherM is large, then the Y 2MY will have high pt from the contribution of MX ,henceYareboostedandtheirdecay products (z) appear to be closer to each other. If z is jet then we call this a merged or fat jet i.e. two z are reconstructed with in the same jet cone of bigger radius
(usually AK8). On the other hand, if rM is small then the Y are approximately at rest, and their decay products (z) are cleanly well separated on detector. This scenario is called resolved case, and jets are well reconstructed even with the small jet cone radius (mostly AK4). Anaiveexampleofboostisshowninfigure4.3[68],whereaparticleofmass(m)
25 Texas Tech University, Kamal Lamichhane, December 2016
Figure 4.3. Schematic Representation of Boosted scenario when gets Lorentz boosted ( )decaystodaughterparticlesmakingsomeangle between them. Since R is the measure of the angular separation in (⌘, )plane,so for the boost, R = 2m , and the decay products are contained in a single fat jet. min pt If the daughter particle are quark or gluon jets (QCD), then quark and gluons individually are considered massless, but the energy of mother particle is actually contained in the angle of separation.
4.3.1 Jet Substructure Generally the decay of a boosted objected to two prongs (X 1, 2) substructure ! within a radius R is characterized by R m 1 ,wherez is the energy sharing pt pz(1 z) pt1 2m fraction . The smallest R to have two prongs within a jet is given by Rmin pt ⇠ pt [69-74]. The jet is highly contaminated by the many radiations from the splitting of quarks and gluons giving additional prongs like structure in the jet. Mass of the ↵sCF,A 2 2 QCD jet is given by:
26 Texas Tech University, Kamal Lamichhane, December 2016 the first discriminating variables we generally use in analysis, for example if we are looking for Z (mass 91 GeV), then making the mass cut let’s say 60 110 GeV not ⇠ only helps to reduce the background, but also reduce the possibility of contribution from the fake jets. In general, substructure study needs to toss out the soft wide angle radiation (QCD dominance), measure the energy fraction (z), identify and test for the number of prongs, and these tasks are performed with algorithm [74]. In 2008, jet substructure technique was introduced to study pp ZH(b¯b)by ! Jonathan Butterworth, Adam Davidson, Mathieu Rubin, Gavin Salam which is widely known as BDRS method [70, 74]. Although the substructure method was in literature before than this; but this one turned out to be the more pronounced and handy. Higgs search on b¯b with this method turned out to be very robust and since then BDRS is the fundamentals of jet substructure techniques. BDRS is basically a recursive, angle based algorithm to de-cluster the boosted object to subjets. This in principle works on three steps as follows:
recursively cluster a jet with nearest neighbor combination of four-vectors in • ⌘ plane till the specified R is reached (CA algorithm).
E2 unwind the jet by requiring : a) with E2
If those requirement fails, throw out the softer subjet (known as gromming), • and continue recursion on harder one. If the test succeeds, we found our jet.
In summary, BDRS method is an algorithmic procedure to unfold, unwind, and find the core boosted object like Z or W or H that might be buried underneath the fat jet, and grooming is a technique to improve mass resolution or signal, and also getting rid of the background that is mostly generated by soft radiation. In current days, the BDRS method is being developed in varieties of manipulations, where grooming techniques like filtering, trimming, soft-drop, pruning and the tagging tools like n-subjettiness are very commonly used. In this work, I used the pruning
27 Texas Tech University, Kamal Lamichhane, December 2016 for jet grooming and n-subjettiness for tagging purpose so I will discuss these two in next subsections. Jet grooming means the active removal of soft wide angle radiation. Softness is tested on parton level instead of hadron level as it is sensitive on hadronization. For example, let’s say we have a pion from pile up (PU) or initial state radiation (ISR), and if we apply some energy cut to remove this pion then this applies to the pion from non-PU also. Hence, the softness test is done on parton level being consistent with collinear and IR safety of QCD.
P runing : The main purpose of pruning is to remove the contamination on jet • mainly due to pile up along with underlying events, and initial state radiation (ISR) e↵ect [71-72]. Pruning follows the similar procedure as in BDRS method i.e.
1. Clustering of jet (originally reconstructed from anti k algorithm) with t CA algorithm as in first step of BDRS method explained before. While
clustering at each step we should check if: z