CERNSeptember/October 2021 cerncourier.com COURIERReporting on international high-energy physics WELCOME CERN Courier – digital edition Welcome to the digital edition of the September/October 2021 issue of CERN Courier. ARTIFICIAL

As data volumes surge, deep learning is becoming increasingly important in INTELLIGENCE particle physics. This special edition on artificial intelligence (AI) captures two new trends: using “unsupervised” deep learning to spot anomalous events, and designing AI that can “think not link”. Community-organised data challenges are leading the way (p27) and deep learning could even be used in the level-one triggers of LHC experiments (p31). To keep up with the cutting edge of AI research, physicists are reaching out to computer science and industry (p36): the latest developments could help explore theory space (p51) and build trust in AI to do more of the heavy lifting throughout the analysis chain (p49). We also explore recent thinking that an ordered simplicity may emerge from the complexity of deep learning in a similar way to statistical mechanics and quantum field theory (p39).

Elsewhere in the issue: a tribute to Steven Weinberg (p65); a SciFi upgrade for LHCb (p43); reports from the summer conferences (p19); the most stable tetraquark yet (p7); quantum gravity in the Vatican (p59); anisotropies point to cosmic-ray origins (p11); and much more.

To sign up to the new-issue alert, please visit: http://comms.iop.org/k/iop/cerncourier

To subscribe to the magazine, please visit: https://cerncourier.com/p/about-cern-courier

Emergence Deep learning in the trigger EDITOR: MATTHEW CHALMERS, CERN DIGITAL EDITION CREATED BY IOP PUBLISHING The most stable tetraquark yet

CCSepOct21_Cover_final_v1.indd 1 04/08/2021 11:50 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 Your DIGITAL CERNCOURIER.COM POWER ELECTRONICS Partner IN THIS ISSUE V o l u m e 61 N u m b e r 5 S e p t e m b e r / o c t o b e r 2 02 1

It’s ALL Sparks!/M Rayner iStock/georgeclerk LHCb about the DETAILS

AI leaders Experts weigh in on what’s next Order in complexity A surprising link between SciFi upgrade LHCb prepares for a in the world of artificial intelligence.36 theoretical physics and AI research. 39 luminous future. 43

NeWS people

ANALYSIS ENERGY FRONTIERS FIELD NOTES CAREERS OBITUARIES

Near stable tetraquark Bs decays remain AI centre stage at Quantum gravity Steven Weinberg • SuperKEKB raises bar anomalous • Charm CHEP • FCC Week • LHCP in the Vatican • Felix H Boehm • Science Gateway • DUNE breaks fragmentation • Strange Quark Matter Residents of the Vatican • Anatoly Efremov cryostats • Surveying universality • Unbalanced • Long-lived particles Observatory describe life • Alexei Onuchin a future collider photon pairs • Machine • IPAC • Sustainable as a full-time physicist • Tord Riemann. 65 • Web code auctioned learning and top-quark HEP. 19 in the church. 59 • Cosmic-ray origins. 7 interactions. 15

FeAtureS

DATA CHALLENGES DEEP LEARNING SCIENCE AND INDUSTRY THEORETICAL PHYSICS LHCb UPGRADE What’s in the box? Hunting anomalies Forging the future Emergence Building the future Grassroots initiatives with an AI trigger of AI Erik Verlinde sizes up of LHCb like the LHC Olympics How ‘unsupervised’ The Sparks! forum will the Standard Model, The experiment boasts and Dark Machines learning could probe gather machine-learning gravity and intelligence a brand-new “SciFi” stimulate innovation every LHC event for experts to encourage as candidates for future tracker and upgraded in AI. 27 signs of new physics. 31 innovation in particle explanation as emergent RICH detectors. 43 physics and AI. 36 phenomena. 39

opINIoN DepArtmeNtS

VIEWPOINT INTERVIEW REVIEWS FROM THE EDITOR 5 Power Supply Systems Designing an Stealing theorists’ An enduring and NEWS DIGEST 13 Precision Current Measurements AI physicist lunch adaptable design LETTERS 53 Jesse Thaler argues that Anima Anandkumar and Resistive gaseous APPOINTMENTS 60 particle physicists must John Ellis explore how detectors Climate Beamline Electronic Instrumentation • CMS-OUTREACH-2019-001/M Rayner & AWARDS go beyond deep learning AI could collaborate with change and energy On the cover: How deep learning could help find new RECRUITMENT 61 FMC and MicroTCA and design AI capable of theoretical physicists. 51 options for a sustainable deep thinking. 49 future. 55 physics at the LHC. 31 BACKGROUND 70

www.caenels.com CERN COURIER SEPTEMBER/OCTOBER 2021 3

CCSepOct21_Contents_v3.indd 3 04/08/2021 12:12 MASTER - February 2021_Paolo.indd 1 03/08/21 12:20 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM FROM THE EDITOR

MMS New trends in AI for particle physics

eep learning has transformed the world over the past i n i Pier i/M Sion S THE UNIVERSAL SYSTEM decade. As data volumes surge, it is also growing D in importance in high-energy physics (HEP). This FOR A LARGE RANGE OF LOW AND HIGH VOLTAGE special edition captures two new trends: using “unsupervised” learning to boost creativity in theory building, and designing ar t ificial intelligence (AI) t hat can “t hin k not lin k”. MULTICHANNEL POWER SUPPLIES Most deep learning is “supervised”. Layer upon layer of Mark Rayner artificial neurons operate on data, abstracting electrical signals Deputy editor into particles and jets, and then event topologies, until the data are compressed into a single label: the Standard Model, or whichever flavour of new physics you are searching for. So far, the stack of new-physics labels has never come close to d i s c o v e r y s i g n i fic a n c e, a n d y e t t h e d e fic i e n c i e s i n t h e St andrd Model are there for all to see. What if we’re looking for the wrong new physics? This situation calls for unsupervised learning, where NS? ASK you don’t need to tell the machine what to look for. Just 40 MHz A data challenge to design artificial intelligence for IO E one unsupervised analysis has been published so far, and the level-one trigger has begun. T R S I community-organised data challenges are leading the way E C H U (p27). But supervised triggering algorithms mean that LHC t h at t h i n k s l i ke a p hy sic i s t i f we a re to t r u s t it to d o more a nd

! Q a n a l y s e r s a r e a l r e a d y l o o k i n g a t a s t a c k e d d e c k o f d a t a a d a p t e d more of the heavy lifting in particle physics (p49). to the needs of the most popular theories. To unlock the full But with all its inscrutable complexity, is AI really a power of unsupervised learning, it must be unleashed on the suitable tool for fundamental science? A recent current of W-IE-NE-R + ISEG f u l l 40 MHz data flow in the level-one triggers (p31). Readers opinion (for example, J Halverson et al. 2021 Mach. Learn.: Sci. Technol. 2 035002) asserts that deep learning may actually SUPPORT OFFICE AT CERN w h o w a nt t o g e t i n v ol v e d c a n t a k e p a r t i n a n e w d a t a c h a l l e n g e launched at the ML4Jets conference in July – you have a year exhibit emergent simplicity in a similar way to statistical Dr. Dipl.-Phys. Erich Schaefer to submit your algorithms. m e c h a n i c s a n d q u a nt u m fie l d t h e or y, w h o s e l a w s a r e n o t fun- Bat.: 13/R-023 | +41 75411 4301 | [email protected] damental but emerge from complexity via a process of data We need to Think not link compression. To e x plore t h i s f a sc i n at i ng l i n k to A I resea rc h, MMS MMS design AI that At their core, deep-learning algorithms are association we asked Erik Verlinde to size up the Standard Model, grav- HV LV thinks like a machines, with limited intelligence. Today, AI research is ity and intelligence as candidates for future explanation as compatible compatible physicist if we dominated by the question of how to think rather than link. emergent phenomena (p39). are to trust it to To make sure HEP is part of the conversation, CERN has invited Elsewhere in this issue: a tribute to Steven Weinberg (p65); EBS HV BOARDS HIGH CLASS 50 world leaders on AI to the inaugural Sparks! Serendipity a S c i F i up g r a d e for L HC b (p43); a n i s o t r o pie s p oi nt t o c o s m ic-r a y do more of the CONTROLLER BOARDS WITH LINUX INSIDE Bipolar 4 quadrant high voltage LOW VOLTAGE BOARDS Forum in September (p36). AI initiatives are springing up or i g i n s ( p1 1); t h e m o s t s t a bl e t e t r a q u a r k y e t ( p 7); r e p or t s f r o m heavy lifting Integrated EPICS IOC, SNMP, OPC-UA module with CFG. EBS perfectly Up to 16 channel LV for on site elsewhere too, with Jesse Thaler recently taking the helm at the summer conferences (p19); Hubble tension questioned (p53); in particle the new Institute for Artificial Intelligence and Fundamental quantum gravity in the Vatican (p59); the flavour anomalies and Webservices, also „bare metal“ use. covers requirements of electron frontend hardware supply saves physics Interactions in the US. He argues that we need to design AI stick around (p15); and much more. Now with onboard Python script runtime. optical systems or capacitive loads. cable and installation space.

Reporting on international high-energy physics

CERN Courier is distributed Editor Laboratory correspondents Jefferson Laboratory SLAC National Technical illustrator General distribution to governments, institutes Matthew Chalmers Argonne National Kandice Carter Accelerator Laboratory Alison Tovey Courrier Adressage, CERN, and laboratories affiliated Deputy editor Laboratory JINR Dubna Melinda Baker Advertising sales 1211 G e ne v a 23, S w it z erl a nd; Up to 10 slots and 480 channels per crate with CERN, and to Mark Rayner Tom LeCompte B St a rc hen ko SNOLAB Samantha Kuula Tom Houlden e-mail courrier- individual subscribers. Editorial assistant Brookhaven National KEK National Laboratory TRIUMF Laboratory Recruitment sales adressage@cern.ch It is published six times Craig Edwards Laboratory Achim Franz Hajime Hikino Marcello Pavan Chris Thomas Highest channel density to save valueable installation space per year. The views Archive contributor Cornell University Lawrence Berkeley Advertisement Published by CER N, 1211 expressed are not Peg g ie R i m mer D G Cassel Laboratory Spencer Klein Produced for CERN by production Katie Graham Geneva 23, Switzerland necessarily those of the Astrowatch contributor DESY Laboratory Los Alamos National Lab IOP Publishing Ltd Marketing and Tel +41 (0) 22 767 61 11 Best ripple and noise characteristics CERN management. Merlin Kole Till Mundzeck Raja n Gupt a Temple Circus, Temple circulation Laura Gillham, E-mail Fermilab Kurt NCSL Ken Kingery Way, Bristol BS1 6HG, UK Jessica Pratten Printed by Warners [email protected] Riesselmann Nikhef Robert Fleischer Tel +4 4 (0)117 929 7481 (Midlands) plc, Bourne, Up to 30 kV, up to 100 W per channel Forschungszentrum Orsay Laboratory Advertising Lincolnshire, UK Advisory board Jülich Markus Buescher Anne-Marie Lutz Head of Media Jo A l len Tel +44 (0)117 930 1026 Peter Jen n i, Ch r ist i ne GSI Darmstadt I Peter PSI Laboratory Head of Media (for UK/Europe display © 2021 CER N Pure LV compatible, pure HV compatible and mixed setup crates available Sutton, Claude Amsler, IHEP, Beijing Lijun Guo P-R Kettle Business Development advertising) or +44 (0)117 ISSN 0304-288X Ph i l ippe Bloc h, Roger IHEP, Serpukhov Saclay Laboratory Ed Jost 930 1164 (for recruitment For t y, M i ke L a mont, Yu Ry abov Elisabeth Locci Content and production advertising); e-mail Interchangeable controller boards for networking and software control Joac h i m Kopp INFN Antonella Varaschin UK STFC Ja ne Bi n k s manager Rut h L eopold [email protected]

CERN COURIER SEPTEMBER/OCTOBER 2021 5 WWW.ISEG-HV.COM/MMS

iseg_cerncourier_2021_09_10_dv_210730.indd 1 30.07.21 11:21 CCSepOct21_Editorial_v2.indd 5 04/08/2021 11:13 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM NEWS ANALYSIS

Hadron spectroscopy New tetraquark a whisker away from stability

All the exotic hadrons that have been D Dominguez a colour-charged cc “diquark” tightly observed so far decay rapidly via the bound via gluon exchange to up and down – strong interaction. The ccu–d tetraquark ant iquarks (see “Jumbled toget her” fig- + (Tcc) just discovered by the LHCb collabo- ure). Diquarks are a frequently employed ration is no exception. However, it is the mathematical construct in low-energy longest-lived state yet, and reinforces quantum chromodynamics (QCD): if expectations that its beautiful cousin, two heav y quarks are sufficiently close – bbu–d, will be stable with respect to the together, QCD becomes perturbative, and strong interaction when its peak emerges they may be shown to attract each other in future data. a n d e x h i bit e ffe c t i v e a nt i- c ol o u r c h a r g e. – “We have discovered a ccu–d tetraquark For example, a red–green cc diquark would with a mass just below the D*+D0 thresh- have a wavefunction similar to an anti- old which, according to most models, blue anti-quark, and could pair up with indicates that it is a bound state,” says a blue quark to form a baryon – or, hypo- LHCb analyst Ivan Polyakov (Syracuse thetically, a blue anti-diquark, to form a University). “It still decays to D mesons colour-neutral tetraquark. via the strong interaction, but much less “The question is if the D and D* are intensively than other exotic hadrons.” more or less separated, jumbled together Most of the exotic hadronic states to such a degree that all quarks are inter- discovered in the past 20 years or so are twined in a compact object, or something cc–qq– tetraquarks or cc–qqq pentaquarks, Jumbled together within the observed resonance’s narrow in between,” says Polyakov. “The first + where q represents an up, down or strange Tcc could be a width of 410 ± 165 keV, prescribed by the scenario resembles a relatively large + quark. A year ago LHCb also discovered a compact cc diquark uncertainty principle. The Tcc tetraquark ~4 fm deuteron, whereas the second can hidden-double-charm cc–cc– tetraquark, tightly bound by can therefore decay via the strong inter- be compared to a relatively compact ~2 fm X(6900) (CERN Courier September/ gluons to much action, but strikingly slowly. By contrast, alpha particle.” October 2020 p25), and two open-charm lighter up and most exotic hadronic states have widths The new T+ tetraquark is an enticing – cc csu–d tetraquarks, X (2900) and X (2900). down antiquarks. from tens to several hundreds of MeV. target for further study. Its narrow decay – 0 1 The new ccu–d state, announced at the “Such closeness to the threshold is not into a D0D0π+ fi n a l s t a t e – t h e v i r t u alD *+ European Physical Society conference on very common in heavy-hadron spectros- decays promptly into D0π+ – includes no high-energy physics in July to have been copy,” says analyst Vanya Belyaev (Kur- particles that are difficult to detect, lead- o b s e r v e d w it h a s i g n i fic a n c e s u b s t a nt i a l l y chatov Institute/ITEP). “Until now, the ing to a better precision on its mass than in excess of five standard deviations, is the only similar closeness was observed for for existing measurements of charmed fi r s t e x o t i c h a d r o n i c s t a t e w it h s o - c a led the enigmatic χ (3872) state, whose mass baryons. This, in turn, can provide a c1 – d o u bl e o p e n h e a v y fl a v o u r – i n t h i s c a se, coincides with the D*0D0 threshold with a stringent test for existing theoretical two charm quarks unaccompanied by precision of about 120 keV.” As it is wider, models and could potentially probe ant ipar t icles of t he same flavour. however, it is not yet known whether pr e v i o u s l y u n r e a c h a bl e Q C D e ffe c t s, s a y s

the χc1(3872) is below or above threshold the team. And, if detected, its beautiful Prime candidate (CERN Courier July/August 2020 p18). cousin would be an even bigger boon. + Tetraquark states with two heavy quarks “The surprising proximity of Tcc and “Observing a tightly bound exotic had- * and two light antiquarks have been χc1(3872) to the D D thresholds must have ron that would be stable with respect to the prime candidates for stable exotic deep reasoning,” adds analyst Mikhail the strong interaction would be a cor- hadronic states since the 1980s. LHCb’s Mikhasenko (ORIGINS, Munich). “I am nerstone in understanding QCD at the discovery, four years ago, of the Ξ++ (ccu) fascinated by the idea that, roughly scale of hadrons,” says Polyakov. “The cc – baryon allowed QCD phenomenologists speaking, a strong coupling to a decay bbu–d, which is believed to satisfy this t o fi r m l y pr e d ic t t h e e x i s t e n c e o f a s table channel might attract the bare mass of the requirement, is produced rarely and is out – bbu–d tetraquark, however the stability of hadron. Tremendous progress in lattice of reach of the current luminosity of the – a potential ccu–d state remained unclear. QCD over the past 10 years gives us hope LHC. However, it may become accessible – Predictions of the mass of the ccu–d state that we will discover the answer soon.” in LHC Run 3 or at the High-Luminosity varied substantially, from 250 MeV The cause of this attraction, says LHC.” In the meantime, there is no short- below to 200 MeV above the D*+D0 mass Mikhasenko, could be linked to a age of work in hadron spectroscopy, jokes threshold, say the team. Astoundingly, “quantum admixture” of two models B e l y a e v. “We d e fi n it e l y h a v e m or e p e a ks its observation by LHCb reveals that it that vie to explain the structure of the than researchers!” is a mere 273 ± 61 keV below the thresh- new tetraquark: it could be a D*+ and a D0 old – a bound state, then, but with the meson, bound by the exchange of colour- Further reading threshold for strong decays to D*+D0 lying neutral objects such as light mesons, or LHCb Collab. 2021 LHCb-PAPER-2021-031.

CERN COURIER SEPTEMBER/OCTOBER 2021 7

CCSepOct21_NewsAnalysis_v2.indd 7 03/08/2021 12:22 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

NEWS ANALYSIS NEWS ANALYSIS

accElEr ator physics NeutriNo detectors SuperKEKB raises the bar CERN to provide two DUNE cryostats KEK On 22 June, the SuperKEKB accelerator at uct of beam currents 3.5 times smaller The Deep Underground Neutrino N-201710-248-1 ice/CER Br M is only one aspect of the ProtoDUNE the KEK laboratory in Tsukuba, Japan set than were used at KEKB when its record Experiment (DUNE) in the US is set to project: instrumentation, readout, a new world record for peak luminosity, ] luminosity was achieved. replicate that marvel of model-making, –1 3 high-voltage supply and many other reaching 3.1 × 1034 cm–2 s–1 in the Belle II s SuperKEKB/Belle II is also reaching the ship-in-a-bottle, on an impressive aspects of detector design have been –2 detector. Until last year, the luminosity super-B-factory-level performance scale. More than 3000 tonnes of steel o p t i m i s e d t h r o u g h m or e t h a n fi v e y e a r s record stood at 2.1 × 1034 cm–2 s–1, shared cm in integrated luminosity, achieving and other components for DUNE’s four of R&D. Two technologies were trialled 34 2 by the former KEKB accelerator and the the highest values collected in a day giant detector modules, or cryostats, at the Neutrino Platform: single- and LHC. In the summer of 2020, however, (1.96 fb–1), in a week (12 fb–1) and in a must be lowered 1.5 km through narrow dual-phase LAr TPCs. The single-phase SuperKEKB/Belle II surpassed this month (40 fb–1). These are about 40% shafts beneath the Sanford Lab in South design has been selected as the design 1 value with a peak luminosity of higher than the old records of KEKB/ Dakota, before being assembled into for t h e fi r s t f u l l-s i z e DU N E m o d u l e. The 2.4 × 1034 cm–2 s–1 (CERN Courier September/ Belle and about twice the level of SLAC’s four 66 × 19 × 18 m3 containers. And Neutrino Platform team is now qualify- luminosity [×10 October 2020 p13). PEP-II/BaBar, which completed opera- the maritime theme is more than a ing a hybrid single/dual-phase version In physics operation since 2019, tions more than a decade ago. metaphor: to realise DUNE’s massive based on a vertical drift, which may prove SuperKEKB is an innovative nano- 6/15 6/20 6/25 6/30 “Despite the challenges brought by cryostats, each of which will keep to be simpler, more cost effective and beam, asymmetric-energy accelerator the COVID-19 pandemic and neces- 17.5 kt of liquid argon (LAr) at a temper- easier-to-install. complex that collides 7 GeV electrons Record rates Instantaneous luminosities recorded in the Belle II sary social-distancing protocols, the ature of –200°, CERN is working closely with 4 GeV positrons, sitting mostly detector in June. SuperKEKB team is making impressive with the liquefied natural gas (LNG) Step change on or near the ϒ(4S) resonance (CERN progress towards an eventual target shipping industry. In parallel with efforts towards the Courier September 2016 p32). It uses a The team is making impressive luminosity of 6.5 × 1035 cm–2 s–1,” says Since it was established in 2013, US neutrino programme, CERN has large crossing angle and strong focusing Belle II physicist Tom Browder of the CERN’s Neutrino Platform has enabled developed the BabyMIND magnetic * at the interaction point (βy = 1 mm), and progress towards a target University of Hawai’i. “The improving significant European participation in spectrometer, which sandwiches mag- has implemented a crab-waist scheme 35 –2 –1 performance of SuperKEKB should ena- long-baseline neutrino experiments netised iron and scintillator to detect to stabilise beam–beam blowup using luminosity of 6.5 × 10 cm s ble Belle II to collect a large data sample in the US and Japan. For DUNE, which Single phase Inside a prototype liquid-argon time-projection relatively low-energy muon neutrinos, carefully tuned sextupole magnets on to clarify the intriguing and potentially will beam neutrinos 1300 km through chamber for the DUNE experiment. and participates in the T2K experi- either side of the interaction point. These rather modest beam currents: 0.8 A in ground-breaking anomalies in the the Earth’s crust from Fermilab to ment, which sends neutrinos 295 km innovations have enabled the SuperKEKB the low-energy positron ring and 0.7 A in fl a v o u r s e c t or, c o n s t r a i n t h e d a r k s e c tor, Sanford, CERN has built and operated ICARUS, which was shipped from Japan’s J-PARC accelerator facil- team to attain record luminosities with the high-energy electron ring – a prod- and search for new physics.” two large-scale prototypes for DUNE’s ity to the Super-Kamiokande detector. LAr time-projection chambers (TPCs). to Fermilab in 2017, is a CERN will contribute to the upgrade of Education and outrEach All aspects of the detectors have been T2K’s near detector, and a proposal has M Brice/CERN mere twentieth of the size of Science Gateway validated. The “ProtoDUNE” detectors’ been made for a new water Cherenkov cryostats will now pave the way for the a single DUNE module test-beam experiment at CERN, to later under construction Neutrino Platform team to design and be placed about 1 km from the neutrino engineer cryostats that are 20 times beam source of the Hyper Kamiokande bigger. CERN had already committed to experiment . Excavation of underground On 21 June, officials and journalists build the first of these giant modules. In would be used. But at the scale of DUNE, caverns for Hyper Kamiokande and DUNE gathered at CERN to mark “first stone” June, following approval from the CERN the cryostats will deform by tens of cm has already begun (see p13). for Science Gateway, CERN’s new flag- Council, the organisation also agreed to when cooled from room temperature, DUNE and Hyper-Kamiokande, ship project for science education and provide a second. potentially imperilling the integrity of along with short-baseline experiments outreach. Due to open in 2023, Science instrumentation, and leading CERN to and major non-accelerator detectors Gateway will increase CERN’s capacity Scaling up use an active foam with an ingenious such as JUNO in China, will enable to welcome visitors of all ages from near Weighing more than 70,000 tonnes, membrane design. high-precision neutrino-oscillation and afar. Hundreds of thousands of peo- DUNE will be the largest ever deploy- “The nice idea from the LNG industry measurements to tackle questions such ple per year will have the opportunity ment of LAr technology, which serves is that they have found a way to have an as leptonic CP violation, the neutrino to engage with CERN’s discoveries and as both target and tracker for neutrino internal membrane, which can deform mass hierarchy, and hints of additional technology, guided by the people who interactions, and was proposed by Carlo like a spring, as a function of the thermal “sterile” neutrinos, as well as a slew of make it possible. Rubbia in 1977. The first large-scale conditions. It’s a really beautiful thing,” questions in multi-messenger astron- The project has environmental sus- LAr TPC – ICARUS, which was refur- says Nessi. “We are collaborating with omy. Entering operation towards the end tainability at its core. Designed by bished at CERN and shipped to Fermi- French LNG firm GTT because there is of the decade, Hyper-Kamiokande and renowned architect Renzo Piano, the lab’s short-baseline neutrino facility in a reciprocal interest for them to opti- DUNE will mark a step-change in the carbon-neutral building will bridge the 2017 – is a mere twentieth of the size of mise the process. They never went to scale of neutrino experiments, demand- Route de Meyrin and be surrounded by a a single DUNE module. LAr temperatures like these, so we are ing a global approach. freshly planted 400-tree forest. Its five Scaling LAr technology to industrial both learning from each other and have “The Neutrino Platform has become linked pavilions will feature a 900-seat First stone (from left) Architect Renzo Piano, CERN Director-General Fabiola Gianotti, Geneva state councillor levels presents several challenges, built a fruitful ongoing collaboration.” one of the key projects at CERN after auditorium, immersive spaces, labora- Antonio Hodgers and chairman of Stellantis and the FCA Foundation (the main donor) John Elkann with the explains Marzio Nessi, who leads CERN’s Having passed all internal reviews at the LHC,” says Nessi. “The whole thing tories for hands-on activities for visitors symbolic foundation stone, adorned with the newly unveiled Science Gateway logo. Neutrino Platform. Typical cryostats CERN and in the US, the first cryostat is a wonderful example – even a proto- from age five upwards, and many other are carved from big chunks of welded is now ready for procurement. Several type – for the global participation interactive learning opportunities. and beyond who are making the CERN her opening speech. “We want the CERN steel, which does not lend itself to a d i ffe r e nt i n d u s t r i e s a c r o s s C E R N’s me- and international collaboration that “I would like to express my deepest Science Gateway possible, in particu- Science Gateway to inspire all those who modular design. Insulation is another ber states and beyond are involved, with will be essential as the field strives to gratitude to the many partners in our lar to our generous donors,” said CERN come to visit with the beauty and the c ha l lenge. In sma l ler set ups, a v ac uum delivery and installation at Sanford Lab build ever more ambitious projects like Member and Associate Member States Director-General Fabiola Gianotti during values of science.” installation comprising two stiff walls expected to start in 2024. The cryostat a future collider.”

8 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 9

CCSepOct21_NewsAnalysis_v2.indd 8 03/08/2021 12:22 CCSepOct21_NewsAnalysis_v2.indd 9 03/08/2021 12:25 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

NEWS ANALYSIS NEWS ANALYSIS

Future CirCular Collider AstrowAtch Surveyors eye up a future collider Anisotropies point to cosmic-ray origins J Ordan/CERN C E R N s u r v e y or s h a v e p e r for m e d t h e fi r s t Such measurements have two main Spanning 13 decades in energy and more Pérez ESO/J geodetic measurements for a possible purposes. The first is to determine than 26 decades in intensity, cosmic rays Future Circular Collider (FCC), a pre- a high-precision surface model, or are one of the hottest topics in astropar- requisite for high-precision alignment “geoid”, to map the height above sea ticle physics today. Spectral features of the accelerator’s components. The level in the FCC region. The second pur- such as a “knee” at a few PeV and an millimetre-precision measurements pose is to improve the present reference “ankle” at a few EeV give insights into a r e o n e o f t h e fi r s t a c t i v it ie s u n d e r t a k en system, whose measurements date back their varying origins, but studies of their by the FCC feasibility study, which was to the 1980s when the tunnel housing arrival direction can also provide valua- launched last year following the rec- the LHC was built. ble information. Though magnetic fields ommendation of the 2020 update of the “The results will help to evaluate if an mean we cannot normally trace cosmic European strategy for particle physics extrapolation of the current LHC geo- rays directly back to their point of origin, (p20). During the next three years, the detic reference systems and infrastruc- angular anisotropies provide important study will explore the technical and ture is precise enough, or if a new design independent evidence towards probable fi n a n c i a l v i a bi l it y o f a100 km collider at is needed over the whole FCC area,” says sources at different energies. In July, at the CERN, for which the tunnel is a top pri- Hélène Mainaud Durand, group leader 37 t h I nt e r n a t io n a l C o s m ic R a y C o n fe r e n c e ority. Geology, topography and surface of CERN’s geodetic metrology group. (ICRC), a range of space- and ground- infrastructure are the key constraints on The FCC feasibility study, which based experiments greatly increased our the FCC tunnel’s position, around which involves more than 140 universities and knowledge of cosmic-ray anisotropies, Spectral Nearby supernova remnants such as Vela could explain sub-TeV spectral features of cosmic rays. civil engineers will design the optimal research institutions from 34 coun- with new results spanning 10 decades in route, should the project be approved. tries, also comprises technological, energy, from GeV to tens of EeV. NASA/ESA Array in the US. These two observatories, The FCC would cover an area about 10 environmental, engineering, political At sub-TeV energies, spectral features which observe different hemispheres, times larger than the LHC, in which every and economic considerations. It is due seen by the AMS-02 and CALET detectors find st rong ev idence for e xcesses in t he geographical reference must be pin- to be completed by the time the next on the International Space Station and c o s m ic-r a y flu x i n c e r t a i n r e g io n s o f the pointed with unprecedented precision. Level headed CERN’s Pierre Valentin performing levelling strategy update gets under way in the t h e C h i n e s e–E u ro p e a n DA M PE s at e l l it e sky at energies exceeding EeV. At ener- To provide a reference coordinate system, measurements between Certoux and Saint-Julien in France. middle of the decade. Should the out- could potentially be explained by a local gies as high as these, cosmic rays point in May the CERN surveyors, in conjunc- come be positive, and the project receive galactic source such as a supernova rem- more clearly to their origin, and galac- t i o n w it h E T H Zü r i c h, t h e Fe d e r a l O ffic e performed geodetic levelling measure- the approval of CERN’s member states, n a nt l i k e Ve l a (s e e “S p e c t r a l” fi g u r e). I f a tic cosmic rays should have very clear of Topography Swisstopo, and the School ments along an 8 km profile across the civil-engineering works could start as nearby source is indeed responsible for point-like sources that are not observed, of Engineering and Management Vaud, Swiss–French border south of Geneva. early as the 2030s. a significant fraction of the cosmic rays providing evidence that they originate observed at such energies, it could show outside of our galaxy. A prime candidate World Wide Web up in the arrival direction of these cos- for such sources are so-called starburst mic rays in the form of a dipole feature, galaxies, wherein star formation happens Web code auctioned as crypto asset despite bending by galactic magnetic unusually rapidly, during a short period fields; however, results from AMS-02 of the galaxy’s evolution (see “Anten- Time-stamped files stated by Tim ERC-721 NFT 70 years ago, in the CERN Convention, at ICRC showed no evidence of a dipole nae galaxies” figure). As presented at Berners-Lee to contain the original which states that results of its work were in the arrival direction of protons or any ICRC 2021, the available data was fitted source code for the web and digitally to be “published or otherwise made gen- other light nucleus. This was confirmed to models where starburst galaxies are signed by him, have sold for US$5.4 mil- erally available” – a culture of openness by DAMPE, which excluded dipole fea- the primary source of EeV cosmic rays. lion at auction. The files were sold as a that continues to this day. tures with amplitudes above about 0.1% The mo del fit s t he a n i sot ropy dat a w it h non-fungible token (NFT), a form of a The auction offer describes the NFT in the 100s of GeV energy range. The more than 4σ s i g n i fic a n c e r e l a t i v e t o t he c r y pto a ss e t t h at us e s blo c kc h a i n te c h- as containing approximately 9555 lines search continues, however, with DAMPE, Antennae c o n fi r m a s w it c h i n t h e p h a s e o f t heani- null hypothesis with normal galaxies, nology to confer uniqueness. of code, including implementations of A M S-02 a n d C A L E T a l l s e t t o t a k e f u r t h e r galaxies sotropy when moving from 100s of TeV to indicating starburst galaxies to likely The web was originally conceived at the three languages and protocols that data over the coming years. Starburst galaxies P eV e n e r g i e s, a s r e p or t e d b y Ic e C u b e a n d be at least one source of EeV cosmic rays. CERN to meet the demand for automated remain fundamental to the web today: Moving to higher energies, clear such as that other experiments in recent years: at PeV W h i l e s o m e o f t h e fe a t u r e s w i l l l i k e l y b e information-sharing between physicists HTML (Hypertext Markup Language), anisotropic dipole excesses have been resulting from energies, close to the knee, the dipole has f u l l y c o n fi r m e d w it h i n t h e c o m i n g y ears spread across universities and institutes HTTP (Hypertext Transfer Protocol) and observed over the last decade by ground- the collision of a maximum rather than a minimum close simply by accumulating statistics, new worldwide. Berners-Lee wrote his first U R I s (Un i for m R e s o u r c e Id e nt i fie r s). T h e based experiments such as the ARGO-YBJ NGC 4038 and to the galactic centre. This could indicate fe a t u r e s a r e a l s o l i k e l y t o a r i s e. O n e e x a m- project proposal in March 1989, and the lot also includes an animated visual- obser vator y in China, t he HAWC obser- NGC 4039 could an excess of “pevatron” sources near the ple is further constraints on the lack of any first website, which was dedicated to the isation of the code, a letter written by v a t or y i n M e x i c o a n d t h e Ic e C u b e o b s e r- explain cosmic-ray galactic centre (CERN Courier July/August observed anisotropy at sub-TeV energies World Wide Web project itself and hosted B e r n e r s-L e e r e fle c t i n g o n t h e pr o c e s sof vatory at the South Pole – though with anisotropies 2021 p11). using data from space-based missions, on Berners-Lee’s NeXT computer, went creating it, and a Scalable Vector Graphics d i ffe r e nt “p h a s e s” a t d i ffe r e nt e n e r g ies. at energies while new data from ground-based exper- live in the summer of 1991. Less than two representation of the full code created The anisotropy in the TeV to the 100s of exceeding EeV. Extragalactic sources iments will start to bridge the measure- years later, on 30 April 1993, and after from t he orig inal files. TeV energy range could point towards a While results up to PeV energies give an ment gap between PeV and EeV energies. several iterations in development, CERN Bidding for the NFT, which auction- nearby source, though models proposing insight into sources within our galaxy, The latter will be especially important in placed version three of the software in house Sotheby’s claims is its first- the structure of the interstellar magnetic it is theorised that the flux starts to be gaining an understanding of the energy the public domain. It deliberately did so ever sale of a digital-born artefact, field as the true origin for the aniso- dominated by extragalactic sources scale at which extragalactic sources start on a royalty-free, “no-strings-attached” opened on 23 June and attracted a tropy also exist. This feature was fur- somewhere between the knee and the to dominate. To fully exploit the data it basis, addressing the memo simply total of 51 bids. The sale will benefit t h e r c o n fi r m e d t h i s y e a r b y t h e L HASO ankle of the cosmic-ray spectrum. Evi- will be necessary to compare complex cos- “To whom it may concern.” initiatives that Berners-Lee and his e x p e r i m e nt i n C h i n a, u s i n g a y e a r o f d a t a dence for this was increased by new mic-ray-propagation simulations with The seed that led CERN to relinquish wife Rosemary Leith support, stated a that was taken while constructing the results from the Pierre Auger Obser- diverse data such as the pevatron sources ownership of the web was planted Unique A portion of the web’s original source code. Sotheby’s press release. detector. The results from LHAASO also vatory in Argentina and the Telescope discovered t his year by LH A ASO.

10 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 11

CCSepOct21_NewsAnalysis_v2.indd 10 03/08/2021 12:25 CCSepOct21_NewsAnalysis_v2.indd 11 03/08/2021 12:26 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM NEWS DIGEST

pro b e d b y s c ie nt i s t s: P b–P b he a v y- excavation for DUNE are being of fusion power by maintaining

M Zacher ion collisions at the LHC (CERN lowered a mile underground a plasma in a sel f-sustaining Courier January/February 2020 in Lead, South Dakota. Both “ignition” phase. Machine p17). Scanning MoEDAL detectors experiments are due to begin assembly began in July 2020, with a SQUID magnetometer taking data in 2027, and will w it h t he goal of ac hiev ing “first excluded monopoles with one, seek to constrain leptonic CP plasma” by late 2025.

two or three Dirac charges up to violation and study astrophysical ITER US masses of 75 G eV. neutrinos (CERN Courier July/ August 2020 p32). Inside the KATRIN spectrometer. ILC pre-lab proposal published In June, the International Instrumentation training probed Neutrino mass less than 1 eV Development Team (IDT) A survey by the European The Karlsruhe Tritium Neutrino for the International Linear Committee for Future (KATRIN) experiment has Collider (ILC) proposed to Accelerators (ECFA) reports announced t he first model- launch a preparatory laboratory that 60% of early-career independent limit on the mass of (pre-lab) to bring the project researchers are eager to take neutrinos with sensitivity below to a point of readiness for on more instrumentation work, 1 eV. The best fit to t he spect ral construction, should the project however, only 15% felt informed data yields a squared neutrino gain international support about training opportunities. The central solenoid (blue) at the mass of 0.26 ± 0.34 eV 2 for the (arXiv:2106.00602). The IDT A quarter feels they have had heart of ITER’s tokamak design. second physics run, thereby will now facilitate a discussion insufficient t raining for t he excluding neutrino masses above among candidate founding instrumentation work that Second superconductor supplier 0.8 eV at 90% confidence when laboratories to decide whether they have undertaken, and a CERN has validated the acceptance

all data is taken into account to proceed with the pre-lab. majorit y believes t hat t here testing of Nb3Sn cables from (arXiv:2105.08533). KATRIN Over a period of approximately should be a greater emphasis Russian company TVEL as part probes the mass of electron four years, the pre-lab would on hands-on instruction. of the conductor-development anti-neutrinos by measuring invite international ideas for Respondents were almost programme for the Future the β-decay spectrum of tritium experiments, facilitate R&D for unanimous in expressing Circular Collider (p20). TVEL is the close to its endpoint at 18.6 keV. superconducting radio-frequency the benefits of peer-to-peer second manufacturer to comply The collaboration targets an cavities and engineering mentoring, and many believe w it h CERN’s specificat ions, w it h ultimate sensitivity of 0.2 eV at studies for the whole accelerator that the quality of training development also ongoing in the 90% confidence by t he end of complex, and perform geological is strongly correlated to both US, Europe, South Korea, Japan 2024, close to complementary surveys at the candidate site in institution and country of and China. model-dependent limits based the Kitakami Mountains in Japan, residence (arXiv:2107.05739). on searc hes for neut rinoless- says the team. The results will inform ECFA’s Galactic geology double-beta decay and detector-R&D roadmap. A team from Stanford and measurements of temperature Hyper-K and DUNE get cracking Stockholm propose using anisotropies of the cosmic Next-generation neutrino Most powerful magnet sets sail “paleo-detectors” (PDs) to microwave background. experiments Hyper- Following testing in the US, the search for dark matter (DM) by Kamiokande (Hy per-K) and t he world’s most powerful magnet observing so-called damage Search for Schwinger monopoles Deep Underground Neutrino is being shipped to international tracks caused by nuclear recoils In a first for an experiment at a Experiment (DUNE) have begun fusion laboratory ITER in in small samples of natural particle collider, the Monopole excavation for their underground southern France. The central minerals (arXiv:2107.02812). and Exotics Detector at the detector caverns. On 28 May, solenoid (CS) stands 18 m tall and Unlike real-time direct-detection LHC (MoEDAL) collaboration 4 m wide, with a maximum field experiments, PDs might have been PUSHING SCIENCE TO ITS LIMITS searched for magnetic monopoles strength of 13 T, and stores 6.4 GJ accumulating tracks for up to a

produced by the Schwinger effect U. Tokyo of field energ y – t hree t imes as billion years, suggesting a unique (arXiv:2106.11933). Following up much as the CMS experiment’s possibility, claim the authors: by an idea from Fritz Sauter in the 14 kton solenoid, which previously reading out PDs of different ages, 1930s, Julian Schwinger showed held the record. The CS uses one ca n e x plore t he t ime-v a r iat ion Sn superconductor Cosylab is an international academic spin-off that builds and integrates state-of-the in 1951 that pairs of particles 43 km of Nb3 of signals on megayear to gigayear art control system solutions for the world’s most complex, precise and advanced with opposite electrical charge manufactured in Japan – the timescales. The team discuss two can be spontaneously created in same material that will be used examples of DM substructure that systems. By working elbow to elbow with some of the best physicists and laboratories a strong electric field. MoEDAL Groundbreaking ritual. for t he High-Luminosit y LHC’s could give rise to time-varying around the world, Cosylab has established great partnerships in the scientific is investigating whether the eight new quadrupole triplets signals: a DM disk through which community and is an active member of the largest international physics projects dual effect might reveal long- Hy per-K broke g round in Hida (CERN Courier January/February the Earth passes every ~45 Myr, hypothesised magnetic charges. City, Gifu Prefecture, Japan. 2021 p8). Formally established and a DM “subhalo” that the For more information, please visit cosylab.com or The collaboration exploited the Meanwhile in the US, 35 pieces of in 2007, ITER is tasked with Earth encountered during the see our career opportunities on cosylab.com/jobs strongest known magnetic fields equipment that will carry out the demonstrating the feasibility past gigayear.

CERN COURIER SEPTEMBER/OCTOBER 2021 13

CCSepOct21_News digest_v2.indd 13 03/08/2021 12:30 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM ENERGY FRONTIERS

Reports from the Large Hadron Collider experiments LHCb

Bs decays remain anomalous The LHCb experiment recently presented range. If lattice-QCD calculations are ×10–8 new results on the b → sμμ d e c a y o f a B s not used in the comparison, increased meson to a φ m e s o n a n d a d i m u o n p a i r, LHCb LHCb 9 fb–1 theory errors reduce the tension to 2 r e i n for c i n g a n a n o m a l y l a s t r e p or t e d i n LHCb 3 fb–1 1.8 standard deviations in the low-q ] 2015 w it h i m pr o v e d s t a t i s t i c s a n d t h e or y –2 12 SM (LCSR+lattice) region. The previous 2015 measure- calculations. Such decays of b hadrons SM (LCSR) m e nt b y L HC b, w h i c h w a s b a s e d e x c lu- [GeV v i a b → s q u a r k t r a n s it i o n s a r e s t r o n g l y 2 SM (lattice) sively on Run-1 data (grey data points), s u p pr e s s e d i n t h e St a n d a r d M o d e l (S M) 8 was reported at the time to be approxi- )/dq –

μ φ J/ψψ(2S) and therefore constitute sensitive + mately three standard deviations below probes for hy pothetical new particles. μ the best theoretical predictions that →φ S I n r e c e nt y e a r s, s e v e r a l m e a s u r e m e nt s 0 4 were available at the time. Since then, of rare semileptonic b → sℓℓ decays theo retical calculations have gener-

have shown tensions with SM predic- dB(B ally become more precise with regard tions. Anomalies have been spotted in to form factors, but more conservatively 0 m e a s u r e m e nt s o f br a n c h i n g f r a c t io n s, evaluated with regard to non-local angular analyses and tests of lepton 0 5 10 15 hadronic effects. 2 2 flavour universality (LFU), leading to q [GeV ] The angular distribution of the

cautious excitement that new physics Bs → φμμ decay products offers com-

might be at play. Fig. 1. Differential branching fractions for the decay Bs → φμμ, plementar y information. At the inter- At t he SM@LHC con ference in Apr i l, compared with SM predictions based on light-cone sum rules national FPCP conference in June, LHCb LHCb presented the most precise deter- (LCSR) and lattice-QCD calculations. pr e s e nt e d a m e a s u r e m e nt o f t h e a n g u l a r m i n a t io n t o d a t e o f t h e br a n c h i n g f r a c- distribution of these decays in different tion for the decay using data collected q2 regions using data collected during during LHC Run 1 and Run 2 (figure 1). 1.0 LHC Run 1 and Run 2. Figure 2 shows LHCb LHCb 8.4 fb–1 The branching fraction is measured the longitudinal polarisation fraction –1 as a function of the dimuon invariant 0.8 LHCb 3 fb FL – one of several variables sensitive m a s s (q 2) a n d fo u n d t o l i e b e l o w t h e S M to anomalous b → sμμ couplings. The SM (LCSR+lattice) prediction at the level of 3.6 standard result s a re consistent w it h SM pred ic- 0.6 deviations in the low-q2 region. This t i o n s a t t h e l e v e l o f t w o s t a n d a r d d e v i- L

defic it of muons is consistent w it h t he F ations, but may also hint at the same pattern seen in LFU tests of b → sℓℓ 0.4 pattern of unexpected behaviour seen in transitions (CERN Courier May/June a n g u l a r a n a l y s e s o f o t h e r b → sμμ d e c a y s φ J/ψψ(2S) 2021 p17), however calculating the SM 0.2 (CERN Courier May/June 2020 p10) and prediction for the B s → φμμ branching in branching-fraction measurements. fraction is more challenging than for For both analyses, LHC Run 3 will LFU test s as it i nvolves t he ca lc ulat ion 0 be crucial to better understanding the of non-per t urbat ive hadronic effects. 0510 15 anomalous behaviour seen so far in 2 2 Calculations based on light-cone sum q [GeV ] Bs → φμμ decays. rules are most precise at low q2, while

lattice-QCD calculations do better at Fig. 2. The longitudinal polarisation fraction FL for the decay Further reading 2 high q . A combination is expected to Bs → φμμ, compared with SM predictions based on light-cone LHCb Collab. 2021 arXiv:2105.14007. give the best precision over the full q2 sum rules (LCSR) and lattice-QCD calculations. LHCb Collab. 2021 arXiv:2107.13428.

ALICE Charm breaks fragmentation universality

The study of heavy-flavour hadron pro- volution of the parton distribution p e r for m e d i n e +e– or e p c ol l i s i o n s, u n d e r duction in proton–proton (pp) collisions functions (PDFs) of the incoming pro- the assumption that the hadronisation provides an important test for quantum t o n s, t h e p a r t o n i c c r o s s s e c t i o n a n d t h e Universal o f c h a r m q u a r k s i nt o c h a r m h a d r o n s i s chromodynamics (QCD) calculations. frag mentation functions t hat describe charm- a u n i v e r s a l pr o c e s s t h a t i s i n d e p e n d e nt Heavy-flavour hadron production is the transition from charm quarks into to-hadron of t he colliding systems. usually computed with perturbative- charm hadrons. The latter are typically fragmentation T h e l a r g e d a t a s a m pl e s c ol l e c t e d d u r- s QCD (pQCD) calculations as the con- parametrised from measurements may not be valid ing Run 2 of the LHC at √s = 5.02 TeV

CERN COURIER SEPTEMBER/OCTOBER 2021 15 www.karlhugo.com

CCSepOct21_EnergyFrontiers_v3.indd 15 03/08/2021 12:41 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

ENERGY FRONTIERS ENERGY FRONTIERS

allowed the ALICE collaboration to The cc– production cross section per picture without the strong interaction, A surprising Extending calculations to encom- of arbitrarily many-quark and gluon 1.0 cc– measure the vast majority of charm unit of rapidity at midrapidity (dσ / the momentum of each photon should role of the pass next-to-next-to-leading order emissions – especially relevant at low ) was calculated by summing quarks produced in the pp collisions ALICE, pp, √s = 5.02 TeV dy||y| < 0.5 perfectly balance in the transverse strong corrections in the strong-interac- energ ies – properly desc r ibe t he ent i re by reconstructing the decays of the B factories, e+e–, √s = 10.5 GeV the cross sections of all measured pl a ne. Howe ver, t h i s si mpl i st ic e x p e c- tion coupling constant (purple line), measured distribution. 0 + interaction ground-state charm hadrons, measur- ground-state charm hadrons (D , D , tation does not match the measurements the impact of the strong interaction T h e r e s u l t s o f t h i s a n a l y s i s, w h i c h a l s o 0.8 + – LEP, e e , √s = mz + + 0 in electro- ing all the charm-meson species and Ds, Λc, and Ξc). The contribution of the (s e e fi g u r e 1). M e a s u r i n g t h e d i ffe r e nt ial becomes manifest. The measured val- i n c lu d e m e a s u r e m e nt s o f o t h e r d i s t r i b u- + HERA, ep, DIS 0 the most abundant charm baryons (Λc, Ξ c w as mult iplied by a factor of t wo, in c r o s s-s e c t io n a s a f u n c t io n o f t h e t r a n s- magnetic u e s a t h i g h t r a n s v e r s e m o m e nt a a r e w e l l tions such as angular variables between 0,+ and Ξc ) down to very low transverse HERA, ep, PHP or d e r t o a c c o u nt for t h e c o nt r i b ut i o n o f verse momentum of the photon pair, diphoton described by these predictions, includ- the two photons, don’t just viscerally 0.6 +

) – momenta. c the Ξc . The resulting cc cross section AT L A S fi n d s t h a t m o s t o f t h e m e a sured production i n g t h e b u m p o b s e r v e d a t 70 GeV, which probe the strong interaction – they also Charm fragmentation fractions, H per unit of rapidity at midrapidity is photon pairs (black points) have low but is another manifestation of higher- pr o v i d e a b e n c h m a r k for t h i s i m p or t a nt → cc– +134 is revealed

f(c → Hc), represent the probability for dσ /dy||y| < 0.5 = 1165 ± 44(stat) –101 (s y s t) μb. non-zero transverse momenta, with a order strong-interaction effects. Monte background process. f(c a charm quark to hadronise into a given 0.4 T h i s m e a s u r e m e nt w a s o b t a i n e d for t h e p ea k at approx i mately 10 G eV, fol lowe d Carlo event generators like Sherpa (red charm hadron. These have now been first time in hadronic collisions at the by a smoothly falling distribution line), which combine similar calcula- Further reading measured for the first time at the LHC L HC i n c lu d i n g t h e c h a r m-b a r y o n s t a t e s. towards higher values. tions with approximate simulations ATLAS Collab. 2021 arXiv:2107.09330. in pp collisions at midrapidity, and, in The cc– c r o s s s e c t i o n m e a s u r e d a t t h e L HC 0 0.2 the case of the Ξc, for t h e fi r s t t i m e i n a ny l ie s at t he up p e r e d ge o f t he t he ore t ic a l CMS collision system (figure 1). The meas- pQCD calculations. u r e d f (c → H ) a r e o b s e r v e d t o b e d i ffe r e nt The measurements described above c Learning to detect new top-quark interactions f r o m t h o s e m e a s u r e d i n e +e– a n d e p c ol- not only provide constraints to pQCD 0 + + + 0 *+ l i sion s – e v ide nc e t h at t he a s su mpt ion D D Ds Λc Ξc D calculations, but also act as important Ever since its discovery in 1995 at the from the data using a strategy that –1 that charm-to-hadron fragmentation references for investigating the interac- Tevatron, the top quark has been con- 138 fb (13 TeV) combines observables specifically is universal is not v alid. Fig. 1. Charm–quark fragmentation fractions into charm hadrons t ion of c ha r m qua rk s w it h t he me d ium s i d e r e d t o b e a h i g h l y e ffe c t i v e pr o b e o f – s e l e c t e d for t h e i r s e n s it i v it y t o E F T. T h e unc. data tZq tWZ ttZ Charm quarks were found to had- measured in pp collisions at √s = 5.02 TeV (black squares), in e+e– created in heav y-ion collisions. These new physics. A key reason is that the CMS k e y fe a t u r e o f t h i s w or k i s it s h e a v y u s e 3 – – ronise into baryons almost 40% of the collisions (purple and yellow markers) and in ep collisions (closed measurements could be extended to l a s t f u n d a m e nt a l fe r m i o n pr e d i c t e d b y 10 SR–ttZ t(t)X WZ VV(V) Xγ NPL of multivariate-analysis techniques time – four times more often than at and open red circles). The D * meson is depicted separately since its include rarer baryons and studied as t h e St a n d a rd M o d e l (S M) h a s a r e m a r k- based on machine learning, which colliders with electron beams. Sev- contribution is also included in the ground-state charm mesons. a function of the event multiplicity ably high mass, just a sliver under the improve its sensit iv it y to new inter ac- 2 eral models have been proposed to in pp and heav y-ion systems in future H i g g s v a c uu m e x p e c t a t io n v a lu e d i v id e d 10 tions. First, to define regions enriched explain this “baryon enhancement”. c o n s i d e r i n g a s e t o f a s-y e t-u n o b s e r v e d LHC r uns. by the square root of two, implying a i n t h e pr o c e s s e s o f i nt e r e s t, a m u l t ic l a s s The explanations feature various higher-mass charm-baryon states, and Yukawa coupling close to unity. This neural network is trained to discrim- events/bin different assumptions, such as includ- including string for mation beyond t he Further reading has far-reaching implications: the top 10 inate between different SM processes. ing hadronisation via coalescence, leading-colour approximation. ALICE Collab. 2021 arXiv:2105.06335. quark impacts the electroweak sector Subsequently, several binary neural net- significantly through loop correc- works learn to separate events generated ATLAS tions, and may couple preferentially 1 according to the SM from events that to new massive states. But while the include EFT effects arising from one or Strongly unbalanced photon pairs t o p q u a r k m a y r e pr e s e nt a w i n d o w i nt o 1.5 more t y p e s o f a nom a lo u s i nt e r a c t ion s. new physics, we cannot know a priori 1.0 For the first time in an analysis using data pred. Most processes resulting from proton– the showering and hadronisation of whether new massive particles could 0.5 L HC d a t a, t h e s e c l a s s i fie r s w e r e t r a i n ed 2 –2 pr o t o n c ol l i s i o n s a t t h e L HC a r e a ffe c t e d ATLAS √s =13TeV,139 fb–1 quarks and gluons. However, the latest ever be produced at the LHC, and direct 10 CtZ/Λ [TeV ] on the full physical amplitudes, includ- data – b y t h e s t r o n g fo r c e – a d i ffic u l t-t o -m o d e l 1 DIPHOX ATLAS analysis instead uses photons, s e a r c h e s h a v e s o f a r b e e n i n c o n c lu s i v e. 1.5 3 ttZ total prediction ing the interference between SM and SM ] NNLOJET 5

part of the Standard Model involving –1 which can be very precisely measured Model-inde pendent measurements EFT components. SHERPA SM+EFT non-perturbative effects. This can be 10–1 by the detector. Although photons do carried out within the framework of The binary classifiers are used to problematic when measuring rare pro- not carry a colour charge, they interact effective field theory (EFT) are therefore 1 234 5 678 construct powerful discriminant var- 0.8 – cesses not mediated by strong inter- [pb GeV 10–2 with quarks as the latter carry electric becoming increasingly important as a NN–CtZ–ttZ output iables out of high-dimensional input

actions, such as those involving the T, γγ 0.4 charge. As a result, strong-interac- means to make the most of the wealth data. Their distributions are fitted to –3 Higgs boson, and when searching for /dp 10 tion effects on the quarks can alter the o f pr e c i s io n m e a s u r e m e nt s at t h e L HC. Fig. 1. The response of a neural network used to target a specific data to constrain up to five types of EFT σ 0 – n e w p a r t i c l e s or i nt e r a c t i o n s. To e n s u r e d characteristics of the measured pho- T h i s a p pr o a c h m a k e s it p o s s i bl e t o s y s- type of EFT interaction in ttZ production. The lower panel shows couplings simultaneously. The widths 0 20 40 60 80 100 such processes are not obscured, pre- 10–4 tons. The conser vation of momentum tematically correlate sparse deviations the change of the event yield in each bin with respect to the SM o f t h e c or r e s p o n d i n g c o n fid e n c e i nt er- cise knowledge of the more dom inant 1.6 allows us to quantify this effect: the o b s e r v e d i n d i ffe r e nt m e a s u r e m e nt s, i n post-fit expectation for t wo benchmark EF T scenarios, both for the vals are significantly reduced thanks to strong-interaction effects, including LHC’s proton beams collide head-on, ord e r t o pi n p oi nt a n y a n o m a l ie s i n t o p - tt–Z process (solid line) and the total prediction (dotted), illustrating the combination of the available kine- those caused by the initial-state par- so the net momentum transverse to quark couplings that might arise from the neural net work’s abilit y to isolate anomalous effects. matic information that was specifically 0.8 tons, is a prerequisite to LHC physics the beam axis must be zero for the un k now n massive par t icles. c ho s e n to b e s e n sit ive to EF T i n t he top

analyses. theory/data final-state particles. Any signs to the A new CMS analysis searches for set, this study targets high-purity quark sector. All results are consistent The electromagnetic production of a 0 contrary indicate additional activity in anomalies in top-quark interactions fi n a l s t a t e s w it h m u l t i pl e e l e c t r o n s a nd with the SM, which indicates either the p h o t o n p a i r i s t h e d o m i n a nt b a c k g r o u n d 4510 0 100 500 t h e e v e nt w it h e q u i v a l e nt b ut o p p o s it e with the Z boson using an EFT frame- m u o n s. It s e t s s o m e o f t h e t i g ht e s t c o n- absence of new effects in the targeted to the H d e c a y c h a n n e l – a pr o c e s s transverse momentum, usually arising work. The cross-section measurements s t r a i nt s to d ate on five ge ne r ic t y p e s of → γ γ pT,γγ [GeV] interactions or that the mass scale of t h at i s i n s t r u me nt a l to t he s t ud y of t he from quarks and gluons radiated from of the rare associated production of EFT interactions that could substantially new physics is too high to be probed – H i g g s b o s o n. D e s pit e it s e l e c t r o m a g n e t ic Fig. 1. Measurements of the differential diphoton cross-section the initial-state partons. Therefore, by either one (tZ) or two (ttZ) top quarks m o d i f y t h e c h a r a c t e r i s t i c s o f a s s o c i a t e d with the current sensitivity. This result nature, diphoton production is affected as a function of the trans verse momentum of the photon pairs m e a s u r i n g t h e t r a n s v e r s e m o m e nt u m with a Z boson were statistically lim- top-Z production, while having negligi- is an important step towards the more by surprisingly large strong-interac- (black points), compared to theoretical predictions including of photon pairs, and related observ- ited until recently. These interactions bl e or n o e ffe c t o n b a c k g r o u n d pr o c e s ses. w idespread use of mac h i ne lea r n i ng to tion effects. Thanks to precise ATLAS those at next-to-leading order (orange) and next-to-next-to ables, the strong interaction may be a re a mong t he lea st c on st r a i ne d b y t he In contrast to the more usual rein- t a r g e t E F T e ffe c t s, t o e ffic i e nt l y e x plore measurements of diphoton processes leading order (purple). indirectly probed. available data in the top-quark sector, terpretations of SM measurements that the enormous volume of LHC data more using the full Run-2 dataset, the col- Comparing the measured values to despite being modified in numerous require assumptions on the nature of globally and comprehensively. l a b or a t i o n i s a bl e t o pr o b e t h e s e e ffe c t s Measurements studying strong predictions reveals the surprising role b e yo n d-S M m o d e l s, s uc h a s c o m p o s it e new physics, this analysis considers and scrutinise state-of-the-art theo- interactions typically employ final of the strong interaction in electromag- Higgs models and minimal supersym- E F T e ffe c t s o n o b s e r v a bl e s a t t h e detc- Further reading s retical calculations. states that include jets produced from n e t i c d i p h o t o n pr o d u c t i o n. I n a s i m pl e metry. Using the full LHC Run-2 data tor level and constrains them directly CMS Collab. 2021 CMS-PAS-TOP-21-001.

16 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 17

CCSepOct21_EnergyFrontiers_v3.indd 16 03/08/2021 12:42 CCSepOct21_EnergyFrontiers_v3.indd 17 03/08/2021 12:42 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

High Voltage Power Solutions FIELD NOTES

Up to 500kVDC output Reports from events, conferences and meetings Computing in HigH-EnErgy pHysiCs with power to 200kW A I and GPUs ta ke centre stage at v CHEP

XP Power offers one of the widest ranges of high voltage power The 25th International Conference on 2021 vCHEP ing calorimeter simulation and liquid- solutions, backed with exceptional technical and customer Computing in High-Energy and Nuclear argon time-projection chamber software support, and a three year warranty. Physics (CHEP) gathered more than 1000 to different toolkits for heterogeneous participants online from 17 to 21 May. programming, a topic that will become Dubbed “vCHEP”, the event took place even more important as computing cen- virtually after this year’s in-person event t res diversi f y t heir offerings. Output voltages from 100VDC to 500kVDC in Norfolk, Virginia, had to be cancelled Keeping up with benchmarking and Output power from 0.5W to 200kW due to the COVID-19 pandemic. Partici- valuing these heterogeneous resources pants tuned in across 20 time zones, is an important topic for the Worldwide Input ranges including DC, single & three phase AC from Brisbane to Honolulu, to live LHC Computing Grid, and a report from Programmable, proportional, local control & various digital options talks, recorded sessions, excellent dis- the HEPiX Benchmarking group pointed cussions on chat apps (to replace the tra- to the future for evaluating modern CPUs Low noise, accurate and reliable high voltage for critical, long term applications ditional coffee-break interactions) and and GPUs for a variety of real-world HEP special sessions that linked job seekers applications. Staying on the facilities For more information about our products visitwww.xppower.com or email us at [email protected] with recruiters. topic, R&D was presented on how to opti- Given vCHEP’s virtual nature this year, On the simulation side, work was vCHEP 2021 mise delivering reliable and affordable t h e r e w a s a d i ffe r e nt fo c u s o n t h e c o nt e nt. pre sented showcasing new neural- More than 1000 storage for HEP, based on CephFS and Plenary speakers are usually invited, but network architectures that use a “bounded participants took the CERN-developed EOS storage sys- this time the organisers invited papers information-bottleneck autoencoder” to part across tem. This will be critical to providing of up to 10 pages to be submitted, and improve training stability, providing a 20 time zones, the massive storage needed in the future. chose a plenary programme from the most solution that replicates important features from Brisbane The network between facilities will likely interesting and innovative. Just 30 had to such as how real minimum-ionising par- to Honolulu. b e c o m e d y n a m i c a l l y c o n fi g u r a bl e i n t he be selected from more than 200 submis- ticles interact with calorimeters. ATLAS future, and how best to take advantage of sions – twice as many as expected – but a l s o s h o w e d o ff t h e i r n e w f a s t-s i m u l a t ion m a c h i n e l e a r n i n g for t r a ffic pr e d i c t i o n i s the outcome was a diverse programme framework, which combines traditional being investigated. tackling the huge issues of data rate and parametric simulation with generative event complexity in future experiments adversarial networks, to provide better Quantum computing in nuclear and high-energy physics (HEP). agreement with Geant4 than ever before. vCHEP was also the first edition of Machine learning is very well suited CHEP with a dedicated parallel session Artificial intelligence to new computing architectures, such as on quantum computing. Meshing very So what were the hot topics at vCHEP? graphics processing units (GPUs), but many well with CERN’s Quantum Initiative, this O n e o ut s t a n d i n g o n e w a s a r t i fic i a l i nt el- other experimental-physics codes are also showed how seriously investigations of ligence and machine learning. There were being rewritten to take advantage of these how to use this technology in the future more papers submitted on this theme new architectures. IceCube are simulating are being taken. Interesting results on than any other, showing that the field is photon transport in the Antarctic ice on using quantum support-vector machines continuing to innovate in this domain. GPUs, and presented detailed work on their to train networks for signal/background Interest in using graph neural net- performance analysis that led to recent classification for B-meson decays were works for the problem of charged- significant speed-ups. Meanwhile, LHCb highlighted. particle tracking was very high, with will introduce GPUs to their trigger farm On a meta note, presentations also three plenary talks. Using a graph to for Run 3, and showed how much this will explored how to adapt outreach events to represent the hits in a tracker as nodes improve the energy consumption per event a virtual setup, to keep up public engage- and possible connections between hits as of the high-level trigger. This will help to ment during lockdown, and how best to edges is a very natural way to represent meet the physical constraints of power use online software training to equip the the data that we get from experiments. and cooling close to the detector, and is a future generation of physicists with the The network can be effectively trained first step towards bringing HEP’s overall advanced software skills they will need. to pick out the edges representing the computing energy consumption to the Was vCHEP a success? So far, the feed- true tracks and reject those that are just table as an important parameter. back is overwhelmingly positive. It was spurious connections. The time needed Encouraging work on porting event a showcase for the excellent work going to get to a good solution has improved generation to GPUs was also presented ATLAS showed on in the field, and 11 of the best papers dramatically in just a few years, and the – particularly appropriately, given the will be published in a special edition of off their scaling of the solution to dense environ- spiralling costs of higher order gener- Computing and Software for Big Science – ments, such as at the High-Luminosity ators for HL-LHC physics. Looking at ne w f a st- anot her first for CHEP in 2021. LHC (HL-LHC), is very promising for this the long-term future of these new code simulation relatively new technique. bases, there were investigations of port- framework Graeme Stewart CERN.

CERN COURIER SEPTEMBER/OCTOBER 2021 19

CCSepOct21_Fieldnotes_v2.indd 19 03/08/2021 12:55 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FIELD NOTES FIELD NOTES

Future CirCular Collider Week LHC PHysiCs FCC feasibility study comes into focus LHCP sees a host of new results ATLAS-PHOTO-2021-052-1 This year’s Future Circular Collider (FCC) OPEN-PHO-ACCEL-2019-001-19 Several talks also pointed out syn- More than 1000 physicists took part in Week took place online from 28 June to ergies with the RF systems needed for the ninth Large Hadron Collider Physics 2 July, attracting 700 participants from the proposed electron–ion collider at (LHCP) conference from 7 to 12 June. The all over the world to debate the next steps Brookhaven and the Powerful Energy- in-person conference was to have been needed to produce a feasibility report Recovery Linac for Experiments (PERLE) held in Paris, however for the second in 2025/2026, in time for the next project at Orsay, and called for stronger year in a row t he organisers efficiently update to the European Strategy for collaboration between the projects. moved the meeting online, without a Particle Physics in 2026/2027. The cur- registration fee, thanks to the support rent strategy, agreed in 2020, sets an Machine design of CERN and IUPAP. While the conference electron–positron Higgs factory as the Another key aspect of the study regards experience cannot be the same over a highest priority facility after the LHC, the machine design. Since the conceptual video link, the increased accessibility for along with the investigation of the tech- design report last year, the pre-injector people from all parts of the international nical and financial feasibility of such a layout for FCC-ee has been simplified, community was evident, with LHCP21 Higgs factory, followed by a high-energy and key FCC-ee concepts have been participants hailing from institutes hadron collider placed in the same 100 km demonstrated at Japan’s SuperKEKB across 54 countries tunnel. The FCC feasibility study will collider, including a new world-record The LHCP format traditionally has fo c us on t he fi rst st age (t un ne l a nd e +e– luminosity of 3.12 × 1034 cm–2 s–1 in June plenary sessions in the mornings and * collider) in t he ne x t five years. with a betatron function of β γ = 1 mm. late afternoons, with parallel sessions Although the FCC is a long-term Separate tests squeezed the beam to just in the middle of the day. This “shape” * project with a horizon up to the 22nd Superconducting allow a similar breakthrough for elec- βγ = 0.8 mm in both rings. was kept for the online meeting, with century, its timescales are rather tight. A key technology troweak symmetry breaking. Regarding Other studies reported during FCC a shorter day to improve the practical- A post-LHC collider should be oper ational for FCC-ee is the recent anomalous measurements, such as Week 2021 demonstrated that hosting ity of joining from distant time zones. around the 2040s, ensuring a smooth development those of the muon’s magnetic moment, four experiments is compatible with This resulted in a dense format with continuation from the High-Luminos- of efficient Reece noted that while these measure- a new four-fold symmetric ring. This seven-fold parallel sessions, allowing ity LHC, so construction would need to superconducting ments could give us the coefficients of redundancy is thought to be essential all parts of the LHC programme, both begin in the early 2030s. Placement stud- radio-frequency one higher dimension operator in an for high-precision measurements, and experimental and theoretical, to be ZH → μμcc candidate ATLAS presented a new search for Higgs decays to charm quarks using powerful new ies to balance geological and territorial cavities, as pictured e ffe c t i v e-fie l d-t h e or y d e s c r i p t io n o f n ew d i ffe r e nt d e t e c t or s olut i o n s w i l l b e i nval- explored in detail. The overall vitality charm-tagging techniques. constraints with machine requirements here with CERN’s physics, only colliders can systematically uable in uncovering hidden systematic of the programme is illustrated by the and physics performance suggest that the Sarah Aull. produce and characterise the nature of biases. The meeting also followed up on raw statistics: a grand total of 238 talks of searches were also reported, some hadron production in proton–proton and most suitable scenarios are based on a any new physics. FCC-ee and FCC-hh the proposal for energy-recovery linacs and 122 posters were presented. using the Higgs as a tool to probe for heavy-ion collisions, for example using 92 km-circumference tunnel with eight both have exciting and complementary (ERLs) at FCC-ee, potentially extending new physics. D m e s o n s. C M S r e p or t e d t h e fi r s t o b s er-

surface sites. roles to play. the energy reach to 600 GeV if deemed From third to second The most puzzling hints from the LHC vation of Bc meson production in heavy- The next steps are subsurface investi- A key technology for FCC-ee is the necessary during the previous physics Nine years on from the discovery of Run 1 seem to strengthen in Run 2. LHCb i o n c ol l i s i o n s, a n d a l s o fi r s t e v i d e n c e for gations of high-risk areas, surface-site development of efficient superconducting runs. First studies for the use of the the 125 GeV Higgs boson, measure- pr e s e nt e d a n a l y s e s r e l a t i n g t o t h e “fl a- top-quark pair production in lead–lead i n it i a l-s t a t e a n a l y s i s a n d v e r i fic a t io n o f radio-frequency (SRF) cavities to com- FCC-ee booster as a photon source were ments have progressed to a new level vour anomalies” found most notably in collisions. ATLAS used heav y-flavour in-principle feasibility with local author- pensate for the 100 MW synchrotron radi- also presented, potentially leading to of precision with the full Run-2 data. b → sµ+µ– decays, updated to the full data decays to muons to compare suppres- ities. A “Mining the Future” competition ation power loss in all modes of operation applications in medicine and industry, Both ATLAS and CMS presented new statistics, in multiple channels (p15). sion of b- and c-hadron production in has been launched to solicit ideas for how from the Z pole up to the top threshold at precision QED studies and fundamental- results on Higgs production, helping While no result yet passes a 5σ d i ffe r e n c e lead–lead and proton–proton collisions. to best use the nine million cubic metres 365 GeV. A staged RF system is foreseen as symmetry tests. constrain the dynamics of the produc- f r o m S M e x p e c t a t i o n s, t h e s i g n i fic a n c e s Beyond the ions, ALICE also showed of molasse that would be excavated from the baseline scenario, with low-imped- Participants also tackled concepts for tion mechanisms via differential and continue to creep upwards. Searches intriguing new results demonstrat- the tunnel. ance single-cell 400 MHz Nb/Cu cavities power reduction and power recycling, “simplified template” cross-section by ATLAS and CMS for potential new ing that the relative rates of different for Z running replaced by four-cell Nb/Cu to ensure that FCC is sustainable and measurements. While the couplings of p a r t ic le s or e ffe c t s at h i g h m a s s e s t h at types of c-hadron production differ in Beyond the LHC cavities for W and Higgs operation, and environmentally friendly. Ideas relating the Higgs to third-generation fermi- could indicate an associated new-phys- proton–proton collisions compared to A highlight of the week was the explo- later augmented by five-cell 800 MHz to FCC-ee include making the magnets ons are now established, last year saw ics mechanism continue to draw a blank, earlier experiments using e+e– and ep ration of the physics case of a post- bulk Nb cavities at the top threshold. superconducting rather than normal a strong focus on the couplings to the however. This remains a dilemma to be collisions at LEP and HERA (p15). LHC collider. Matthew Reece (Harvard As well as investigations into the use conducting, improving the klystron second generation. After first evidence studied with more precision and data Looking forward, the experiments University) identified dark matter, the of HIPIMS coating and the fabrication of e ffic i e n c y, u s i n g E R L s a n d o t h e r e n e rgy- for Higgs decays to muons was reported in Run 3. Other results in the flavour reported on their preparations for the baryon asymmetry and the origin of copper substrates, an innovative slot- storage devices, designing “twin” from CMS and ATLAS results earlier in sector from LHCb included a new meas- coming LHC Run 3, including substantial

primordial density perturbations as key ted waveguide elliptical (SWELL) cavity dipole and quadrupole magnets with a the year, ATLAS presented a new search urement of the lifetime of the Ωc, four upgrades. While some work has been experimental motivations, and the elec- design was presented that would oper- factor-two power saving, and coating with the full Run-2 data for Higgs decays times longer than previous measure- slowed by the pandemic, recommis-

troweak hierarchy problem, the strong FCC ate at 600 or 650 MHz. SWELL cavities SRF cavities with a high-temperature to charm quarks using powerful new ments (CERN Courier July/August 2021 sioning of the detectors has begun in CP problem and the mystery of flavour optimise the surface area, simplify the superconductor. charm-tagging techniques. Both CMS p17) a n d t h e fi r s t o b s e r v a t i o n o f a m as preparation for physics data taking in – mixing patterns as key theoretical moti- coating process and avoid the need for All in all, FCC Week 2021 saw tremen- and ATLAS showed updated searches difference between the mixed D0–D0 spring 2022, with the brighter beams vations. The present situation in particle welding in critical areas, which could dous progress across different areas of for Higgs-pair production, with ATLAS meson mass eigenstates (CERN Courier Last year expected from the upgraded CERN accel- physics is reminiscent of the early days reduce the performance of the cavity. the study. The successful completion of being able to exclude a production rate July/August 2021 p8). saw a strong erator chain. One constant to rely on, of superconductivity, he noted, when we T h e d e s i g n pr o fit s f r o m pr e v i o u s w or k o n the FCC Feasibility Study (2021–2025) will more than 4.1 times the Standard Model A wealth of results was presented from focus on the however, is that LHCP will continue to had a phenomenological description of CLIC, a nd m ay offer a si mpl i fie d i n st al- be a crucial milestone for the future of (S M) pr e d ic t io n a t 95% c o n fid e n c e. T h i s heavy-ion collisions. Measurements . showcase the fantastic panoply of phys- symmetry breaking in superconductivity, l a t i o n s c h e d u l e w h i l e a l s o fi n d i n g a p pl i- CERN and t he field. is a process that should be observable with heavy quarks were prominent here couplings to ics at the LHC. but no microscopic picture. Constraining cations outside of high-energy physics. with High-Luminosity LHC statistics, as well. ALICE reported various stud- the second the shape of the Higgs potential could A prototype will be tested later this year. Panos Charitos CERN. if it is as predicted in the SM. A host ies of the differences in heav y-flavour generation Dave Charlton University of Birmingham.

20 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 21

CCSepOct21_Fieldnotes_v2.indd 20 03/08/2021 12:56 CCSepOct21_Fieldnotes_v2.indd 21 03/08/2021 12:56 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FIELD NOTES FIELD NOTES

Str ange Quark M atter entirely outside of the detector volume. Simulating currently shaping the future of US par- ers Pythia module, highlighting the To ensure that this discovery avenue is dark showers is ticle physics. Dark showers are a generic growing interest in this subject in the Experiment and theory trade blows at SQM 2021 also covered for the future of the LHC’s a longstanding and poorly understood feature of a poten- LLP community. operation, there is a rich set of dedicated challenge tial BSM dark sector with similarities LLP9 was raucous and stimulating, The 19th international conference on Laboratory National Brookhaven new expansion schemes to extend the LLP detectors either approved or pro- to QCD, which could have its own “dark and identified multiple new avenues of strangeness in quark matter (SQM) lattice-QCD equation of state to larger posed, and LLP9 featured updates from hadronisation” rules. Simulating dark research. LLPX, the tenth workshop in the was hosted virtually by Brookhaven net baryon chemical potentials that are MoEDAL, FASER, MATHUSLA, CODEX-b, showers is a longstanding challenge. series, will be held in November this year. National Laboratory from 17 to 22 May, not computable directly, because of the MilliQan, FACET and SND@LHC, as well More than 50 participants joined for a attracting more than 300 participants. fermion-sign problem. Another study as a presentation about the proposed hands-on demonstration of simulation James Beacham Duke University and Anz_spec_CERN_8-2021_sp 29.07.21 09:34 Seite 1 The series deals with the role of strange included the lattice-QCD equation of forward physics facility for the High- tools and a discussion of the dark-show- Albert De Roeck CERN. and heavy-flavour quarks in high- state and susceptibilities in a hydrody- Luminosity LHC (HL-LHC). energy heavy-ion collisions and astro- namic calculation to allow for a more physical phenomena. direct comparison to experimental Reinterpreting machine learning New results on the production of measurements of net-proton fluctua- The liveliest parts of any LLP commu- strangeness in heavy-ion collisions tions. Significant differences between nity workshop are the brainstorming New Digital I/O card were presented for a variety of collision net-proton and net-baryon fluctuations and hands-on working-group sessions. energies and systems. In an experimen- were quantified. LLP9 included multiple vibrant discus- tal highlight, the ALICE collaboration The study of the quark–gluon plas- sions and working sessions, including Perfect for logic analysis and pattern generation reported that the number of strange ma’s vorticity via the measurement of on heavy neutral leptons and the ability baryons depends more on the final- the polarisation of hyperons was also of physicists who are not members of state multiplicity than the initial-state a major topic. Theoretical calculations experimental collaborations to be able energy. On the theory side, it was shown STAR detector quantum systems coupled with potential obtain the opposite sign to the data for to re-interpret LLP searches – a key issue that several models can explain the Experiments at non-relativistic QCD calculations were the angular differential measurement. for the LLP community. At LLP9, partici- suppression of strange particles at low BNL’s Relativistic used to compute both the suppression and Attempts to solve this discrepancy pants examined the challenges inherent multiplicities. ALICE also presented new Heavy-Ion Collider a n i s o t r o pic flo w o f b o t t o m o n iu m s t a t es. presented at SQM 2021 featured shear- in re-interpreting LLP results that use measurements of the charm cross section are seeking to dependent terms and a stronger “mem- machine learning techniques, by now and fragmentation functions in proton– confirm hints of Hints of extrema ory” of the strange-quark spin. a common feature of particle-physics proton (pp) collisions. When compared extrema as a Another important goal of the field is Various new applications of machine analyses. For example, boosted deci- to e+e– collisions, these results suggest function of to determine experimentally whether a l e a r n i n g a n d a r t i fic i a l i nt e l l i g e n c e w e re sion trees (BDTs) and neural networks that the universality of parton-to- beam energy. critical point exists in the phase diagram also discussed, for example, for deter- (NNs) can be quite powerful for either hadron fragmentation may be broken. of strongly interacting matter, and, if mining the order of the phase transi- o bje c t id e nt i fic at ion or e ve nt-le ve l d i s- Moving on to heavy flavours, the so, where it is located. The STAR exper- tion and constraining the neutron-star crimination in LLP searches, but it’s not ATLAS collaboration presented results iment at the Relativistic Heavy Ion Col- equation of state. entirely clear how best to give theorists for the suppression of heavy-flavour lider (RHIC) presented results on higher Overall, there were 41 plenary and 96 access to the full original BDT or NN used production compared to pp collisions order cumulants of net-proton fluctua- parallel talks at SQM 2021, poignantly internally by the experiments. and the angular anisotropy of heavy tions over a range of collision energies. including presentations in memory of LLP searches at the LHC often must mesons in heavy-ion collisions. These Extrema as a function of beam energy Jean Cleymans, Jean Letessier, Dick Majka also grapple with background sources measurements are crucial for constrain- are expected to indicate critical behav- and Jack Sandweiss, who all made excep- that are negligible for the majority of ing models of in-medium energy loss. iour. New data from the Beam Energy t ional impacts on t he field. searches for prompt objects. These Interestingly, while charm seems to fol- Scan II programme at RHIC is expected The next SQM conference will be backgrounds – such as cosmic muons, l o w t h e flo w o f t h e q u a r k–g lu o n pl a s ma, to provide much-needed statistics to held from 13 to 18 June 2022 in Busan, beam-induced backgrounds, beam-halo b e a ut y d o e s n o t s e e m t o flo w. B e t t e r s ta- confirm hints of extrema in the data. South Korea. effects and cavern backgrounds – are Synchronize up to 16 different tistics are needed to constrain theoretical On the theory side, new lattice QCD cal- reasonably well-understood for Run 2 PCIe cards of the M2p family 32 channels for input or output models. On the theory side, extremely culations of second-order net-baryon Bjoern Schenke Brookhaven and Run 3, but little study has been per- in one PC system! interesting new calculations using open cumulants were presented, as well as National Laboratory. formed for the upcoming HL-LHC, and Speeds up to 125 MHz Choose from 24 different digitizer LLP9 featured a brainstorming session 3.3 V and 5 V TTL compatibility models, 14 different AWG variants LLP9 about what such non-standard back- 1 GByte of on-board memory CMS / cms3d.web.cern.ch/EXO-20-003 CMS and the new Digital I/O card! be produced in a proton–proton collision grounds might look like in the future. PCIe x4 with FIFO speed up to 700 MB/s Long-lived and then get stuck in the detector before Also looking to the future, two very decaying minutes, days or weeks later. forward-thinking working-group ses- SDKs for C++, LabVIEW, MATLAB, particles Good hypothetical examples are the sions were held on LLPs at a potential VB.NET, Python, Java, Delphi etc. so-called gluino R-hadrons that occur future muon collider and at the proposed Up to 512 synchronized channels gather interest in supersymmetric models. Also fea- Future Circular Collider (FCC). Hadron (16 cards) tured was a new CMS search for displaced collisions at ~100 TeV in FCC-hh would From 25 to 28 May, the long-lived particle di-muon resonances using “data scout- open up completely unprecedented (LLP) community marked five years of ing” – a unique method of increasing the discovery potential, including for LLPs, stretching the limits of searches for new number of potential signal events kept but it’s unclear how to optimise detector SPECTRUM physics with its ninth and best-attended at the trigger level by reducing the event designs for both LLPs and the full slate INSTRUMENTATION workshop yet, with more than 300 reg- information that is retained. Both exper- of prompt searches. Perfect fit – modular designed solutions istered participants. iments presented new results searching Finally, LLP9 hosted an in-depth LLP9 played host to six new results, for the Higgs boson decaying to LLPs (see working-group session dedicated to three each from ATLAS and CMS. These LLP candidate A CMS candidate event for the decay of a Higgs “LLP candidate” fig ure). the simulation of “dark showers”, in Europe / Asia: Phone +49 (4102) 695 60 | US: Phone (201) 562 1999 included a remarkable new ATLAS paper boson into a pair of LLPs, which decay into jets (yellow bands) Long-lived particles can also be pro- collaboration with the organisers of the searching for stopped particles: beyond- containing secondary vertices (orange circles), in association duced in a collision inside ATLAS, CMS dark-showers study group connected www.spectrum-instrumentation.com s the-Standard Model (BSM) LLPs that can with Z → e+e– (green track s). or LHCb and live long enough to drift to the Snowmass process, which is

22 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 23

CCSepOct21_Fieldnotes_v2.indd 22 03/08/2021 12:56 CCSepOct21_Fieldnotes_v2.indd 23 03/08/2021 12:57 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

FIELD NOTES

InternatIonal PartIcle acceler ator conference Fast Pulse Generators & LNLS on the future of high-power free- IPAC thrives electron lasers; Houjun Qian (DESY) on radio-frequency photocathode guns; Arbitrary Waveform Generators online and Young-Kee Kim (University of Chi- cago) on future directions in US particle The annual International Particle Accel- physics. The closing plenary talk was a erator Conference (IPAC) promotes col- sobering presentation on climate change laboration among scientists, engineers, and the Brazilian Amazonia region by technicians, students and industrial Paulo Artaxo (University of São Paulo). partners across the globe. Originally to The remainder of the talks were be hosted this year by the Laboratório pre-recorded with live Q&A sessions, and Nacional de Luz Síncrotron (LNLS) in 400 teleconferencing rooms per day were Campinas, Brazil, the conference was set up to allow virtual poster sessions. moved online when it became clear Highlights in topical sessions included that the global pandemic would pro- “Women in Science: The Inconvenient hibit travel. IPAC21 was nevertheless Truth” by Márcia Barbosa (Universi- highly successful, attracting more than dade Federal do Rio Grande do Sul) and 1750 participants online from 24 to 28 Sirius storage ring A newly built fourth-generation light an industrial forum hosted by Raffaella May. Despite the technical and logistical source at the Laboratório Nacional de Luz Síncrotron, which Geometrante (KYMA) on the intersection challenges, the virtual platform provided virtually hosted IPAC this year. between government accelerator projects many advantages, including low or zero and industry. registration fees and a larger, younger Highlights included Harry Westfahl, Jr IPAC22 is currently planned as an and more diverse demographic than (LNLS) on the scientific capabilities of in-person conference in Bangkok, typical in-person events, which tend to fourth-generation storage-ring light Thailand, from 17 to 22 June next year. attract about 1000 delegates. sources; Thomas Glasmacher (FRIB) on In order to allow worldwide virtual the newly commissioned Facility for John Byrd Argonne National Laboratory, Rise / Fall Time Sampling Rate Amplitude Range participation, live plenary presenta- Rare Isotope Beams at Michigan State Liu Lin LNLS and tions were limited to two hours daily. University; Norbert Holtkamp (SLAC) Regis Neuenschwander CNPEM. www.activetechnologies.it Phone: +39 0532 1772145 | E-Mail: [email protected] Designed and Manufactured in Italy SuStHeP 2021 SustHEP 2021 SustHEP Sustainable the true meaning of online meetings to Sustainable participants her: everyone wants to belong. It was Before the only during the first online meetings high-energy pandemic, a during the pandemic that she truly felt typical HEP a real sense of belonging to the global physics researcher scientific community. was expected to The third day was dedicated to exist- cross the world ing sustainability initiatives and new COVID-19 put the community on a steep several times a year technologies. Mike Seidel (PSI) presented learning curve regarding new forms of for conferences, studies on energy-recovery linacs and online communication and collabo- collaboration discussed energy-management con- ration. Before the pandemic, a typical meetings and cepts for future colliders, including daily high-energy physics (HEP) researcher detector shifts. “standby modes”. Other options include was expected to cross the world several beam dynamics explicitly designed to times a year for conferences, collabo- maximise the ratio of luminosity to ration meetings and detector shifts, at power, more efficient radio-frequency the cost of thousands of dollars and a and collaboration meetings. Instead, cavities, the use of permanent magnets, sizeable carbon footprint. The online the community needs to rethink what and high-temperature superconductor workshop Sustainable HEP – a new ini- virtual scientific exchange could look cables and cavities. He concluded his talk tiative this year – attracted more than like in the 21st century. This might, for by asking thought-provoking questions 300 participants from 45 countries instance, include replacing traditional such as whether the HEP community from 28 to 30 June to discuss how the live presentations by pre-recorded should engage with its international lessons learned in the past two years talks that are pre-watched by the audi- networks to help establish sustainable might help HEP transition to a more ence at their own convenience, leav- energy-supply solutions. sustainable future. ing more precious conference time for The workshop ended by drafting a T h e fi r s t d a y o f t h e w or k s h o p fo c used in-depth discussions and interactions closing statement that calls upon the on how new forms of online interaction among the participants. HEP community to align its activities could change our professional travel with the Paris Climate Agreement and culture. Shaun Hotchkiss (University Social justice the goal of limiting global warming to of Auckland) stressed in a session ded- The second day highlighted social- 1.5 degrees. This statement can be signed icated to best-practice examples that justice issues, and the potential for by members of the HEP community until the purpose of online meetings should greater inclusivity using online for- 20 August. not simply be to emulate traditional mats. Alice Gathoni (British Institute 20th-century in-person conferences in Eastern Africa) powerfully described Kai Schmitz CERN. 614.891.2243 | www.lakeshore.com

24 CERN COURIER SEPTEMBER/OCTOBER 2021

CCSepOct21_Fieldnotes_v2.indd 24 03/08/2021 12:59 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

FEATURE DATA CHALLENGES B Nachman (inspired by arxiv:0811.4622) by (inspired B Nachman WHAT’S IN THE BOX?

The LHC Olympics and Dark Machines data challenges stimulated innovation in the use of machine learning to search for new physics, write Benjamin Magnifying the unexpected Artist’s impression of a neural network probing Nachman and Melissa van Beekveld. a black box of complex final states at the LHC.

he need for innovation in machine learning (ML) set of black-box datasets composed mostly of Standard transcends any single experimental collaboration, Model (SM) background events. Contestants were charged T and requires more in-depth work than can take with identifying any anomalous events that would be a place at a workshop. Data challenges, wherein simu- sign of new physics. These datasets focused on resonant lated “black box” datasets are made public, and contest- anomaly detection, whereby the anomaly is assumed to ants design algorithms to analyse them, have become b e lo c a l i se d – a “bump hunt”, i n effe c t. Th i s i s a gener ic essential tools to spark interdisciplinary collaboration feature of new physics produced from massive new par- and innovation. Two have recently concluded. In both ticles: the reconstructed parent mass is the resonant cases, contestants were challenged to use ML to figure feature. By assuming that the signal is localised, one out “what’s in the box?” can use regions away from the signal to estimate the background. The LHCO provided one R&D dataset with LHC Olympics labels and three black boxes to play with: one with an The LHC Olympics (LHCO) data challenge was launched anomaly decaying into two two-pronged resonances, one in autumn 2019, and the results were presented at the without an anomaly, and one with an anomaly featuring ML4Jets and Anomaly Detection workshops in spring and t wo d i ffe r e nt d e c a y m o d e s (a d ij e t d ecayX → qq and a trijet ® summer 2020. A final report summarising the challenge decay X → gY, Y → qq). There are currently no dedicated was posted to arXiv earlier this year, written by around 50 searches for these signals in LHC data. HiPace turbopumps authors from a variety of backgrounds in theory, the ATLAS and CMS experiments, and beyond. The name of this No labels community effort was inspired by the first LHC Olympics About 20 algorithms were deployed on the LHCO datasets, For perfect vacuum solutions to challenging applications that took place more than a decade ago, before the start including supervised learning, unsupervised learning, of the LHC. In those olympics, researchers were worried weakly supervised learning and semi-supervised learning. about being able to categorise all of the new particles Supervised learning is the most widely used method across ■ Unique rotor design with proven hybrid bearing that would be discovered when the machine turned on. science and industry, whereby each training example has ■ High compression for light gases Since then, we have learned a great deal about nature at a label: “background” or “signal”. For this challenge, the TeV energy scales, with no evidence yet for new particles data do not have labels as we do not know exactly what we ■ Minimal space needs due to compact design or forces of nature. The latest LHC Olympics focused on are looking for, and so strategies trained with labels from THE AUTHORS ■ Extended maintenance intervals a d i fferent c ha l lenge – bei ng able to find ne w physics i n a different dataset often did not work well. By contrast, Benjamin ■ Bearing replacement on site the first place. We now know that new physics must be unsupervised learning generally tries to identify events Nachman Lawrence Berkeley ■ Low vibration due to patented Laser Balancing™ technology rare and not exactly like what we expected. that are rarely or never produced by the background; National In order to prepare for rare and unexpected new physics, weakly supervised methods use some context from data Laboratory and organisers Gregor Kasieczka (University of Hamburg), to provide noisy labels; and semi-supervised methods use Melissa van Are you looking for a perfect vacuum solution? Please contact us: Benjamin Nachman (Lawrence Berkeley National Lab- some simulation information in order to have a partial set Beekveld oratory) and David Shih (Rutgers University) provided a of labels. Each method has its strengths and weaknesses, University of Oxford. Pfeiffer Vacuum (Schweiz) AG · T +41 44 444 22 55 · F +41 44 444 22 66 · [email protected] www.pfeiffer-vacuum.com CERN COURIER SEPTEMBER/OCTOBER 2021 27

21.08.03_ID2168_CernCourier_Sept-Oct_HiPace_turbopumps_213x282mm_EN.indd 1 03.08.2021 18:07:10 CCSepOct21_DATAUPGRADE_v3.indd 27 03/08/2021 15:52 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

FEATURE DATA CHALLENGES

Br y a n O s t id ie k (Ha r v a rd Un i v e r s it y) i n t h e d e v e lo p m e nt 5 10 of a common software infrastructure, and Melissa van 5 Beekveld (University of Oxford) for dataset generation. arXiv:2101.08320 arXiv:2101.08320 In total, 39 participants arranged in 13 teams explored various unsupervised techniques, with each team sub- 4

round mitting multiple algorithms. 0 1 2 3 4 kg 10 10 10 10 10 By cont r ast w it h LHCO, t he Da rk M ac h i nes dat a c ha l-

bac counts

aled lenge focused on developing algorithms broadly sen- /p 3

eve sitive to non-resonant anomalies. Good examples of signal

unr anomaly non-resonant new physics include many supersymmetric 2 models and models of dark matter – anything where α = p “invisible” particles don’t interact with the detector. In such a situation, resonant peaks become excesses in 1 the tails of the missing-transverse-energy distribution. 2750 3000 3250 3500 3750 4000 4250 101 103 105 Two datasets were provided: R&D datasets including a MJJ [GeV] concoction of SM processes and many signal samples Olympian algorithm The “anomaly score”, α, as a function of the invariant for contestants to develop their approaches on; and a mass of the leading two jets of events in “black box 1” of the LHCO data challenge, black-box dataset mixing SM events with events from in the analysis of Stein, Seljak and Dai, who used an early form of a technique u n s p e c i fie d s i g n a l pr o c e s s e s. T h e c h a l le n g e h a s n owfor- that is now called “Gaussianising iterative slicing”. A number of anomalous mally concluded, and its outcome was posted on arXiv in events are seen near 3750 GeV. May, but the black-box has not been opened to allow the community to continue to test ideas on it. and multiple approaches are usually needed to achieve a A wide variety of unsupervised methods have been broad coverage of possible signals. deployed so far. The algorithms use diverse representa- T h e b e s t p e r for m a n c e o n t h e fi r s t bl a c k b o x i n t h e LHCO t io n s o f t h e c ol l id e r e v e nt s (for e x a m ple, l i s t s o f p a r t ic le c h a l le n g e, a s m e a s u r e d b y fi n d i n g a n d c or r e c t l y c h arc- four-momenta, or physics quantities computed from terising the anomalous signals, was by a team of cosmol- them), and both implicit and explicit approaches for ogists at Berkeley (George Stein, Uros Seljak and Biwei estimating the probability density of the background Dai) who compared the phase-space density between (for example, autoencoders and “normalising flows”). a sliding signal region and sidebands (see “Olympian While no single method universally achieved the highest algorithm” figure). Overall, the algorithms did well on sensitivity to new-physics events, methods that mapped RELIABLE t h e R& D d at a s e t, a n d s o m e a l s o d id w e l l o n t h e fi r s t bl a c k the background to a fixed point and looked for events b o x, w it h m e t h o d s t h at that were not described well by this mapping generally The Dark Machines data m a d e u s e o f l i k e l i h o o d did better than techniques that had a so-called dynamic UNDER ALL challenge focused on ratios proving par- embedding. A key question exposed by this challenge t ic u l a rl y e ffe c t i v e. B ut that will inspire future innovation is how best to tune and CONDITIONS. developing algorithms no method was able combine unsupervised machine-learning algorithms in to detect the anoma- a way that is model independent with respect to the new broadly sensitive to lies in the third black physics describing the signal. Heading into the unknown to open new horizons non-resonant anomalies box, and many teams The enthusiastic response to the LHCO and Dark demands reliable tools. Help turn your resarch reported a false signal Machines data challenges highlights the important goals into reality. for the second black future role of unsupervised ML at the LHC and else- Vacuum valve solutions and bellows box. This “placebo where in fundamental physics. So far, just one analy- from VAT provide unfailing reliability effect’’ illustrates the need for ML approaches to have sis has been published – a dijet-resonance search by a n a c c u r at e e s t i m at io n o f t h e b a c k g r o u n d a n d n o t ju s t a the ATLAS collaboration using weakly-supervised M L – and enhanced process safety – procedure for identifying signals. The challenge for the but many more are under way, and these techniques are under all conditions. third black box, however, required algorithms to identify even being considered for use in the level-one triggers multiple clusters of anomalous events rather than a single of LHC experiments (p31). And as the detection of out- cluster. Future innovation is needed in this department. liers also has a large number of real-world applications, from fraud detection to industrial maintenance, fruitful Dark Machines cross-talk between fundamental research and industry A second data challenge was launched in June 2020 within is possible. t h e D a r k M a c h i n e s i n it i at i v e. D a r k M a c h i n e s i s a r e s e a rc h The LHCO and Dark Machines data challenges are a collective of physicists and data scientists who apply stepping stone to an exciting experimental programme ML techniques to understand the nature of dark mat- t hat is just beginning.  ter – as we don’t know the nature of dark matter, it is critical to search broadly for its anomalous signatures. Further reading The challenge was organised by Sascha Caron (Radboud T Aarrestad et al. 2021 arXiv:2105.14027. University), Caterina Doglioni (University of Lund) and ATLAS Collaboration 2020 Phys. Rev. Lett. 125 131801. Maurizio Pierini (CERN), with notable contributions from G Kasieczka et al. 2021 arXiv:2101.08320.

28 CERN COURIER SEPTEMBER/OCTOBER 2021 www.vatvalve.com

CCSepOct21_DATAUPGRADE_v3.indd 28 03/08/2021 15:53 CERNCOURIER 202009_Sept2020_CERNCourier_101x282mm_rightPage_rightSide.inddwww. 1 03.06.2020 10:54:24 V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

FEATURE DEEP LEARNING S Sioni/CMS-PHO-EVENTS-2021-004-2/M Rayner

THE 15th VIENNA CONFERENCE Trigger warning Artist’s impression of an FPGA in the level-one trigger scanning for anomalies at a rate of 40 million events per second. ON INSTRUMENTATION 18 – 22 FEBRUARY 2019 HUNTING ANOMALIES WITH AN AI TRIGGER Jennifer Ngadiuba and Maurizio Pierini describe how ‘unsupervised’ machine learning could keep watch for signs of new physics at the LHC that have not yet been dreamt up by physicists.

n the 1970s, the robust mathematical framework of Blind faith the Standard Model (SM) replaced data observation To maximise discovery potential while minimising the Ias the dominant starting point for scientific inquiry risk of false discovery claims, physicists design rigorous in particle physics. Decades-long physics programmes data-analysis protocols to minimise the risk of human bias. were put together based on its predictions. Physicists built Data analysis at the LHC is blind: physicists prevent them- complex and highly successful experiments at particle selves from combing through data in search of surprises. colliders, culminating in the discovery of the Higgs boson Simulations and “control regions” adjacent to the data of at the LHC in 2012. interest are instead used to design a measurement. When Along this journey, particle physicists adapted their the solidity of the procedure is demonstrated, an internal methods to deal with ever growing data volumes and rates. review process gives the analysts the green light to look at the To handle the large amount of data generated in collisions, result on the real data and produce the experimental result. they had to optimise real-time selection algorithms, or A blind analysis is by necessity a supervised approach. triggers. The field became an early adopter of artificial The hypothesis being tested is specified upfront and tested intelligence (AI) techniques, especially those falling under against the null hypothesis – for example, the existence the umbrella of “supervised” machine learning. Verifying of the Higgs boson in a particular mass range versus its the SM’s predictions or exposing its shortcomings became absence. Once spelled out, the hypothesis determines other THE AUTHORS the main goal of particle physics. But with the SM now aspects of the experimental process: how to select the Jennifer apparently complete, and supervised studies incrementally data, how to separate signals from background and how to Ngadiuba Caltech excluding favoured models of new physics, “unsuper- interpret the result. The analysis is supervised in the sense and Fermilab, vised” learning has the potential to lead the field into the that humans identify what the possible signals and back- and Maurizio uncharted waters beyond the SM. grounds are, and label examples of both for the algorithm. Pierini CERN.

CERN COURIER SEPTEMBER/OCTOBER 2021 31

CCSepOct21_AI_v2.indd 31 04/08/2021 12:01 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FEATURE DEEP LEARNING FEATURE DEEP LEARNING

cal, using a ranking metric learned during training. One Puljak E detector high-level data of the advantages of this approach is that the algorithm M Pier i n i collisions L1 trigger trigger analysis Standard Model can be trained on data recorded by the experiment rather LQ → bτ than simulations. This could, for example, be a control 40 A → 4ℓ sample that we know to be dominated by SM processes: the algorithm will learn how to reconstruct these events “exactly” – and conversely how to rank unknown processes 20 as atypical. As a proof of principle, this strategy has already been applied to re-discover the top quark using the first open-data release by the CMS collaboration.

40,000,000 100,000 1000 2

z 0 events/sec events/sec events/sec This approach could be used in the online data processing at the LHC and applied to the full 40 million collision events Big data The data flow from the ATLAS and CMS experiments must be filtered down to just 1000 events per second for the data to be handled by produced every second. Clustering techniques commonly the available downstream computing resources. This is done by a two-stage real-time filtering process. The level-one (L1) trigger is coded on used in observational astronomy could be used to highlight –20 application-specific integrated circuits and field-programmable gate arrays underground near the detectors. The high-level trigger operates on the recurrence of special kinds of events. CPUs at ground level. Anomaly detection would be most beneficial at the L1 trigger, where all produced events could be inspected. In case a new kind of process happens in an LHC colli- sion, but is discarded by the trigger algorithms serving –40 The data flow at the LHC makes the need to specify a image is a cat. An ML algorithm is a function of this kind, the traditional physics programme, an anomaly-detection signal hypothesis upfront even more compelling. The LHC whose parameters are fixed by looking at a given dataset algorithm could save the relevant events, storing them in a special stream of anomalous events (see “Anomaly produces 40 million collision events every second. Each for which the correct labels are known. Through a training –60 –40 –20 0 20 40 60 overlaps with 34 others from the same bunch crossing, process, the algorithm is tuned to minimise the number of hunting” figure). The ultimate goal of this approach would z1 on average, like many pictures superimposed on top of wrong answers by comparing its prediction to the labels. be the creation of an “anomaly catalogue” of event topol- each other. However, the computing infrastructure of a Now replace the dogs with photons from the decay of a ogies for further studies, which could inspire novel ideas Anomaly hunting In this illustrative simulation of the unsupervised detection of typical experiment is designed to sustain a data flow of Higgs boson, and t he cats w it h detector noise t hat is mis- for new-physics scenarios to test using more traditional leptoquark (orange squares) and neutral-scalar-boson (red triangles) decays on a ju s t 1000 e v e nt s p e r s e c o n d. To a v oid b e i n g o v e r w h e l m e d taken to be photons. Repeat the procedure, and you will techniques. With an anomaly catalogue, we could return SM background (purple circles), LHC collisions are compressed by an autoencoder by the data pressure, it’s necessary to select these 1000 obtain a photon-identification a lgor it h m t hat you ca n use to the first stage of the scientific method, and recover and then further compressed to a two-dimensional representation (z1,z2), which out of e ve r y 40 million events in a short time. But how do on LHC data to improve the search for Higgs bosons. This is a data-driven alternative approach to the theory-driven is suitable for human observation, using the t-SNE algorithm. While most you decide what’s interesting? w h at h a p p e n e d i n t h e C M S e x p e r i m e nt b a c k i n 201 2. T h a n k s investigation that we have come to rely on. new-physics events overlap with the SM events, the most anomalous populate the This is where the supervised nature of data analysis to t he us e of a s p e c i a l k i nd of M L a l gor it h m c a l le d b o o s te d This idea comes with severe technological challenges. outlying regions. These outliers could be used to define a stream of potentially at the LHC comes into play. A set of selection rules – the decision trees, it was possible to maximise the accuracy of To apply this technique to all collision events, we would interesting events to be further scrutinised in future data-taking campaigns. t r i g g e r a l gor it h m s – a re d e s i g n e d s o t h at t h e k i n d o f c ol- the Higgs-boson search, exploiting the rich information need to integrate the algorithm, typically a special kind of lisions predicted by the signal hypotheses being studied provided by the experiment’s electromagnetic calorimeter. neural network called an autoencoder, into the very first on the traditionally triggered samples. are present among the 1000 (see “Big data” figure). As The ATLAS collaboration developed a similar procedure to stage of the online data selection, the level-one (L 1) t r i g g e r. If this strategy is endorsed by the experimental col- lo n g a s yo u k n o w w h at t o lo o k for, t h i s s t r at e g y o p t i m i s e s identify Higgs bosons decaying into a pair of tau leptons. The L1 trigger consists of logic algorithms integrated onto laborations, it could create a public dataset of anomalous your resources. The discovery in 2012 of the Higgs boson Photon and tau-lepton c l a s si fie r s a re b ot h e x a mple s of c u s t o m e le c t r o n ic b o a rd s b a s e d o n fie ld pr o g r a m m a ble g at e data that could be investigated during the third LHC long demonstrates this: a mission considered impossible in supervised learning, and the success of the discovery of arrays (FPGAs) – a highly parallelisable chip that serves shutdown, from 2025 to 2027. By studying those events, the 1980s was accomplished with less data and less time the Higgs boson was also a success story for applied ML. as a programmable emulator of electronic circuits. Any L1 phenomenologists and theoretical physicists could for- than anticipated by the most optimistic guesses when S o f a r s o g o o d. B ut w h at a b o ut s e a rc h i n g for n e w p h y s ic s? trigger algorithm has to run within the order of one micro- mulate creative hypotheses about new-physics scenarios the LHC was being designed. Machine learning played a T y pic a l e x a m ple s o f n e w p h y s ic s s uc h a s s up e r s y m m e t r y, second, and take only a fraction of the available computing to test, potentially opening up new search directions for crucial role in this. extra dimensions and the underlying structure for the Higgs resources. To run in the L1 trigger system, an anomaly the High-Luminosity LHC. b o s o n h a v e b e e n e x t e n s i v e l y i n v e s t i g at e d at t h e L HC, w it h detection network needs to be converted into an electronic Blind analyses minimise human bias if you know what Machine learning no evidence for them found in data. This has told us a great circuit that would fulfill these constraints. This goal can to look for, but risk yielding diminishing returns when the Machine learning (ML) is a branch of computer science that d e a l a b o ut w h at t h e p a r t ic le s pr e d ic t e d b y t h e s e s c e n a r io s be met using the “hls4ml” (high-level synthesis for ML) theoretical picture is uncertain, as is the case in particle deals with algorithms capable of accomplishing a task with- cannot look like, but what if the signal hypotheses are library – a tool designed by an international collaboration physics after the first 10 years of LHC physics. Unsupervised out being explicitly programmed to do so. Unlike traditional simply wrong, and we’re not looking for the right thing? of LHC physicists that exploits automatic workflows. AI techniques such as anomaly detection could help the field break beyond the limits of human creativity in theory algorithms, which are sets of pre-determined operations, This situation calls for “unsupervised” learning, where building. In the big-data e n v i r o n m e nt o f t h e L HC, t h e y offe r an ML algorithm is not programmed. It is trained on data, so hu m a n s a r e n o t r e q u i r e d t o l a b e l d at a. A s w it h s up e r v i s e d Computer-science collaboration Recently, we collaborated with a team of research- a powerful means to move the field back to data-driven that it can adjust itself to maximise its chances of success, le a r n i n g, t h i s id e a d o e s n’t or i g i n at e i n p h y s ic s. M a r k e t i n g ers from Google to integrate the hls4ml library into discovery, after 50 years of theory-driven progress. To as defined by a quant itat ive fig ure of merit. teams use clustering algorithms based on it to identify Google’s “QKeras” – a tool for developing accurate ML maximise their impact, they should be applied to every To explain further, let’s use the example of a dataset of customer segments. Banks use it to detect credit-card fraud models on FPGAs with a limited computing footprint. collision produced at the LHC. For that reason, we argue i m a ge s of c at s a nd dog s. We’l l l a b e l t he c at s a s “0” a nd t he by looking for anomalous access patterns in customers’ Thanks to this partnership, we developed a workflow that that anomaly-detection algorithms should be deployed dogs as “1”, and represent the images as a two-dimensional accounts. Similar anomaly detection techniques could be can design a ML model in concert with its final imple- in the L1 triggers of the LHC experiments, despite the a r r a y o f c olo u r e d pi x e l s, e a c h w it h a f r a c t io n o f r e d, g r e e n a n d used at the LHC to single out rare events, possibly origi- AI techniques mentation on the experimental hardware. The resulting technological challenges that must be overcome to make blue. Each dog or cat is now a stack of three two-dimensional nating from new, previously undreamt of, mechanisms. QKeras+hls4ml bundle is designed to allow LHC physicists that happen.  could help the arrays of numbers between 0 and 1 – essentially just the to deploy anomaly-detection algorithms in the L1 trigger field break animal pictured in red, green and blue light. We would Unsupervised learning system. This approach could practically be deployed in L1 Further reading beyond the like to have a mathematical function converting this stack Anomaly detection is a possible strategy for keeping watch trigger systems before the end of LHC Run 3 – a powerful O Knapp et al. 2021 Eur. Phys. J. Plus 136 236. l i m it s of hu m a n of arrays into a score ranging from 0 to 1. The larger the for n e w p h y s ic s w it h o ut h a v i n g t o s p e c i f y a n e x a c t s i g n a l. complement to the anomaly-detection techniques that O Cerri et al. 2019 J. High Energy Phys. 2019 36. creativity in score, the higher the probability that the image is a dog. A kind of unsuper vised ML, it involves ranking an unla- are already being considered for “offline” data analysis J Duarte et al. 2018 J. Instrum. 13 P07027. theory building The smaller the score, the higher the probability that the belled dataset from the most typical to the most atypi-

CERN COURIER SEPTEMBER/OCTOBER 2021 33 32 CERN COURIER SEPTEMBER/OCTOBER 2021

CCSepOct21_AI_v2.indd 33 04/08/2021 12:02 CCSepOct21_AI_v2.indd 32 04/08/2021 12:01 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 NEW! Best Model B35adp Alpha/Deuteron/Proton Cyclotron for Medical Radioisotope Production & Other Applications NEW! Best 6–15 MeV Proton Particle Beam: 30 MeV Energy, 500 µA Beam Current Compact/High Current/ Deuteron Particle Beam: 15 MeV Energy, 250 µA Beam Current Variable Energy Proton Alpha Particle Beam: 35 MeV Energy, 100 µA Beam Current Cyclotron 1–1000 µA extracted beam current COMING SOON! Best Model 150p Capable of producing the following isotopes: 18F, 68Ga, 89Zr, 99mTc, 11C, 13N, 15O, 64Cu, 67Ga, 111In, 124I, 225Ac, 103Pd and more! Cyclotron for Proton Therapy Simultaneuous 18F and 68Ga at high yield (Patent Pending) Up to 5 x 1013 neutrons per second from external target From 70 MeV up to 150 MeV 21 stripping foils at each stripping port for 2 minute rapid change Non-Variable Energy Dedicated for Proton Therapy with two beam lines & two treatment rooms For all Medical Treatments including: Benign & Malignant Tumors for Neurological, Eye, Head/Neck, Pediatric, Lung Cancers, NEW! Best Sub-Compact Model 200 Vascular/Cardiac/Stenosis/Ablation, etc. Self-Shielded Cyclotron with Optional Second Chemistry Module Best Particle Therapy 400 MeV ion Rapid Cycling Medical Synchrotron (iRCMS) for Proton-to-Carbon, Variable Energy Heavy Ion Therapy, with Low Energy Compact System Capable of producing: 18FDG, or without Gantries — Upgradeable Single & Multi-Room Solutions Easy to operate push-button graphic interface Na18F, 18F-MISO, 18FLT, 18F-Choline, Automated quality control testing 18F-DOPA, 18F-PSMA, 11C, 13N, Intrinsically small beams facilitating Ideal for Nuclear Cardiology and other Medical Applications 68Ga and more! Upgradeable Single Room beam delivery with precision PARTICLE & Multi-Room Solutions THERAPY NUCLEAR Small beam sizes – small magnets, PHARMACY light gantries – smaller footprint NEW! Best Model B3d Sub-Compact Low Energy Deuteron/Proton Cyclotron Highly efficient single turn extraction NUCLEAR Accelerated Deuteron Particle: 1 to 3 MeV Energy MEDICINE Efficient extraction – less shielding Accelerated Proton Particle: 1 to 6 MeV Energy Flexibility – heavy ion beam therapy Maximum Beam Current of 2µA (protons and/or carbon), beam delivery modalities System is self-shielded Small footprint (less than 5 m x 5 m) Includes sample and detector chamber Automated turn-on and turn-off control sequences PATIENT PREP Remote monitoring for diagnostics and scheduled BRACHYTHERAPY TELETHERAPY maintenance BESTBest RADIATIONMedical International, THERAPY & DIAGNOSTIC Inc. 7643 CENTERFullerton Road,ion Rapid Springfield, Cycling Medical VA Synchrotron 22153 (iRCMS) USA Racetrack www.bestcyclotron.com • www.bestabt.com • www.bestproton.com TeamBest703 451 Global 2378 Companies 800 © 2021 336* 4970Some of the www.bestmedical.com products shown are under development www.teambest.com and not available for sale currently. BCS tel: 604 681 3327 • Best ABT tel: 865 982 0098 • BPT tel: 703 451 2378

CERN_Courier_BCS-BestABT-BPT_TBG-R-logo_2pageSpread_Collection_v67_07272021_working.indd 1 7/28/21 5:11:27 PM CERN_Courier_BCS-BestABT-BPT_TBG-R-logo_2pageSpread_Collection_v67_07272021_working.indd 2 7/28/21 5:12:45 PM CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

FEATURE SCIENCE AND INDUSTRY FORGING THE FUTURE OF AI The first Sparks! Serendipity Forum at CERN will bring together world experts in artificial intelligence in a spirit of multidisciplinary collaboration. M a rk R ay ner sp oke to some of t he pa r t ic ip a nt s i n t he r un-up to t he Septemb er e vent. C Marcelloni/OPEN-PHO-ACCEL-2020-003-10

Launching the forum The CMS collaboration’s Jennifer Ngadiuba speaks to fellow Sparks! participant and machine-learning expert Michael Kagan of the ATLAS experiment (right) and Bruno Giussani (left), the global curator of the TED conference series, who will host the public Sparks! event on 18 September.

ield lines arc through the air. By chance, a cosmic ray journal Machine Learning: Science and Technology. T h e for u m Back to the future from self-driving c a r s t o s e a rc he s for e x o t ic a at t he L HC k n o c k s a n e le c t r o n off a m ole c u le. It hu r t le s a w a y, reflects the growing use of machine-learning techniques In the 1980s, AI research was dominated by code that (see p31). But many Sparks! participants think that the F crashing into other molecules and multiplying the in particle physics and emphasises the importance that emulated logical reasoning. In the 1990s and 2000s, time has come to reintegrate causal logic into AI. effect. The temperature rises, liberating a new supply of CERN and the wider community places on collaborating at tent ion t ur ned to sof tening its st rong syl log isms into electrons. A spark lights up t he dark. probabilistic reasoning. Huge strides forward in the Geneva is the home not only of CERN

This is an excellent metaphor for the Sparks! Serendipity AI is orders of magnitude past decade have rejected logical reasoning, however, S Russel l but also of the UN negotiations on For u m – a n e w a n nu a l e v e nt at C E R N d e s i g n e d t o e n c o u r a g e faster than traditional instead capitalising on computing power by letting layer lethal autonomous weapons. The interdisciplinary collaborations between experts on key Anandkumar A numerical simulations. major powers must put the evil genie scientific issues of the day. The first edition, which will On the other side of the coin, 2020 revealed unexpectedly back in the bottle before it’s too late simulations are being used to T Osbor ne fragile technological and Stuart Russell is professor of computer science The absence of causal train AI in domains such as socio-cultural infrastructures. at the University of California, Berkeley and V M i ng inference in practical robotics where real data is How we locate our coauthor of the seminal text on AI machine learning touches on very scarce conversations and research every aspect of AI research, Anima Anandkumar is Bren professor about AI in those contexts application, ethics and policy at Caltech and director of machine feels as important as the “A p u r e l y pr e d ic t i v e s y s t e m, s uc h a s t h e c u r r e nt m a c h i n e Vivienne Ming is a theoretical learning research at NVIDIA research itself learning that we have, that lacks a notion of causality, seems neuroscientist and a serial Genevieve Bell is director of the School t o b e v e r y s e v e r e l y l i m it e d i n it s a bi l it y t o s i mu l at e t h e w a y AI entrepreneur w it h d i v e r s e t e c h n olo g ic a l s e c t or s. Suc h i nt e r a c t io n s a r e of Cybernetics at the Australian National that people think,” says Nobel-prize-winning cognitive essent ial to t he long-ter m success of t he field. University and vice president at Intel p s yc h olo g i s t D a n ie l K a h n e m a n. “C u r r e nt A I i s b u i lt t o s ol v e take place from 17 to 18 September, will focus on artifi- T h e l i k e l i h o o d o f s p a r k s fly i n g d e p e n d s o n t h e w e at h er. o n e s p e c i fic t a s k, w h ic h u s u a l l y d o e s n o t i n c lud e r e a sonig cial intelligence (AI). Fifty leading thinkers will explore To take the temperature, CERN Courier spoke to a sam- upon layer of artificial neurons discern the relation- about that task,” agrees AAAI president-elect Francesca the future of AI in topical groups, with the outcomes of ple of the Sparks! participants to preview themes for the ships inherent in vast data sets. Such “deep learning” has Rossi. “Leveraging what we know about how people rea- their exchanges to be written up and published in the September event. been transformative, fuelling innumerable innovations, s on a nd b e h a ve c a n he lp bu i ld more ro bu s t, a d apt a ble a nd

36 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 37

CCSepOct21_SPARKS_v4.indd 36 04/08/2021 12:05 CCSepOct21_SPARKS_v4.indd 37 04/08/2021 12:05 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FEATURE SCIENCE AND INDUSTRY FEATURE THEORETICAL PHYSICS

generalisable AI – and also AI that can support humans in This idea of partnership, that making better decisions.” worries me. It looks to me like G o o g le’s N y a l le n g M o or o s i id e nt i fie s a n o t h e r w e a k n e s D Ka h nema n a very unstable equilibrium. If of deep-learning models that are trained with imperfect the AI is good enough to help d at a: w h e t h e r A I i s d e c id i n g w h o d e s e r v e s a lo a n or w h e t h e r the person, then pretty soon it an event resembles physics beyond the Standard Model, EMERGENCE will not need the person its decisions are only as good as its training. “What we call the ground truth is actually a system that is full of Daniel Kahneman is a renowned Many physical laws ‘emerge’ from complexity thanks to a process of data errors,” she says. cognitive psychologist and a winner of the 2002 Nobel Prize in Economics F u r t h e r m or e, s a y s i n flue nt i a l c o m p ut at io n a l n e u r o s c ien- compression. Erik Verlinde sizes up the Standard Model, gravity and intelligence t i s t To m a s o P o g g io, w e d o n’t y e t u n d e r s t a n d t h e s t at i s t ic a l as candidates for future explanation as emergent phenomena. b e h a v io u r o f d e e p -le a r n i n g a l g or it h m s w it h m at h e m at ic a l L i a o. “T h e r e’s a b a s ic m i n i mu m t h at w e n e e d t o m a k e s u r e e v e r yo n e h a s a c c e s s t o. I f w e s t a r t f r o m t h e r e, a lo t o f t h e s e AI is converging on forms of problems become more tractable.” T Poggio intelligence that are useful but Far-term ethical considerations will be even more pro- very likely not human-like fo u n d i f A I d e v e lo p s hu m a n-le v e l i nt e l l i g e n c e. W h e n S p a r k s! Tomaso Poggio is a cofounder of participants were invited to put a confidence interval on computational neuroscience and when they expect human-level AI to emerge, answers Eugene McDermott professor at MIT We need to understand ethical

S M Liao principles, rather than just list them, because then there’s precision. “There is a risk in trying to understand things a worry that we’re just doing l i k e p a r t ic le p h y s ic s u s i n g t o ol s w e d o n’t r e a l l y u n d e r s t a n d,” ethics washing – they sound he explains, also citing attempts to use artificial neural good but they don’t have any bite networks to model organic neural networks. “It seems a very S Matthew Liao is a philosopher and i r o n ic s it u at io n, a n d s o m e t h i n g t h at i s n o t v e r y s c ie nt i fic.” the director of the Center for Bioethics St u a r t R u s s e l l, o n e o f t h e w orld’s m o s t r e s p e c t e d v oic e s at New York University on AI, echoes Poggio’s concerns, and also calls for a greater focus on controlled experimentation in AI research ranged from [2050, 2100] at 90% confidence to [2040,∞ ] it sel f. “In stead of t r y i ng to comp ete b et we en De ep M i nd at 99% confidence. Other participants said simply “in 100 and OpenAI on who can do the biggest demo, let’s try to y e a r s” or n o t e d t h at t h i s i s “d e l i g ht f u l l y t h e w r o n g q ue s t io n” answer scientific questions,” he says. “Let’s work the way as it’s too human-centric. But by any estimation, talking scientists work.” about AI cannot w ait. “With Sparks!, we plan to give a nudge to serendip- iStock/georgeclerk Good or bad? ity in interdisciplinary science by inviting experts from Emergent simplicity The evolution of a murmuration of starlings cannot be described by following the motion of any individual bird. Though most Sparks! participants firmly believe that AI a range of fields to share their knowledge, their visions benefits humanity, ethical concerns are uppermost in and their concerns for an area of common interest, first article physics is at its heart a reductionistic endeav- very small, but by its very nature skips across orders of their minds. From social-media algorithms to autono- with each other, and then with the public,” says Joachim our that tries to reduce reality to its most basic magnitude in scale. It is even evident, often mesmeris- Mnich, CERN’s director for research and computing. “For P building blocks. This view of nature is most evident ingly so, at scales much larger than atoms or elementary We always had privacy the first edition of Sparks!, we’ve chosen the theme of AI, in the search for a theory of everything – an idea that particles, for example in the murmurations of a flock of

N Moorosi violation, we had people w h ic h i s a s i m p or t a nt i n p a r t ic l e p h y s ic s a s it i s i n s o c ie t y at is nowadays more common in popularisations of phys- birds – a phenomenon that is impossible to describe by being blamed falsely for large. Sparks! is a unique experiment in interdisciplinarity, ics than among physicists themselves. If discovered, all following the motion of an individual bird. Another striking crimes they didn’t do, we had physical phenomena would follow from the application of example may be intelligence. The mechanism by which mis-diagnostics, we also had Only a multi-stakeholder and its fundamental laws. artificial intelligence is beginning to emerge from the F Rossi Rossi F false news, but what AI has multi-disciplinary approach A complementary perspective to reductionism is that of complexity of underlying computing codes shows sim- done is amplify all this, and can build an ecosystem of trust e m e r g e n c e. E m e r g e n c e s a y s t h at n e w a n d d i ffe r e nt k i n d s ilarities with emergent phenomena in physics. One can make it bigger around AI. Education, cultural of phenomena arise in large and complex systems, and that argue that intelligence, whether it occurs naturally, as change, diversity and governance these phenomena may be impossible, or at least very hard, in humans, or artificially, should also be viewed as an Nyalleng Moorosi is a research software to derive from the laws that govern their basic constitu- emergent phenomenon. engineer at Google and a founding are equally as important as member of Deep Learning Indaba making AI explainable, robust ents. It deals with properties of a macroscopic system that and transparent have no meaning at the level of its microscopic building Data compression mous weapons, current AI overwhelmingly lacks com- blocks. Good examples are the wetness of water and the Renormalisable quantum field theory, the foundation of Francesca Rossi co-leads the World passion and moral reasoning, is inflexible and unaware of Economic Forum Council on AI for superconductivity of an alloy. These concepts don’t exist the SM, works extraordinarily well. The same is true of its fallibility, and cannot explain its decisions. Fairness, humanity and is IBM AI ethics global at the level of individual atoms or molecules, and are very general relativity. How can our best theories of nature be so inclusivity, accountability, social cohesion, security and leader and the president-elect of AAAI difficult to derive from the microscopic laws. successful, while at the same time being merely emergent? i nt e r n at io n a l l a w a r e a l l i m p a c t e d, d e e p e n i n g l i n k s b e t w e e n As physicists continue to search for cracks in the Stand- Perhaps these theories are so successful precisely because THE AUTHOR THE AUTHOR the ethical responsibilities of individuals, multinational which I hope will inspire continued innovative uses of AI ard Model (SM) and Einstein’s general theory of relativity, they are emergent. Erik Verlinde Mark Rayner corporations and governments. “This is where I appeal to in high-energy physics. I invite the whole community to could these natural laws in fact be emergent from a deeper As a warm up, let’s consider the laws of thermodynam- University of deputy editor. the human-rights f r a m e wor k,” s a y s p h i lo s o p h e r S M at t h e w get involved in the public event on 18 September.”  reality? And emergence is not limited to the world of the ics, which emerge from the microscopic motion of many Amsterdam.

38 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 39

CCSepOct21_SPARKS_v4.indd 38 04/08/2021 12:06 CCSepOct21_EMERGENCE_v3.indd 39 04/08/2021 12:07 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FEATURE THEORETICAL PHYSICS FEATURE THEORETICAL PHYSICS

r e d e fi n i n g t h e s pi n s t o b e t h e a v e r a g e o f t h e or i g i nalspis. even empty space develops a horizon, from behind which Holographic renormalisation One then rescales the coarser lattice so that the distance light rays will not be able to catch up. These horizons between lattice points takes the old value, and repeats also carry a temperature and entropy, and obey the same The AdS/CFT correspondence black-hole event horizon one Planck this step many times. A key insight was that, for quantum thermodynamic laws as black-hole horizons. incorporates a principle called area

M Weiler/arXiv:2106.06020 statistical systems that are close to a phase transition, It was shown by Stephen Hawking that the thermal “holography”: the gravitational physics you can take a continuum limit in which the expectation radiation emitted from a black hole originates from pair inside a region of space emerges from values of t he spins tur n into t he local quantum fields on creation near the black-hole horizon. The properties of the a microscopic description that, just the continuum spacetime. pair of particles, such as spin and charge, are undetermined like a hologram, lives on a space with This procedure is analogous to the compression algo- due to quantum uncertainty, but if one particle has spin up one less dimension and thus can be rithms used in machine learning. Each renormalisation (or positive charge), then the other particle must have spin viewed as living on the boundary of the one unit of entropy step creates a new layer, and the algorithm that is applied down (or negative charge). This means that the particles are spacetime region. The extra dimension between two layers amounts to a form of data compression. quantum entangled. Quantum entangled pairs can also be of space emerges together with the Black-hole maths Jacob Bekenstein and The goal is similar: you only keep the information that found in flat space by considering accelerated obser vers. gravitational force through a process Stephen Hawking related entropy to the area is required to describe the long-distance and low-energy Crucially, even the vacuum can be entangled. By sep- called “holographic renormalisation”. of the event horizon. Credit: J Bekenstein behav iour of t he system in t he most efficient w ay. arating spacetime into two parts, you can ask how much One successively adds new layers of So quantum field theory can be seen as an effective entanglement there is between the two sides. The answer spacetime. Each layer is obtained not have a boundary. Instead, it has emergent description of one of a large universality class to this was found in the last decade, through the work of from the previous layer through a cosmological horizon whose size is Gauge molecules. These laws are not fundamental but are derived of many possible underlying microscopic theories. But many theorists, and turns out to be rather surprising. If “coarse-graining”, in a similar way determined by the rate of the Hubble symmetry by statistical averaging – a huge data compression in which w h at a b o ut t h e S M s p e c i fic a l l y, a n d it s p o s s i ble s up e r sym- you consider two regions of space that are separated by a to both renormalisation in quantum expansion. One proposed explanation To ensure that a the individual motions of the microscopic particles are metric extensions? Gauge fields are central ingredients two-dimensional surface, the amount of quantum entan- field theory anddata-compression for this qualitative difference is neural network compressed into just a few macroscopic quantities such as of the SM and its extensions. Could gauge symmetries glement between the two sides turns out to be precisely algorithms in machine learning. that, unlike for negatively curved recognises a temperature. As a result, the laws of thermodynamics are and their associated forces emerge from a microscopic given by the Bekenstein–Hawking entropy formula: it is Unfortunately, our universe is spacetimes, the microscopic quantum pixelated lizard, its universal and independent of the details of the microscopic description in which there are no gauge fields? Similar equal to a quarter of the area of the surface measured in not described by a negatively curved state of our universe is not unique, algorithm should theory. This is true of all the most successful emergent questions can also be asked about the gravitational force. Planck units. spacetime. It is much closer to a but secretly carries a lot of quantum be invariant theories; they describe universal macroscopic phenomena Could the curvature of spacetime be explained from an This raises an interesting prospect: if the microscopic so-called de Sitter spacetime, which information. The amount of this under rotations. whose underlying microscopic descriptions may be very emergent perspective? quantum data of our universe may be thought of as many has a positive curvature. The main quantum information can once again be different. For instance, two physical systems that undergo String theory seems to indicate that this is indeed pos- entangled qubits, could our current theories of spacetime, difference between de Sitter space counted by an entropy: the Bekenstein– a second-order p h a s e t r a n s it io n, w h i le b e i n g v e r y d i ffe r- sible, at least theoretically. While initially formulated in particles and forces emerge via data compression? Space, and the negatively curved anti-de Hawking entropy associated with the ent microscopically, often obey exactly the same scaling terms of vibrating strings moving in space and time, it for example, could emerge by forgetting the precise way Sitter space is that de Sitter space does cosmological horizon. laws, and are at the critical point described by the same became clear in the 1990s that string theory also contains in which all the individual qubits are entangled, but only emergent theory. In other words, an emergent theory can many more extended objects, known as “branes”. By study- preserving the information about the amount of quantum often be derived from a large universality class of many ing the interplay between branes and strings, an even more entanglement present in the microscopic quantum state. the Bekenstein–Hawking entropy, and the derivation of underlying microscopic theories. microscopic theoretical description was found in which the This compressed information would then be stored in the the Einstein equations from entanglement, are best under- Entropy is a key concept here. Suppose that you try to coordinates of space and time themselves start to dissolve: form of the areas of certain surfaces inside the emergent stood. I have also contributed to these developments in store the microscopic data associated with the motion of instead of being described by real numbers, our familiar curved spacetime. a paper in 2010 that emphasised the role of entropy and some particles on a computer. If we need N bits to store (x, y, z) coordinates are replaced by non-commuting matri- In this description, gravity would follow for free, information for the emergence of the gravitational force. all that information, we have 2N possible microscopic ces. At low energies, these matrices begin to commute, expressed in the curvature of this emergent spacetime. Over the last decade a lot of progress has been made in our states. The entropy equals the logarithm of this number, and give rise to the normal spacetime with which we are What is not immediately clear is why the curved spacetime understanding of these connections, in particular the deep and essentially counts the number of bits of information. familiar. In these theoretical models it was found that would obey the Einstein equations. As Einstein showed, connection between gravity and quantum entanglement. Entropy is therefore a measure of the total amount of data both gauge forces and gravitational forces appear at low the amount of curvature in spacetime is determined by the Quantum information has taken centre stage in the most that has been compressed. In deriving the laws of thermo- energies, while not existing at the microscopic level. amount of energy (or mass) that is present. It can be shown recent theoretical developments. dynamics, you throw away a large amount of microscopic While these models show that it is theoretically possible that his equations are precisely equivalent to an application data, but you at least keep count of how much information for gauge forces to emerge, there is at present no emergent of the first law of thermodynamics. The presence of mass Emergent intelligence has been removed in the data-compression procedure. theory of the SM. Such a theory seems to be well beyond us. or energy changes the amount of entanglement, and hence But what about viewing the even more complex problem Gravity, however, being universal, has been more amenable the area of the surfaces in spacetime. This change in area of human intelligence as an emergent phenomenon? Since Emergent quantum field theory to emergence. can be computed and precisely leads to the same space- s c ie nt i fic k n o wle d g e i s c o n d e n s e d a n d s t or e d i n o u r c u rent Successful One of the great theoretical-physics paradigm shifts of the time curvature that follows from the Einstein equations. theories of nature, the process of theory formation can itself emergent 20th century occurred when Kenneth Wilson explained Emergent gravity The idea that gravity emerges from quantum entangle- be viewed as a very efficient form of data compression: it theories the emergence of quantum field theory through the appli- In the early 1970s, a group of physicists became interested ment goes back to the 1990s, and was first proposed by Ted only keeps the information needed to make predictions describe cation of the renormalisation group. As with thermody- in the question: what happens to the entropy of a thermo- Jacobson. Not long afterwards, Juan Maldacena discovered about reproducible events. Our theories provide us with a namics, renormalisation compresses microscopic data dynamic system that is dropped into a black hole? The sur- that general relativity can be derived from an underlying way to make predictions with the fewest possible number universal into a few relevant parameters – in this case, the fields prising conclusion was that black holes have a temperature microscopic quantum theory without a gravitational force. of free parameters. macroscopic and interactions of the emergent quantum field theory. and an entropy, and behave exactly like thermodynamic His description only works for infinite spacetimes with The same principles apply in machine learning. The way phenomena W i l s o n d e m o n s t r at e d t h at q u a nt u m fie ld t h e or ie s a p p e ar s y s t e m s. I n p a r t ic u l a r, t h e y o b e y t h e fi r s t l a w o f t h e rmo- negative curvature called anti-de Sitter (or AdS–) space, as an artificial-intelligence machine is able to predict whether whose naturally as an effectivelong-distance and low-energy dynamics: when the mass of a black hole increases, its opposed to the positive curvature we measure. The micro- an image represents a dog or a cat is by compressing the underlying description of systems whose microscopic definition is (Bekenstein–Hawking) entropy also increases. scopic description then takes the form of a scale-invariant microscopic data stored in individual pixels in the most microscopic given in terms of a quantum system living on a discretised The correspondence between the gravitational laws and quantum field theory – aso-called c o n for m a l fie ld t h e or y efficient way. This decision cannot be made at the level of descriptions spacetime. As a concrete example, consider quantum spins the laws of thermodynamics does not only hold near black (CFT) – that lives on the boundary of the AdS–space (see individual pixels. Only after the data has been compressed may be very on a lattice. Here, renormalisation amounts to replacing holes. You can artificially create a gravitational field by “Holographic renormalisation” panel). It is in this context and reduced to its essence does it becomes clear what the different the lattice by a coarser lattice with fewer points, and a c c e le r at i n g. For a n o b s e r v e r w h o c o nt i nue s t o a c c e le r at e, that the connection between vacuum entanglement and picture represents. In this sense, the dog/cat-ness of a pic-

40 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 41

CCSepOct21_EMERGENCE_v3.indd 40 04/08/2021 12:07 CCSepOct21_EMERGENCE_v3.indd 41 04/08/2021 12:08 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FEATURE THEORETICAL PHYSICS FEATURE LHCb UPGRADE

ture is an emergent property. This is even true for the way conventional “brute force” chess program by coming up humans process the data collected by our senses. It seems with smarter ideas and showing a deeper understanding easy to tell whether we are seeing or hearing a dog or a cat, of the game. Top grandmasters use its ideas in their own but underneath, and hidden from our conscious mind, our games at the highest level. brains perform a very complicated task that turns all the In its basic material design, an artifi cial-intelligence BUILDING THE neural data that come from our eyes and ears into a signal machine looks like an ordinary computer. On the other that is compressed into a single outcome: it is a dog or a cat. hand, it is practically impossible to explain all aspects of Can intelligence, whether artificial or human, be human intelligence by starting at the microscopic level of explained from a reductionist point of view? Or is it an the neurons in our brain, let alone in terms of the elemen- emergent concept that only appears when we consider a tary particles that make up those neurons. Furthermore, FUTURE OF LHCb complex system built out of many basic constituents? There the intellectual capability of humans is closely connected are arguments in favour of both sides. As human beings, to the sense of consciousness, which most scientists would The brand-new “SciFi” tracker and upgraded ring-imaging Cherenkov detectors our brains are hard-wired to observe, learn, analyse and agree does not allow for a simple reductionist explanation. solve problems. To achieve these goals the brain takes the Emergence is often summarised with the slogan that are currently being installed are vital for the higher LHC luminosities ahead, large amount of complex data received via our senses and “the whole is more than the sum of its parts” – or as write Christoph Frei, Silvia Gambetta and Blake Leverington. reduces it to a very small set of information that is most condensed-matter theorist Phil Anderson put it, “more

relevant for our purposes. This capacity for effi cient data is diff erent”. It counters the reductionist point of view, M Brice compression may indeed be a good defi nition for intel- reminding us that the laws that we think to be funda- ligence, when it is linked to making decisions towards mental today may in fact emerge from a deeper underlying Emergence reaching a certain goal. Intelligence defi ned in this way is reality. While this deeper layer may remain inaccessible is often e x h i bit e d i n hu m a n s, b ut c a n a l s o b e a c h ie v e d a r t i fi c i a l l y. to experiment, it is an essential tool for theorists of the summarised Artifi cially intelligent computers beat us at problem mind and the laws of physics alike.  with the solving, pattern recognition and sometimes even in what slogan “the appears to be “generating new ideas”. A striking example Further reading whole is more is DeepMind’s AlphaZero, whose chess rating far exceeds G Musser 2018 Nature 557 S3–S6. than the sum that of any human player. Just four hours after learning M Van Raamsdonk 2020 Science 370 198. of its parts” the rules of chess, AlphaZero was able to beat the strongest M Weiler et al. 2021 arXiv:2106.06020.

Virtual Lab Series for CERN with COMSOL Multiphysics® Sept 28, Oct 26 & Nov 30 2021 ONLINE t was once questioned whether it would be possible to at departures from the Standard Model. Fast track successfully operate an asymmetric “forward” detector At the core of the LHCb upgrade project are new detec- Planes of LHCb’s Join fellow engineers and simulation Iat a hadron collider. In such a high-occupancy envi- tors capable of sustaining an instantaneous luminosity SciFi tracker being specialists for a free educational event. ronment, it is much harder to reconstruct decay vertices up to five times that seen at Run 2, and which enable assembled at and tracks than it is at a lepton collider. Following its a pioneering software-only trigger that will enable CERN in late 2019. comsol.ch/c/c6nf successes during LHC Run 1 and Run 2, however, LHCb LHCb to process signal data in an upgraded computing has rewritten the forward-physics rulebook, and is now farm at the frenetic rate of 40 MHz. The vertex locator preparing to take on bigger challenges. (VELO) will be replaced with a pixel version (CERN Courier THE AUTHORS During Long Shutdown 2, which comes to an end early July/August 2021 p36), the upstream silicon-strip tracker next year, the LHCb detector is being almost entirely will be replaced with a lighter version (the UT) located Christoph Frei rebuilt to allow data to be collected at a rate up to 10 times closer to the beamline, and the electronics for LHCb’s CERN, Silvia Gambetta higher during Run 3 and Run 4 (CERN Courier January/ muon stations and calorimeters are being upgraded for University of February 2019 p34). This will improve the precision of 40 MHz readout. Edinburgh and numerous world-best results, such as constraints on Recently, three further detector systems key to dealing Blake Leverington the angles of the CKM triangle, while further scrutinis- with the higher occupancies ahead were lowered into the Heidelberg ing intriguing results in B-meson decays, which hint LHCb cavern for installation: the upgraded ring-imaging University.

42 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 43

CCSepOct21_EMERGENCE_v3.indd 42 04/08/2021 12:08 CCSepOct21_LHCbUPGRADE_v2.indd 43 03/08/2021 16:35 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

FEATURE LHCb UPGRADE FEATURE LHCb UPGRADE och / itrzyk z r Piet v/M he Hopc P B Leverington LHCb S Gambetta y side view ECAL HCAL M5 M3 M4 5 m magnet SciFi RICH2 M2 RICH1 tracker UT

vertex locator (VELO)

5 m 10 m 15 m 20 m z

Asymmetric Cherenkov detectors RICH1 and RICH2 for sharper particle decided to use a new single-phase liquid coolant called Tracking tall anatomy identification, and the brand new “SciFi” (scintillating Novec-649 from 3M for its non-toxic properties and low Completed SciFi

LHCb’s successive fibre) trac ker. greenhouse warming potential (GWP = 1). Historically, C6F14 C-frames in the detector layers – which has a GWP = 7400 – was the thermo-transfer flu id assembly hall at located SciFi tracking of choice. Although several challenges had to be faced Point 8. downstream The components of LHCb’s SciFi tracker may not seem in learning how to work with the new fluid, wider use from the futuristic at first glance. Its core elements are constructed of Novec-649 and similar products could contribute sig- interaction point from what is essentially paper, plastic, some carbon fibre nificantly to the reduction of CERN’s carbon footprint. in the heart of and glue. However, its materials components conceal Additionally, since the narrow envelope of the tracking the VELO. advanced technologies which, when coupled together, pro- stations precludes the use of standard foam insulation d uc e a v e r y l i g ht a n d u n i for m, high-performance detector of the coolant lines, a significant engineering effort has t hat i s ne e de d to cop e w it h t he h ig her numb er of pa r t ic le been required to vacuum insulate the 48 transfer lines tracks expected during Run 3. from the 24 rows of SiPMs and 256 cold-bars where leaks Located behind the LHCb magnet (see “Asymmetric are possible at every connection. anatomy” image), the SciFi represents a challenge, not To date, LHCb collaborators have tirelessly assembled o n l y d ue t o it s c o m ple x it y, b ut a l s o b e c a u s e t h e t e c h n olo g y and tested nearly half of the SciFi tracker above ground, – plastic scintillating fibres and silicon photomultiplier w h e r e o n l y t wo d e fe c t i v e c h a n n e l s o ut o f t h e 262,1 4 4 t e s t e d a r r ay s – h a s ne ve r b e e n us e d for suc h a l a rge a rea i n suc h SciFi spools Scintillating-fibre mat production at EPFL, RICH2 to go One of the two photon detector planes of RICH2, in the full signal chain were unrecoverable. Four out of 12 a harsh radiation environment. Many of the underlying Lausanne, in 2017. fully assembled before installation. “C-frames” c ont a i n i ng t he fibre mo du le s (s e e “T r ac k i ng t e c h n olo g ie s h a v e b e e n p u s h e d t o t h e e x t r e m e d u r i n g t h e tall” image) are now installed and waiting to be connected p a s t d e c a d e t o a l lo w t h e S c i F i t o s uc c e s s f u l l y o p e r at e u n d e r custom-made s i l ic o n p h o t o mu lt i pl ie r s (Si PM s). Si PM s h a v e LHCb’s SciFi tracker is the firstlarge-scale use of SiPMs and commissioned, with a further two installed in mid-July. LHC conditions in an affordable and effective way. become a strong alternative to conventional photomul- for tracking, and takes advantage of improvements in the The remaining six will be completed and installed before More t h a n 1 1,000 k m of 0.25 mm-diameter p oly s t y re ne tiplier tubes in recent years, due to their smaller channel technology in the 10 years since the SciFi was proposed. The the start of operations early next year. fi br e w a s d e l i v e r e d t o C E R N b e for e u n d e r g oi n g m e t ic ulos s i z e s, e a s ie r o p e r at io n a n d i n s e n s it i v it y t o m a g n e t ic fie ld s. photon-detection efficiency of SiPMs has nearly doubled q u a l it y c h e c k s. E x c e s s i v e d i a m e t e r v a r i at io n s w e r e r e m o v e d This makes them ideal to read out the higher number of thanks to improvements in the design and production New riches to prevent disruptions of the closely packed fibre matrix c h a n n e l s n e c e s s a r y t o id e nt i f y s e p a r at e b ut n e a r b y t r a c k s of the underlying pixel structures, while the probability One of the key factors in the success of LHCb’s produced during the winding procedure, and clear improve- in LHCb during Run 3. of crosstalk between the pixels (which creates multiple flavour-physics programme is its ability to identify charged ments from the early batches to the production phase were The width of the SiPM channels, 0.25 mm, is designed fake signals by causing a single pixel to randomly fire particles, which reduces the background in selected final m a d e b y w or k i n g c lo s e l y w it h t h e i n d u s t r i a l m a nu f a c t u r e r. to match that of the fibres. Though they need not align without incident light following radiation damage) has st ates a nd a ssi st s i n t he flavour t ag g i ng of b qua rk s. T wo From the raw fibres, nearly 1400multi-layered fi br e m at s perfectly, this provides a better separation power for track- been reduced from more than 20% to a few percent by the ring-imaging Cherenkov (RICH) detectors, RICH1 and w e r e wo u n d i n fo u r o f t h e L HC b c ol l a b or at io n’s i n s t it ut e s i ng t ha n t he pre v iously use d 5 m m g a s st r aw t ub es i n t he introduction of microscopic trenches between the pixels. RICH2, located upstream and downstream of the LHCb (see “SciFi spools” image), before being cut and bonded outer regions of the detector, while providing a similar The dark-single-pixel firing rate can also be reduced by magnet 1 and 10 m away from the collision point, provide One of the i n m o d u le s, t e s t e d, a n d s h i p p e d t o C E R N w h e r e t h e y w e r e performance to the silicon-strip tracker. The tiny chan- cooling the SiPM. Together, these two methods greatly excellent particle identification over a very wide momen- key factors in assembled with the cold boxes. The SciFi tracker contains nel size results in over 524,288 SiPM channels to collect reduce the number of fake-signal clusters such that the tum range. They comprise a large volume of fluorocarbon the success 1 2 8 s t i ff a n d r o b u s t5 × 0.5 m2 m o d u le s m a d e o f e i g ht m at s light from 130 m of fibre-mat e d g e s. A c u s t o m A SIC, c a l le d tracker can effectively function after several years of oper- gas (the radiator), in which photons are emitted by charged of LHCb’s bonded with two fire-resistant h o n e yc o m b a n d carbon-fibre the PACIFIC, outputs two bits per channel based on three ation in the LHCb cavern. particles travelling at speeds higher than the speed of light flavour- pa nels, a long w it h some mec ha n ics a nd a light-injection signal-amplitude t h r e s h old s. A field-programmable g at e The LHCb collaboration assembled commercial SiPMs on in the gas; spherical and flat mirrors to focus and reflect 2 physics system. In total, the design produces nearly 320 m of array (FPGA) assigned to each SiPM then groups these flex cables and bonded them in groups of 16 to a 0.5 m-long this Cherenkov light; and two photon-detector planes detector surface over the 12 layers of the tracking stations. signals into clusters, where the location of each cluster is 3D-printed titanium cooling bar to form precisely assem- where the Cherenkov rings are detected and read out by programme The scintillating fibres emit photons atblue-green w a v e- sent to the computing farm. Despite clustering and noise bled photodetection units for the SciFi modules. By cir- the front-end electronics. is its ability lengths when a particle interacts with them. Secondary suppression, this still results in an enormous data rate culating a coolant at a temperature of -50 °C through the The original RICH detectors are currently being refur- to identify s c i nt i l l at or d y e s a d d e d t o t h e p ol y s t y r e n e a m pl i f y t h e l i g ht of 20 Tb/s – nearly half of the total data bandwidth of the cold bar, the dark-noise rate was reduced by a factor of bished to cope with the more challenging data-taking charged and shift it to longer wavelengths so it can be read out by upgraded LHCb detector. 60. Furthermore, in a first for a CERN experiment, it was conditions of Run 3, requiring a variety of technological particles

44 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 45

CCSepOct21_LHCbUPGRADE_v2.indd 44 03/08/2021 16:36 CCSepOct21_LHCbUPGRADE_v2.indd 45 03/08/2021 16:36 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM WE ENABLE UNMATCHED DATA ACQUISITION PERFORMANCE. FEATURE LHCb UPGRADE BUY WITH CONFIDENCE FROM A TRUSTED BIG PHYSICS SUPPLIER.

To keep the new RICHes as compact as possible, the host- C Frei ing mechanics has been designed to provide both struc- tural support and active cooling. Recent manufacturing techniques have enabled us to drill two 6 mm-diameter ducts over a length of 1.5 m into the spine of the sup- port, through which a coolant (the more environmentally friendly Novec649, as in the SciFi tracker) is circulated. Each element of the opto-electronics chain has been produced and fully validated within a dedicated quality-assurance programme, allowing the position of the photon detectors and their operating conditions to be fine-tuned across the RICH detectors. In February, the first photon-detector pl a n e o f R IC H 2 (s e e “R IC H 2 t o g o” i m a g e) b e c a m e t h e fi r s t active element of the LHCb upgrade to be installed in the cavern. The two planes of RICH2, located at the sides of the beampipe, were commissioned in early summer and will see fi r s t C h e r e n k o v l i g ht d u r i n g a n L HC b e a m t e s t i n O ctober. RICH1 presents an even bigger challenge. To reduce the number of photons in the hottest region, its optics have been redesigned to spread the Cherenkov rings over a larger surface. The spatial envelope of RICH1 is also constrained by its magnetic shield, demanding even more compact When Compromise is Not an Option. mechanics for the photon-detector planes. To accommo- date the new design of RICH1, a new gas enclosure for the radiator is needed. A volume of 3.8 m3 of C F is enclosed 4 10 High-Performance Digitizers for in an aluminium structure directly fastened to the VELO tank on one side and sealed with a low-mass window on the other, with particular effort placed on building aleak-less Big Physics Applications system to limit potential environmental impact. Installing these fragile components in a very limited space has been Digitizers from Teledyne SP Devices utilize patented a delicate process, and the last element to complete the gas-enclosure sealing was installed at the beginning of June. calibration technology, the latest data converters, and Cherenkov challenges to be overcome. The photon detection system, The optical system is the final element of the RICH1 state-of-the-art FPGAs in order to achieve an unrivaled 2 curves for example, has been redesigned to adapt to the highly mechanics. The ~2 m spherical mirrors placed inside the combination of high resolution and sampling rate. Their The spherical non-uniform occupancy expected in the RICH system, g a s e n c lo s u r e a r e m a d e o f c a r b o n fi br e c o m p o s it e t o l i m it mirrors of the running from an unprecedented peak occupancy of ~35% the material budget (see “Cherenkov curves” image), while versatility makes them ideal for applications such as new RICH1. in the central region of RICH1 down to 5% in the peripheral the two 1.3 m2 pl a n e s o f fl at m i r r or s a r e m a d e o f b or o s i l icate beam position monitoring, Thomson scattering plasma region of RICH2. Two types of 64-channel multi-anode glass for high optical quality. All the mirror segments diagnostics, and more. photomultiplier tubes (MaPMTs) have been selected for the are individually coated, glued on supports and finally task which, thanks to their exceptional quantum efficiency aligned before installation in the detector. The full RICH1 in the relevant wavelength range, are capable of detecting installation is expected to be completed in the autumn, Supported features include: single photons while providing excellent spatial resolution followed by the challenging commissioning phase to tune and very low background noise. These are key requirements the operating parameters to be ready for Run 3. • Up to 10 GSPS sampling rate with 14 bits resolution to allow pattern-recognition algorithms to reconstruct • Open FPGA for custom real-time signal processing Cherenkov rings even in the high-occupancy region. Surpassing expectations More than 3000 MaPMT units, for a total of 196,608 In its first 10 years of operations, the LHCb experiment • Multiple form factors including MTCA.4, PXIe, and PCIe channels, are needed to fully instrument both upgraded has already surpassed expectations. It has enabled phys- RICH detectors. The already large active area (83%) of the icists to make numerous important measurements in the • Multi-channel synchronization capabilities devices has been maximised by arranging the units in heavy-flavour sector, including the first observation 0 + – • White Rabbit synchronization (MTCA.4 only) a compact and modular “elementary cell” containing a of the rare decay Bs → µ µ , precise measurements of custom-developed, radiation-hard eight-channel ASIC quark-mixing parameters, the discovery of CP violation • Peer-to-peer streaming to GPU (PCIe only) called the Claro chip, which is able to digitise the MaPMT in the charm sector, and the observation of more than 50 signal at a rate of 40 MHz. The readout is controlled by new hadrons including tetraquark and pentaquark states. • Application-specific firmware shortens design time FPGAs connected to around 170 channels each. The prompt However, many crucial measurements are currently statis- nature of Cherenkov radiation combined with the per- tically limited, including those underpinning the so-called formance of the new opto-electronics chain will allow fl a v o u r a n o m a l ie s (p15). To g e t h e r w it h t h e t r a c k e r, t rige the RICH systems to operate within the LHC’s 25 ns time and other upgrades taking place during LS2, the new SciFi window, dictated by the bunch-crossing period, while and revamped RICH detectors will put LHCb in prime posi- applying a time-gate of less than 6 ns to provide back- tion to explore these and other searches for new physics ground rejection. for the next 10 years and beyond.  Learn more on our website www.spdevices.com 46 CERN COURIER SEPTEMBER/OCTOBER 2021

CCSepOct21_LHCbUPGRADE_v2.indd 46 03/08/2021 16:37 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM OPINION VIEWPOINT Partner with the experts in advanced Silicon Drift Detectors Designing an AI physicist Science at the speed of light Jesse Thaler argues that particle D Dominguez/CERN phy sic i st s must go b e yond de ep lea r n i ng a nd desig n A I c apable of de ep t h i n k i ng. With over 40 years’ experience in XRF-detector R&D and manufacturing, Hitachi High-Tech detector products are used worldwide. Proven reliability and radiation hardness, low power Can we trust physics decisions made by machines? In recent applications of arti- J Charney consumption and competitive pricing make Hitachi detectors the complete package. From at ficial intelligence (AI) to particle phys- our production facilities in the USA and Finland, our teams offer flexibly scalable manufacturing ics, we have partially sidestepped this question by using machine learning to capabilities and agile R&D to deliver the custom solutions to your requirements. augment analyses, rather than replace them. We have gained trust in AI deci- Now, our expanded detector product range includes the SAANA as well as the Vortex SDD; sions through careful studies of “control r e g io n s” a n d p a i n s t a k i n g nu m e r ic a l s i m- offering a wider choice of designs, fully customizable to your needs: u l a t i o n s. A s o u r p h y s i c s a m bit i o n s g r o w, Jesse Thaler h o w e v e r, w e a r e u s i n g “d e e p e r” n e t w or k s is a professor at MIT with more layers and more complicated SAANA SDD Vortex SDD and director of the a r c h it e c t u r e s, w h i c h a r e d i ffic u l t t o v al- Resolution <140 eV <140 eV US National Science id a t e i n t h e t r a d it io n a l w a y. A n d t o m it i- Best of both worlds We need to merge the insights gained from artificial intelligence Foundation’s gate 10 to 100-fold increases in computing and physics intelligence. Sensor Thickness 1.0 mm 0.5, 1.0 and 2.0 mm Institute for costs, we are planning to fully integrate Artificial AI into data collection, simulation and It is equally important, however, that l i k e a m a c h i n e a n d t h i n k i n g l i k e a p h y s- Sensor sizes 50 mm2 50 to 100mm2 Intelligence and analysis at t he high-luminosit y LHC. we physicists teach ourselves how to icist. Recently, computer scientists and Fundamental To build trust in AI, I believe we need t hin k li ke a mac hine. physicists have begun to discover that Input Count rates 2000 Kcps >5,000 Kcps Interactions. to teac h it to t hin k li ke a physicist. Modern AI tools yield results that are t h e a p p a r e nt c o m pl e x it y o f d e e p l e a r n i n g I am the director of the US National often surprisingly accurate and insight- m ay m a sk a n emergent si mpl ic it y. Th i s Science Foundation’s new Institute for ful, but sometimes unstable or biased. idea is familiar from statistical physics, Artificial Intelligence and Fundamen- This can happen if the problem to be where the interactions of many atoms tal Interactions, which was founded last solved is “underspecified”, meaning that or molecules can often be summarised y e a r. O u r g o a l i s t o f u s e a d v a n c e s i n d e e p we have not provided the machine with i n t e r m s o f s i m pl e r e m e r g e nt pr o p e r t i e s learning with time-tested strategies for a complete list of desired behaviours, of materials. In the case of deep learn- “d e e p t h i n k i n g” i n t h e p h y s ic a l s c ie n c e s. such as insensitivity to noise, sensible ing, as the width and depth of a neural M a n y pr o m i s i n g o p p or t u n it i e s a r e o p e n ways to extrapolate and awareness of network grows, its behaviour seems to to us. Core principles of fundamental u n c e r t a i nt ie s. A n e v e n m or e c h a l l e n g i n g be describable in terms of a small num- physics such as causality and spacetime situation arises when the machine can ber of emergent parameters, sometimes symmetries can be directly incorporated identify multiple solutions to a problem, just a handful. This suggests that tools into the structure of neural networks. but lacks a guiding principle to decide from statistical physics and quantum Symbolic regression can often trans- w h i c h i s m o s t r o b u s t. B y t h i n k i n g l i k e a field theory can be used to understand late solutions learned by AI into com- machine, and recognising that modern AI dynamics, and yield deeper insights pact, human-interpretable equations. AI solves problems through numerical into their power and limitations. In experimental physics, it is becom- optimisation, we can better understand Ultimately, we need to merge the SAANA SDD SAANA TO8 VORTEX ME-7 SDD ing possible to estimate and mitigate the intrinsic limitations of training neu- insights gained from artificial intel- systematic uncertainties using AI, even ral networks with finite and imperfect ligence and physics intelligence. If we when there are a large number of nui- datasets, and develop improved opti- d o n’t e x pl oit t h e f u l l p o w e r o f A I, w e w i l l sance parameters. In theoretical physics, misation strategies. By thinking like a not maximise the discovery potential Find out more about partnering with Hitachi for your detector supply: we are finding ways to merge AI with machine, we can better translate first of the LHC and other experiments. But traditional numerical tools to satisfy principles, best practices and domain if we don’t build trustable AI, we will Hitachi High-Tech Science America, Inc. stringent requirements that calculations knowledge from fundamental physics lack scientific rigour. Machines may To build trust b e e x a c t a n d r e pr o d u c i bl e. H i g h- e n e r g y into the computational language of AI. ne ver t h i n k l i ke hum a n phy sic i st s, a nd 20770 Nordhoff St. Chatsworth, CA 91311 in AI, we need physicists are well positioned to develop Beyond these innovations, which human physicists will certainly never t r ust wor t hy A I t h at c a n b e s c r ut i n i s e d, echo the logical and algorithmic AI that match the computational ability of AI, www.hitachi-hightech.com/hhs-us/ to teach it to verified and interpreted, since the preceded the deep-learning revolution b ut t o g e t h e r w e h a v e e n or m o u s p o t e nt i a l Email: [email protected]; Call: +44 747 1086 241 think like a fi v e-s i g m a s t a n d a r d o f d i s c o v e r y i n our of the past decade, we are also finding to learn about the fundamental structure physicist field necessitates it. s u r pr i s i n g c o n n e c t i o n s b e t w e e n t h i n k i n g of the universe.

CERN COURIER SEPTEMBER/OCTOBER 2021 49

CCSepOct21_Viewpoint_v2.indd 49 03/08/2021 16:43 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM OPINION HIGH VOLTAGE POWER SUPPLIES INTERVIEW MAGNET POWER SUPPLIES HIGH CURRENT POWER SUPPLIES Stealing theorists’ lunch HIGH PRECISION Artificial-intelligence techniques have been used in experimental particle physics for 30 years, POWER SUPPLIES and are becoming increasingly widespread in theoretical physics. Anima Anandkumar and John Ellis explore the possibilities.

POWER SUPPLIES FOR BASIC RESEARCH How might artificial intelligence make J Guillaune/CERN-PHOTO-201410-199-5/A Anandkumar can come up with a combination of AT THE HIGHEST LEVEL an impact on theoretical physics? particles and interactions that could John Ellis (JE): To phrase it simply: solve our problems. I t hin k t hat in t his Heinzinger has played a role in research for a very long time where do we go next? We have the sort of problem space, the creativity thanks to its expertise and passion. In addition to the European Standard Model, which describes would come in the testing of the CERN facility in Geneva, companies, universities, colleges and all the visible matter in the universe t heor y. The AI might find a par t icle state institutions are investing in research. Whether it be testing successfully, but we know dark matter and a set of interactions that would and measuring equipment, accelerator technology or HV pow- must be out there. There are also deal w it h t he anomalies t hat I w as er supplies for particle detectors: High-specification technical puzzles, such as what is the origin of talking about, but how do we know the matter in the universe? During my what’s the right theory? We have to equipment from Heinzinger helps to produce reliable and resilient Rosenheim (Germany) 4 I 83026 GmbH I Anton-Jakob-Str. Heinzinger electronic I www.heinzinger.com 2458 0 I email: info @heinzinger.de +49 8031 P: results in laboratories and testing centers all over the world. lifetime we’ve been playing around propose some other experiments that with a bunch of ideas for tackling might test it – and that’s one place those problems, but haven’t come where the creativity of theoretical up with solutions. We have been able physicists will come into play. to solve some but not others. Could ar t ificial intelligence (AI) help us find AA: Absolutely. And many theories new paths towards attacking these John Ellis is Clerk Maxwell Professor of Theoretical Physics at are not directly testable. That’s where questions? This would be truly stealing King’s College London, and Anima Anandkumar is Bren professor the deeper knowledge and intuition theoretical physicists’ lunch. at Caltech and a director of machine-learning research at NVIDIA. that theoretical physicists have is so critical. Anima Anandkumar (AA): I think that, if you gave it a set of concepts the first steps are whether you can that we have, and the experimental Is human creativity driven by our understand more basic physics and be anomalies t hat we have, t hen t he AI consciousness, or can contemporary able to come up with predictions as could point the way? AI be creative? well. For example, could AI rediscover AA: Humans are creative in so the Standard Model? One day we can AA: The devil is in the details. How many ways. We can dream, we can hope to look at what the discrepancies do we give the right kind of data and hallucinate, we can create – so how are for the current model, and hopefully knowledge about physics? How do we do we build t hose capabilit ies into AI? come up with better suggestions. express those anomalies while at the Richard Feynman famously said “What same time making sure that we don’t I cannot create, I do not understand.” JE: A n interest ing e xercise might be bias the model? There are anomalies It appears that our creativity gives us to take some of the puzzles we have at suggesting that the current model the ability to understand the complex t he moment and somehow equip an AI is not complete – if you are giving inner workings of the universe. system with a theoretical framework that prior knowledge then you could With the current AI paradigm that we physicists are trying to work be biasing the models away from this is very difficult. Current AI is w it h, let t he AI loose and see whet her discovering new aspects. So, I geared towards scenarios where the it comes up with anything. Even think that delicate balance is the training and testing distributions over the last few weeks, a couple main challenge. are similar, however, creativity of experimental puzzles have been requires extrapolation – being able reinforced by new results on B-meson JE: I t hin k t hat t heoret ical physicists to imagine entirely new scenarios. So decays and the anomalous magnetic could propose a framework with extrapolation is an essential aspect. moment of the muon. There are many boundaries t hat AI could e x plore. We Can you go from what you have learned theoretical ideas for solving these could tell you what sort of particles and extrapolate new scenarios? For puzzles but none of them strike me are allowed, what sort of interactions that we need some form of invariance those could have and what would as being particularly satisfactory in Could AI or understanding of the underlying the sense of indicating a clear path still be a well-behaved theory from laws. That’s where physics is front and towards the next synthesis beyond the point of view of relativity and rediscover centre. Humans have intuitive notions t he Standard Model. Is it imag inable quantum mechanics. Then, let’s the Standard of physics from early childhood. We t hat one could dev ise an AI system just release the AI to see whether it Model? slowly pick them up from physical

CERN COURIER SEPTEMBER/OCTOBER 2021 51

CCSepOct21_Interview_v2.indd 51 03/08/2021 16:51 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM OPINION INTERVIEW OPINION interactions with the world. That Similarly to what happens when you put many now passed the Turing test when it understanding is at the heart of what we see of these fundamental constituents comes to generating human faces LETTERS getting AI to be creative. in the brain, together, and they behave in a way – you can no longer tell very easily that you could often not anticipate if whether it is generated by AI or if it is we recently JE: It is often said that a child learns you just looked at the fundamental a real person. So, I think those kinds more laws of physics than an adult ever built feedback laws of physics. One of the interesting of mechanisms that have competition Shanghai City Hall will! As a human being, I think that I mechanisms developments in physics over the built into the objective they optimise Hubble tension questioned are caused by a misunderstanding of think. I think that I understand. How into AI past generation is to recognise that are fundamental to creating more In your recent interview, Licia Verde how distance–ladder measurements can we introduce those things into AI? systems there are some universal patterns that robust and more intelligent systems. highlights the “Hubble tension” between actually work”. emerge. I’m thinking, for example, of the “local measurements of the current There are indeed serious problems with AA: We need to get AI to create images, phase transitions that look universal, JE: All this is very impressive – but expansion rate of the universe, for exam- the ΛCDM model – not least the complete and other kinds of data it experiences, even though the underlying systems there are still some elements that ple, based on supernovae as standard absence of any theoretical foundation for and then reason about the likelihood of are extremely different. So, I wonder, I am missing, which seem very candles, which... give values that cluster dark energy – but the “Hubble tension” the samples. Is this data point unlikely is there something similar in the important to theoretical physics. Take around 73 km s–1 Mpc–1 ” and the value of is not one of them! Indeed “the universe versus another one? Similarly to what field of intelligence? For example, the chess: a very big system but finite “around 67 km s–1 Mpc–1 ” from “measur- could be much, much more interesting we see in the brain, we recently built brain structure of the octopus is very nevertheless. In some sense, what I ing what we believe is the same quantity than we assumed,” as Verde says, and feedback mechanisms into AI systems. different from that of a human, so to try to do as a theoretical physicist has but only within a model, the lambda- there is emerging evidence that it is. For When you are watching me, it’s not what extent does the octopus think in no boundaries. In some sense, it is cold-dark-matter (ΛCDM) model, which example, the rest frames of the CMB and just a free-flowing system going from the same way that we do? infinite. So, is t here any hope t hat AI is looking at the baby universe via the cosmologically distant matter do not the retina into the brain; there’s also would eventually be able to deal with cosmic microwave background (CMB)” coincide, implying a gross violation of the a feedback system going from the AA: There’s a lot of interest now in problems that have no boundaries? (CERN Courier July/August 2021 p51). “The cosmological principle itself (N Secrest Not 1982 c r y s t a l s, fi r s t for L3, t h e n i n C r y s t a l C l e ar, inferior temporal cortex back into the st udy ing t he octopus. From what I Hubble tension is certainly one of the et al. 2021 ApJL 908 L51). Paul Lecoq and and finally for CMS. With Paul, and to a visual cortex. This kind of feedback is learned, its intelligence is spread out AA: That’s the difficulty. These are most intriguing problems in cosmology,” Sam Ting (third and large extent thanks to him, it was the fundamental to us being conscious. so that it’s not just in its brain but also infinite-dimensional spaces… so how says Verde, who asserts that “it is becom- Subir Sarkar University of Oxford. fifth from right) most enriching period of my career. Building these kinds of mechanisms in its tentacles. Consequently, you have do we decide how to move around ing increasingly unlikely that it is only stand either side of The inset picture is a good illustration into AI is the first step to creating this distributed notion of intelligence there? What distinguishes an expert due to dumb systematics”. • Author’s reply Jiang Zemin, who of the importance of links that are forged conscious AI. that still works very well. It can be like you from an average human is that In fact, systematic discrepancies In this Hubble constant (H0) matter, the went on to become between first-rank personalities in e x t remely camouflaged – imag ine you build your knowledge and develop that have been noted between super- fie l d m o v e s s o q u i c k l y t h a t pl o t s n e e dto the president of science and politics – links that were JE: A lot of the things that you just being in a wild ocean without a shell to intuition – you can quickly make nova datasets are sufficient to under- have a full date – not just the year, but China. Michel crucial to L3’s success. mentioned sound like they’re going protect yourself. That pressure created judgments and find which narrow mine the “Hubble tension”. Note that also month and day – and publications Lebeau is at the 1982 is cer ta i n ly t he first m i lestone i n to be incredibly useful going forward the need for intelligence such that it can part of the space you want to work the absolute peak magnitude of type-Ia in many cases become obsolete before right of the picture. L3 history, but it cannot be the date of this in our systems for analysing data. But be extremely aware of its surroundings on compared to all the possibilities. supernovae (assumed to be standard can- they are published. Preprints, seminars picture, which reminds me of exceptional how is AI going to devise an experiment and able to quic k ly camouflage itsel f or That’s the aspect that is so difficult dles) are obtained by calibrating their and talks are the way to stay up to date circumstances. The date is 7 February that we should do? Or how is AI going to manipulate different tools. for AI to fig ure out. The space is distances in the “cosmic distance ladder” with the latest developments. While this 1986, Chinese New Year, in Shanghai. At devise a theory that we should test? enormous. On the other hand, AI does to local distance anchors via the period- opens up a Pandora’s box of discussions that time S.C.C. Ting had already suc- JE: If intelligence is the way that a have a lot more memory, a lot more luminosity relation of Cepheid variable we could have about the (perhaps too) ceeded in involving the Shanghai Institute AA: Those are the challenging aspects living thing deals with threats and computational capacity. So can we stars. The magnitude is then trans- fast pace of science, let me stick to the of Ceramics (SIC) in the L3 experiment, for an AI. A data-driven method using feeds itself, should we apply the same create a hybrid system, with physicists lated into the Hubble parameter via the point here. As of late July, the consen- in the production of the BGO crystals for a standard neural network would evolutionary pressure to AI systems? and machine learning in tandem, magnitude–redshift relation of super- sus among the SH0ES and TRGB groups its electromagnetic calorimeter, which perform really poorly. It will only We threaten them and only the fittest to help us harness the capabilities novae in the redshift range 0.023–0.15. is that the data agree extremely well in started the previous year. However, the think of the data that it can see and not will survive. We tell them they have of both AI and humans together? However, the redshifts of 150 supernovae relative distances but not on absolute initial production rate of the crystals was about data that it hasn’t seen – what to go and find their own electricity We’re currently exploring theorem (all measured in the Sloan Digital Sky Sur- distances for objects that host Cepheid not sufficient to guarantee that L3’s cal- we call “zero-short generalisation”. or silicon or something like that – I provers: can we use the theorems v e y) a r e s i g n i fic a nt l y d i s c r e p a nt b e t w e e n variables and/or supernovae which are orimeter would be ready for the planned To me, the past decade’s impressive understand that there are some first that humans have proven, and then the JLA and Pantheon catalogues. Focus- far away enough to be directly useful for start of LEP, and a way to double the BGO

progress is due to a trinity of data, steps in this direction, computer add reinforcement learning on top to ing on these, the latter sample yields constraining H0. It is crucially important output had to be found. At this meeting neural networks and computing programs competing with each other create very fast theorem solvers? If we 72 km s–1 Mpc–1, while the former favours to remember that supernovae alone can- we discussed processing methods, tool- –1 –1 infrastructure, mainly powered by at chess, for example, or robots that can create such fast theorem provers 68 km s Mpc (M Rameez and S Sarkar not measure H0; they need an external ing and machinery, and the assembly GPUs [graphics processing units], have to find w all soc kets and plug in pure mathematics, I can see them 2021 Class. Quantum Grav. 38 154005). This calibration. It is this calibration that is of a workforce. I was present to address coming together: the next step for AI themselves in. Is this something being very useful for understanding is equivalent to the “systematic bias of uncertain. The moment that the absolute e n g i n e e r i n g m a t t e r s. My fi r s t v i s it t o SIC is a wider generalisation to the ability that one could generalise? And then the Standard Model and the gaps and 0.1–0.15 mag in the intercept of the Cep- distance of a few key objects that host had been a two-month stay the previous to extrapolate and predict hitherto intelligence could emerge in a way that discrepancies in it. It is much harder heid period-luminosity relations of supernovae/Cepheids is agreed upon at year, to initiate the crystal machining.

unseen scenarios. we hadn’t imagined? than chess, for example, but there are SH0ES galaxies” noted by Efstathiou (G the 1 to 2% level, the local H0 tension I would like to mention two dis- exciting programming frameworks Efstathiou 2020 arXiv:2007.10716). Thus will be resolved. Whether that will yield tinguished members of the Chinese –1 –1 –1 –1 Across the many tens of orders of AA: That’s an excellent point. and data sets available, with efforts there is no firm basis for the claimed H0 near 73 km s Mpc , 68 km s Mpc , Academy of Sciences (CAS) whose magnitude described by modern Because what you mentioned broadly to bring together different branches tension. In particular, an independent or elsewhere, remains to be seen. Data presence was essential to this meeting. physics, new laws and behaviours is competition – different kinds of of mathematics. But I don’t think calibration of supernovae using the “tip release 4 from the Gaia spacecraft will Next to and left of S.C.C. Ting is Profes- “emerge” non-trivially in complexity pressures that drive towards good, humans will be out of the loop, at least of the red giant branch” (TRGB) by Wendy provide the needed precision. sor Yan Dongsheng, CAS vice-president (p39). Could intelligence also robust objectives. An example is for now. Freedman et al. (W Freedman et al. 2019 and special advisor, and SIC honorary be an emergent phenomenon? generative adversarial models, which ApJ 882 34) indicates consistency with Licia Verde Universit y of Barcelona. director, who passed away in 2016; and JE: As a theoretical physicist, my main can generate very realistic looking This conversation is abridged from one of a the CMB value. Nevertheless, more than to the far left of the picture is Professor field of interest is the fundamental images. Here you have a discriminator series of podcasts recorded in anticipation a hundred theoretical solutions to the L3 diplomacy Yin Zhiwen, vice-director of SIC, who building blocks of matter, and the roles that challenges the generator to of the inaugural Sparks! Serendipity Hubble tension have been proposed. As It was a real pleasure to read your passed away in 2006. that they play very early in the history generate images that look real. These forum, which will take place at CERN from Efstathiou comments (Mon. Not. R. Astron. article about Paul Lecoq (CERN Courier of the universe. Emergence is the word kinds of competitions or games are 17 to 18 September(p36). Mark Rayner and Soc 2021 505 3866): “many of the claims May/June 2021 p59). I was Paul’s part- Michel Lebeau retired senior mechanical that we use when we try to capture getting a lot of traction and we have Abha Eli Phoboo were asking the questions. in the literature favouring such solutions ner in the development of scintillating engineer, IN2P3 and CERN.

52 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 53

CCSepOct21_Interview_v2.indd 52 03/08/2021 16:48 CCSepOct21_Letters_v2.indd 53 04/08/2021 10:24 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

Scintinel™ Stilbene Scintillation Crystals are OPINION the gold standard in neutron-gamma Pulse Shape Discrimination. REVIEWS Unlike most neutron detectors, stilbene is sensitive only to high-energy fast neutrons. Directly detecting fast neutrons preserves timing, neutron energy distribution, and Enabling Fast Neutron Detection directional information. Scintinel stilbene is superior to liquid and plastic scintillators for An enduring and adaptable design

neutron/gamma discrimination. M Brice/CERN-PHOTO-201903-071-3 Resistive Gaseous Detectors: Designs, • Superior Pulse Shape Discrimination Performance, and Perspectives • Lower Gamma false rate By Marcello Abbrescia, Vladimir Peskov and Paolo Fonte • Lower Threshold for gamma rejection Wiley • Excellent Neutron Detection Efficiency T h e fi r s t t r u l y r e s i s t i v e g a s e o u s d e t e c- Crystals are available in sizes up to 5 inches tor was invented by Rinaldo Santon- and in various format shapes. Stilbene ico and Roberto Cardarelli in 1981. A kind of parallel-plate detector with crystals are easily integrated with PMT and electrodes made of resistive materials SiPM devices. such as Bakelite and thin-float glass, t h e d e s i g n i s s o m e t i m e s a l s o k n o w n a s a resistive-plate chamber (RPC). Resistive gaseous detectors use electronegative For more information go to gases and electric fields that typically www.inradoptics.com/scintinel exceed 10 k V/c m. W h e n a c h a r g e d p a r t i- c l e i s i n c i d e nt i n t h e g a s g a p, t h e w or k i n g operational gas is ionised, and primary e l e c t r o n s c a u s e a n a v a l a n c h e a s a r e s u l t of the high electric field. The induced charge is then obtained on the read- out pad as a signal. RPCs have several SUPERCON, Inc. unique and important practical fea- t u r e s, c o m bi n i n g g o o d s p a t i a l r e s olut i o n Superconducting Wire Products with a time resolution comparable to t h a t o f s c i nt i l l a t or s. T h e y a r e t h e r e for e Standard and Speciality designs are available to meet your most demanding well suited for fast spacetime particle Muon detector Resistive-plate chambers in the CMS detector (grey disk), pictured during the ongoing tracking, as a cost-effective way to second long shutdown of the LHC. superconductor requirements. instrument large volumes of a detec- tor, for example in muon systems at RPCs. A modern design of double-gap tomography. For homeland security, SUPERCON, Inc. has been producing niobium-based superconducting wires and cables for 58 years. collider experiments. RPCs and examples for the muon sys- RPCs can be used in muon-scattering the original SUPERCON Resistive Gaseous Detectors: Designs, t e m s l i k e t h o s e a t AT L A S a n d C M S a t t h e tomography with cosmic-ray muons We are – the world’s first commercial producer of niobium-alloy based wire and Performance, and Perspectives, a n e w b o o k LHC, the STAR detector at the Relativistic to scan spent nuclear fuel containers cable for superconducting applications. b y M a r c e l l o A b br e s c i a, V l a d i m i r P e s k o v Heav y-Ion Collider at Brookhaven and without opening them, or to quickly a n d P a u l o Fo nt e, c o v e r s t h e b a s i c pr i n c i- the multi-gap timing RPC for the scan incoming cargo trucks without Standard SC Wire Types Product Applications ples of their operation, historical devel- time-of-flight system of the HADES disrupting the traffic of logistics. A key NbTi Wires Magnetic Resonance Imaging opment, the latest achievements, and e x p e r i m e nt a t G SI a r e d e t a i l e d. A d v a n c e d subject not covered in detail, however, their growing applications in various designs with new materials for elec- is the need to search for environmen- Nb3Sn —Bronze Nuclear Magnetic Resonance fields from hadron colliders to astro- t rodes for h ig h-r ate detec tors a re t hen tally friendly alternatives to gases with Nb3Sn —Internal Tin High Energy Physics p h y s ic s. T h i s b o o k i s n o t o n l y a s u m m a r y introduced, and ageing and longevity h i g h g l o b a l-w a r m i n g p o t e nt i a l, w h ic h CuNi resistive matrix wires SC Magnetic Energy Storage of numerous scientific publications are elaborated upon. A new generation are often needed in resistive gaseous on many different examples of RPCs, of gaseous detectors with resistive detectors at present to achieve stable

Fine diameter SC Wires Medical Therapeutic Devices Wiley but also a detailed description of their e l e c t r o d e s t h a t c a n b e m a d e w it h m i c r o - and sustained operation (CERN Courier Aluminum clad wire Superconducting Magnets and Coils desig n, operat ion and per for mance. electronic technology is then intro- July/Aug ust 2021 p20). Wire-in-Channel Crystal Growth Magnets The book has nine chapters. The duced: these large-area electrodes Abbrescia, Peskov and Fonte’s book Innovative composite wires Scientific Projects o p e r a t i o n a l pr i n c i pl e o f g a s e o u s d e t e c- can easily be manufactured while still will be useful to researchers special- tors and some of their limitations, achieving high spatial resolutions up ising in high-energy physics, astron- most notably the efficiency drop in a to 12 microns. omy, astrophysics, medical physics and high-particle-rate environment, are radiation measurements, and at all “We deliver when you need us!” d e s c r i b e d. T h i s i s fol l o w e d b y a h i s t or y Homeland security academic levels, from students to of parallel-plate detectors, the first The final chapter covers applications seasoned professionals. c l a s s i c a l B a k e l it e R P C, d o u bl e-g a p R P C s outside particle physics such as those in www.SUPERCON-WIRE.com and glass-electrode multi-gap timing medicine exploiting positron-emission Yi Wang Tsinghua University.

CERN COURIER SEPTEMBER/OCTOBER 2021 55

CCSepOct21_Reviews_v2.indd 55 04/08/2021 10:30 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

OPINION REVIEWS World Scientific Climate Change and Energy Options ent the basic principles of alternative for a Sustainable Future Human wellbeing renewable sources, and offer many examples, including agrivoltaics in By Dinesh Kumar Srivastava and is decided by the Africa, a floating solar-panel station V S Ramamurthy rate at which power in California and wind-turbines in World Scientific the Netherlands and India. Drawing is consumed on their own expertise, they discuss In Climate Change and Energy Options nuclear energy and waste-management, for a Sustainable Future, nuclear phys- description of how each green-energy accelerator-driven subcritical systems, icists Dinesh Kumar Srivastava and option could be integrated into a and the use of high-current electron V S Ramamurthy explore global policies global-energy strategy. accelerators for water purification. The for an eco-friendly future. Facing the In the first part of the book, the book t hen fina l ly t ur ns to sust a i nabi l- world’s increasing demand for energy, authors provide a wealth of evidence ity, showing by means of a wealth of the authors argue for the replacement demonstrating the pressing reality of s c i e nt i fic d a t a t h a t i n c r e a s i n g t hesup - of fossil fuels with a new mixture of climate change and the fragility of the ply of renewable energy, and reducing green energy sources including wind environment. Srivastava and Rama- carbon-intensive energy sources, can energy, solar photovoltaics, geothermal murthy then examine unequal access lead to sustainable power across the energy and nuclear energy. Srivastava to energy across the globe. There should globe, both reducing global-warming is a theoretical physicist and Rama- be no doubt that human wellbeing is emissions and stabilising energy prices murthy is an experimental physicist decided by the rate at which power is for a fairer economy. The authors stress with research interests in heavy-ion consumed, they write, and providing that any solution should not compromise physics and the quark–gluon plasma. enough energy to everyone on the planet quality of life or development opportu- To g e t h e r, t h e y a n a l y s e s olut i o n s o ffe r e d to reach a human-development index nities in developing countries. by science and technology with a clar- of 0.8, which is defined by the UN as This book could not be more timely. ity that will likely surpass the expecta- high human development, calls for about It is an invaluable resource for scientists, tions of non-expert readers. Following 30 trillion kWh per year – roughly double policymakers and educators. a pedagogical approach with vivid the present global capacity. illustra tions, the book offers an in-depth Srivastava and Ramamurthy pres- Panos Charitos CERN.

Compact Photomultipliers for scintillation applications

Our compact range of end window photomultipliers with blue-green sensitive bialkali photocathode and 10 high gain, high stability, SbCs dynodes of linear focused design for good linearity and timing

The compact nature of these photomultiplierNew Compact tubes make 25mm these Diameter Photomultiplier designs ideal for use in scintillation applications.

These are available in 38mm, 9181B 52mm and 78mm diameter. – ,, "',.... Our wide range of products cover photomultipliers, photodetector modules, light-tight housings, HV supplies and associated custom built electronics.

-2 -1 0 1 2 3 Input Power {dBm)

www.et-enterprises.com

56 CERN COURIER SEPTEMBER/OCTOBER 2021

CCSepOct21_Reviews_v2.indd 56 04/08/2021 10:31 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM PEOPLE CAREERS

THE LATEST JOBS IN HIGH-ENERGY PHYSICS Quantum gravity in the Vatican

Residents of the Vatican Vatican Observatory Obser v ator y desc r ib e l i fe a s a full-time physicist in the church.

“O u r j o b i s t o b e p a r t o f t h e s c i e nt i fic c o m mu- nity and show that there can be religious people and priests who are scientists,” says Gabriele G i o nt i, a R o m a n C a t h ol i c pr i e s t a n d t h e or e t i c a l physicist specialising in quantum gravity who is resident at t he Vat ican Obser v ator y. “Our mission is to do good science,” agrees G u y C o n s ol m a g n o, a n o t e d pl a n e t a r y s c i e nt i s t, Jesuit brother and the observatory’s director. “I like to say we are missionaries of science to the believers.” No t o n l y m i s s io n a r ie s o f f a it h, t h e n, b ut a l s o of science. A nd t here are adv antages. “At the Vatican Observatory, we don’t have to write proposals, we don’t have to worry about t e nu r e a n d w e d o n’t h a v e t o h a v e r e s u lt s i n t h r e e years to get our money renewed,” says Consol- magno, who is directly appointed by the Pope. Missionary of science Gabriele Gionti presents Pope Francis with a journal containing a recent “It changes the nature of the research that is publication on black holes, gravitational waves and spacetime singularities. av ailable to us.” “Here I have had time to just study,” says faith and science were fully compatible. “The A n d b e s i d e s, a nt a g o n i s m b e t w e e n s c i e n c e a n d Gionti, who explains that he was able to extend first thing that the Vatican Observatory did was r e l i g i o n i s l a r g e l y b a s e d o n a f a l s e d i c h o t o m y, s a y s his research to string theory as a result of this to take part in the Carte du Ciel pr o g r a m m e,” h e C o n s ol m a g n o. “T h e G o d t h a t m a n y a t h e i s t s d o n’t extra freedom. “If you are a postdoc or under s a y s, h i nt i n g a t a s e c o n d a r y m o t i v a t i o n. “Ev e r y b e l i e v e i n i s a G o d t h a t w e a l s o d o n’t b e l i e v e i n.” tenure, you don’t have t his oppor t unit y.” national observatory was given a region of the T h e o b s e r v a t or y’s d i r e c t or p u s h e s b a c k h a rd “I remember telling a friend of mine that I sky. Italy was given one region and the Vati- o n t h e i d e a t h a t f a it h i s i n c o m p a t i bl e w it h p h y s- d o n’t h a v e t o w r it e g r a nt pr o p o s a l s, a n d h e s a id, can was given another. So, de facto, the Vatican ics. “It doesn’t tell me what science to do. It ‘h o w d o I g e t i n o n t h i s?’” j o k e s C o n s ol m a g n o, a became seen as an independent nat ion state.” doesn’t tell me what the questions and answers native of Detroit. “I said that he needed to take The observatory quickly established itself are going to be. It gives me faith that I can under- a vow of celibac y. He replied, ‘it’s wor t h it!’.” as a respected scientific organisation. Though stand t he universe using reason and log ic.” it is staffed by priests and brothers, there is Cannonball moment an absolute rule that science comes first, says Surprised by CERN C l a d i n T-s h i r t s, G i o nt i a n d C o n s ol m a g n o d o n’t Consolmagno, and the stereotypical work of a Due to light pollution in Castel Gandolfo, a new Search our comprehensive job listings and recruiter resemble the priests and monks seen in mov- priest or monk is actually a temptation to be out post of t he Vat ic a n Obser v ator y w as est ab- ies. They are connected to monastic tradition, resisted. “Day-to-day li fe as a scient ist can be lished in Tucson, Arizona, in 1980. A little later information to find the perfect high-energy physics job but do not withdraw from the world. As well as tedious, and it can be a long time until you see in the day, when the Sun was rising there, I spoke being full-time physicists, both are members a reward, but pastoral life can be rewarding to Paul Gabor – an astrophysicist, Jesuit priest for you. Whether it’s working in academia, industry or of the Society of Jesus – a religious order that immediately,” he e x plains. and deputy director for the Tucson observatory. at a research facility, take the next step in your career traces its origin to 1521, when Saint Ignatius of Consolmagno was a planetary scientist for Born in Košice, Slovakia, Gabor was a summer Loyola was struck in the leg by a cannonball at 20 years before becoming a Jesuit. By contrast, student at CERN in 1992, working on the devel- with CERN Courier Jobs t h e B a t t l e o f P a m pl o n a. To d a y t h e y h e l p s t a ff a t Gionti, who hails from Capua in Italy, joined o p m e nt o f t h e e l e c t r o m a g n e t ic c a l or i m e t e r o f t h e an institution that was founded in 1891, though after his first postdoc at UC Irvine in Califor- ATLAS experiment, a project he later continued it s or i g i n s a r g u a bl y d a t e b a c k t o a t t e m p t s t o fi x n i a. Ne it h e r r e p or t s e n c o u nt e r i n g pr o fe s s i o n a l in Grenoble, thanks to winning a scholarship t he date for Easter in 1582. prejudice as a result of their vocation. “I think at the university. “We were making prototypes cerncourierjobs.com “It was at the end of the 19th century that the that’s a generational thing,” says Consolmagno. and models and software. We tested the actual myth began that the church was anti-science, “Scientists working in the 1970s and 1980s were physical models in a couple of test-beam runs and they would use Galileo as the excuse,” says more likely to be anti-religious, but nowadays – t hat w as f un,” he recalls. Consolmagno, explaining that the Pope at the it’s n o t t h e c a s e. Yo u a r e l o o k e d o n a s p a r t o f t h e Gabor was surprised at how he found the time, Pope Leo XIII, wanted to demonstrate that multicultural nature of the field.” laboratory. “It was an important part of my

CERN COURIER SEPTEMBER/OCTOBER 2021 59

CCSepOct21_Careers_v2.indd 59 04/08/2021 10:41 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM PEOPLE CAREERS RECRUITMENT

journey, because I was quite surprised that I Vatican Observatory do, so I urge a little bit of restraint. Religious found CERN to be full of extremely nice people. I zeal is a great thing, but if you are in the third was expecting everyone to be driven, ambitious, year of a doctorate, don’t just pack up your bags competitive and not necessarily collaborative, and join a seminary. That is not a very prudent For advertising enquiries, contact CERN Courier recruitment/classified, IOP Publishing, Temple Circus, Temple Way, Bristol BS1 6HG, UK. but people were very open,” he says. “It was a t h i n g t o d o. T h a t i s t o n o b o d y’s b e n e fit. T h i s i s a Te l +4 4 (0)1 17 930 1 264 E-m a i l s a le s@c e r nc ou r ie r.c om. really good human experience for me.” scenario that is all too common unfortunately.” Please contact us for information about rates, colour options, publication dates and deadlines. “When I finally caved in and joined the Consolmagno also offers words of caution. Jesuit order in 1995, I always thought, well, these “50% of Jesuits leave the order,” he notes. “But scientists definitely are a group that I got to this is a sign of success. You need to be where know and love, and I would like to, in one way or you belong.” another, be a minister to them and be involved But Gionti, Consolmagno and Gabor all agree with them in some way.” that, if properly discerned, the life of a scientific “Something that I came to realise, in a begin- The Pope’s astronomer Guy Consolmagno, religious is a rewarding one in a community ning, burgeoning kind of way at CERN, is the the director of the Vatican Observatory, poses like the Vatican Observatory. They describe a idea of science being a spiritual journey. It forms with a summer student. close-knit group with a common purpose and your personality and your soul in a way that any little superficiality. sustained effor t does.” grow as human beings. I think this point is quite “Faith gives us the belief that the universe is important. In a way it explains my experience at good and worth studying,” says Consolmagno. Scientific athletes CERN as a place full of nice, generous people.” “If you believe that the universe is good, then “Experimental science can be a sapiential journey Surprisingly, however, despite being happy you are justified in spending your life study- to wisdom,” says Gabor. “We are subject to con- with life as a scientific religious and religious ing things like quarks, even if it is not useful. stant frustration, failure and errors. We are con- scientist, Gabor is not recruiting. Believing in God gives you a reason to study fronted with our limitations. This is something “There is a certain tendency to abandon sci- science for the sake of science.” that scientists have in common with athletes, ence to join the priesthood or religious life,” he for example. These long labours tend to make us says. “This is not necessarily the best thing to Mark Rayner deputy editor. Appointments and awards

D Biedenbender; M Hirsch D Biedenbender; will see Hirsch use computer his contributions to the novel of the thesis manuscript and the models and coding for his art, GPU-based first-level trigger importance of the contribution to and he aims to draw on his data of LHCb Upgrade I. Pereima the collaboration. The award was visualisation background to make was recognised for his thesis presented online in June. Fermilab science more accessible on the search for new decays and intriguing to the public. of beauty particles at the LHCb Beamline for Schools experiment, and he made Team “EXTRA’’ from Liceo LHCb recognises theses significant contributions to the Scientifico Statale “A Scacchi” The 2021 winners of the understanding of the X(3872) (Italy) and team “Teomiztli’’ from LHCb thesis prize have been particle and the calibration of the the Escuela Nacional Preparatoria New artists at Fermilab announced as Tom Boettcher hadronic calorimeter. “Plantel 2” (Mexico) have won the Fermilab have announced their (left) (Massachusetts Institute 2021 edition of CERN’s Beamline new 2021–2022 guest composer of Technology) and Dmitrii ALICE PhD winner for Schools competition. Every as David Biedenbender (left) Pereima (right) (Kurchatov Jonatan Adolfsson (Lund year, the Beamline for Schools and 2021–2022 artist-in- Institute, Moscow). The prize is University) has been awarded competition challenges teams residence as Mark Hirsch (right). awarded annually to early-career the 2021 ALICE PhD thesis of high-school students across Biedenbender becomes the second scientists for the best PhD thesis prize for his thesis “Study of the world to submit proposals

guest composer at the institute, and outstanding contributions P Christiansen for an experiment that utilises following on from David Ibbett in to the LHCb collaboration. a beamline. Team EXTRA 2020, who is currently working Boettcher’s thesis is on the proposed to investigate the on neutrino-inspired music. The LHCb GPU high-level trigger transition-radiation effect by role gives a guest composer an and measurements of neutral discriminating signals produced opportunity to interpret Fermilab pion and photon production by the particles in a beam from through music and celebrate the with the LHCb detector, and he those produced by X-ray photons, relationship between art and was particularly commended for while the goal of team Teomiztli

science. Biedenbender has written CERN was to compare the production of pieces inspired by accelerators Cherenkov radiation in different before, having in 2017 composed materials. The two winning “Cyclotron” – a 10-minute Ξ-hadron correlations in pp teams, which were chosen from wind ensemble piece inspired collisions at √s = 13 TeV using a pool of 289 teams representing by the particle accelerator at the ALICE detector”. The annual 57 countries, will travel to DESY the National Superconducting prize is awarded by the ALICE later this year to carry out their Cyclotron Laboratory at MSU. collaboration to the best PhD proposed experiments with the Now in its seventh year, the thesis based on the excellence of support of scientists from CERN artist-in-residence programme the results obtained, the quality and DESY.

60 CERN COURIER SEPTEMBER/OCTOBER 2021

CCSepOct21_Careers_v2.indd 60 04/08/2021 10:42 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

We provide international scientific teams with access to unique technologies ELI Beamlines is a part of the ELI (Extreme Light Infrastructure) pan-European laser centre representing a unique tool of support of scientific excellence in Europe. ELI Beamlines operates the world’s most intense laser system. For our location in Hamburg we are seeking: With ultra-high power 10 PW and concentrated intensities DESY-Fellowships – experimental of up to 1024 W/square cm, we offer our users a unique source of radiation and rays of accelerated particles. These The Research Division of GSI Helmholtzzentrum für particle physics Schwerionenforschung in Darmstadt, Germany, searches Further Informations: www.desy.de/career/job_offers beamlines are enable pioneering research not only in at the earliest possible date for a https://v22.desy.de/e67416/records106397/index_eng.html physics and material science, but also in biomedicine and ID: FHFE003/2021 | Postdoc | EG 13 | 2 years | fulltime/part-time | laboratory astrophysics and many other fields. Deadline: 30.09.2021 Head of the Experiment In our team we have the following positions available: Electronics Department (EEL) • Postdoc Fellow (all genders) • Scientist Ref. No.: 21.77-6220 • Designer of control systems • Web developer Your Tasks: • Software engineer The successful candidate will lead the EEL department of • Phd GSI with in total 30 scientists, engineers and technicians JOIN US! developing and supporting instrumentation for a wide The future is in laser technologies variety of physics experiments in the fields of nuclear For more information see our website www.eli-beams.eu and and particle physics, atomic physics, material sciences, send your application, please. plasma physics and biophysics. The department acts as a competence center for the international collaborati- ons active at GSI/FAIR as well as for local units like the accelerator department and the detector laboratory. The department conducts a broad range of activities within a portfolio comprising analog and digital electronics, sys- tem design, ASIC development, embedded software (e.g. FPGA), data acquisition (DAQ) and analysis, control sys- tems as well as PCB layout, assembly and rapid prototy- THE LATEST JOBS IN ping. Your vision and expertise will guide the department HIGH-ENERGY PHYSICS to strengthen these roles by using forefront technologies.

Your Profile: We are looking for a person with the following qualification: ▪ university degree in Electronic Engineering or Physics (M.Sc.) ▪ outstanding expertise in modern front-end electronics for nuclear and particle physics experiments ▪ several years of work experience in a supervising position in electronics development ▪ professional experience in one of these areas: stringent real-time applications / low-latency systems / low-noise systems ▪ excellent written and oral English language skills; good German language skills will be essential ▪ goal-oriented and analytical mind-set, strong skills in people management and resource handling

Please provide us with a short summary of your ideas on the future of electronics for experiments in the framework of GSI and FAIR.

Further information on this position can be found at https://www.gsi.de/en/jobscareer/job_offers Search our comprehensive job Please also feel free to contact Dr. Yvonne Leifels listings and recruiter information (y.leifels(at)gsi.de) in case of questions. to find the perfect high-energy If you find this position interesting, please send your physics job for you. complete application incl. the usual documents, your salary expectation and your earliest start date, stating the Ref. No., by 12. September 2021 at the latest to: cerncourierjobs.com bewerbung(at)gsi.de

CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM PEOPLE Open access for CERN researchers OBITUARIES

Thanks to a transformative agreement between CERN and 50+ journals including: IOP Publishing, you can: Steven Weinberg 1933–2021 • Publish open access in more than 50 IOP Publishing journals at no cost to you A mind to rank with the greatest • Increase your paper’s visibility and impact* M Valentine M Comply with any open access mandate and policy, even when Steven Weinberg, one of the greatest theo- • retical physicists of all time, passed away on publishing in non-SCOAP3 journals 23 July, aged 88. He revolutionised particle physics, quantum field theory and cosmol- • Be covered whether your affiliation is primary or secondary ogy with conceptual breakthroughs that still Include articles from CERN experimental collaborations form the foundation of our understanding of • physical reality. such as ATLAS, CMS, LHCb and ALICE Weinberg is well known for the unified theory of weak and electromagnetic forces, which earned him the Nobel Prize in Physics in 1979, jointly awarded with Sheldon Glashow Find out more at http://bit.ly/CERNIOP and Abdus Salam, and led to the prediction of the Z and W vector bosons, later discovered at CERN in 1983. His breakthrough was the real- isation that some new theoretical ideas, ini- tially believed to play a role in the description of nuclear strong interactions, could instead explain the nature of the weak force. “Then it suddenly occurred to me that this was a per- fectly good sort of theory, but I was applying it to the wrong kind of interaction. The right place to apply these ideas was not to the strong interactions, but to the weak and electromag- netic interactions,” as he later recalled. With his work, Weinberg had made the next step in t he u n i fic at ion o f p hy s ic a l l a w s, a f t e r Ne w t on Steven Weinberg radically changed the way we look at the universe. understood that the motion of apples on Earth and planets in the sky are governed by the same In my life, I have built physicists, and will certainly continue to serve gravitational force, and Maxwell understood future generations. that electric and magnetic phenomena are the only one model Steven Weinberg is among the very few indi- expression of a single force. viduals who, during the course of the history In his research, Weinberg always focused of civilisation, have radically changed the way on an overarching vision of physics and not we look at the universe. on a model description of any single phe- theoretical concepts, reinterprets them in orig- nomenon. At a lunch among theorists, when inal ways, and applies them to the description Gian Giudice CERN. a colleague referred to him as a model of the physical world. A good example is his builder, he jokingly retorted: “I am not a construction of effective field theories, which • Steven Weinberg passed away shortly model builder. In my life, I have built only are still today the basic tool to understand before the Courier went to press. A featured one model.” Indeed, Weinberg’s greatest the Standard Model of particle interactions. article exploring his lifetime of extraordinary legacy is his visionary approach to vast areas His inimitable way of thinking has been the contributions to fundamental physics will of physics, in which he starts from complex inspiration and guidance for generations of appear in the next issue.

Felix H boeHm 1924–2021 Born in Basel, Switzerland, in 1924, Felix had developed the bent-crystal spectrometer, SCAN ME TO FIND OUT MORE Always ahead studied physics at ETH Zürich, earning a an instrument with unrivalled energy resolu- diploma in 1948 and a PhD in 1951 for a measure- tion in gamma-ray spectrometry. He used it of the game ment of the (p,n) reaction at the ETH cyclotron. to determine nuclear radii by measuring X-ray In 1951 he moved to the US and joined the group isotope shifts in atoms. Later, he installed such * H Piwowar, J Priem, L V arivière, J P Alperin, L Matthias, B Norlander, A Farley, J West, S Haustein (2018) Felix H Boehm, who was William L Valentine of Chien-Shiung Wu at Columbia University, devices at LAMPF, SREL and PSI to investigate The state of OA: a large-scale analysis of the prevalence and impact of Open Access articles. Professor of Physics at Caltech, passed away on which was investigating beta decay. He joined pionic atoms, which led to a precise determi- PeerJ 6:e4375 https://doi.org/10.7717/peerj.4375 25 May in his Altadena home. He was a pioneer Caltech in 1953 and spent the rest of his aca- nation of the strong-interaction shift in pionic in the study of fundamental laws in nuclear- demic career there. hydrogen. At Caltech, he also became inter- s physics experiments. Felix worked first with Jesse DuMond, who ested in parity violation and time-reversal

CERN COURIER SEPTEMBER/OCTOBER 2021 65

CCSepOct21_Obits_v4.indd 65 04/08/2021 10:51 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM CERNCOURIER.COM

PEOPLE OBITUARIES PEOPLE OBITUARIES Caltech Archives invariance. In 1957, in an experiment performed disappearance mode at a still-longer baseline. structure function, which entails the presence Throughout his life, Anatoly continued to A n a t ol y Va s i l i e v i c h E f r e m o v w a s t h e u n d i s­ with Aaldert Wapstra, he demonstrated that In parallel, Felix decided to probe neutrino in the nucleus of a hard collective quark sea. develop concrete manifestations of his ideas puted scientific leader, who initiated studies of electrons in beta decay have a predominantly masses by searching for neutrinoless double- This naturally explains both the EMC nuclear based on fundamental theory, becoming a quantum chromodynamics and spin physics in negative helicity. beta decay. He led a small collaboration that effect and cumulative particle production, t e a c h e r a n d a d v i s or o f m a n y p h y s i c i s t s a t J IN R, Dubna, one of the key BLTP JINR staff, and at the In the mid 1970s, discussions with Harald installed a germanium detector in the Got- and unambiguously indicates the existence of in Russia and abroad. In 1991 he initiated and same time a modest and very friendly person, Fritzsch and Peter Minkowski convinced Felix thard underground laboratory in Switzerland multi- quark density fluctuations (fluctons) – became the permanent chair of the organising enjoying the highest authority and respect of that the study of neutrino masses and mix- to probe 76Ge, and then searched for the pro- a prediction that was later confirmed and led committee of the Dubna International Work­ colleagues. It is this combination of scien­ ings might provide answers to fundamental cess using a time-projection chamber (TPC) to the so­called nuclear super­scaling phe­ shops on Spin Physics at High Energies. He tific and human qualities that made Anatoly questions. From then on, long before it was filled with xenon enriched with136 Xe. The TPC, n o m e n o n. To d a y, s i m i l a r e ffe c t s o f flu c t o n sor was a long­term and authoritative member Efremov’s personality unique, and this is how fashionable, it became his main field of activ- a novel device at the time, improved the event short­range correlations are investigated in a of the International Spin Physics Committee we will remember him. ity. He first looked at neutrino oscillations and signature and thus reduced the background, fixed-target experiment at NICA and in several coordinating work in this area, and a regular initiated an electron–neutrino disappearance allowing stringent constraints to be placed on experiments at Jlab in the US. v i s it or t o t h e C E R N t h e or y u n it s i n c e t h e 1970 s. His colleagues and friends. experiment with Rudolf Mössbauer. t h e e ffe c t i v e n e ut r i n o m a s s. T h e o n g oi n g E XO Theirs was the first dedicated search for experiment can be seen as a continuation of a lexei onuchin 1934–2021 neutrino oscillations, beginning with a this programme, vastly improving the sensi- short-baseline phase at the ILL reactor in t i v it y i n it s fi r s t p h a s e (E XO -200 a t W I PP, Ne w Grenoble. The concept of the experiment was Mexico) and expected to do even better in the An outstanding experimentalist presented at the Neutrino ′79 conference in second phase, nEXO. Family photograph Family Bergen, at which the Gargamelle collabora- Felix Boehm had a talent to identify impor- Alexei Onuchin, one of the pioneers of exper­ c a r r i e d o ut a t B u d k e r IN P a n d o t h e r l a b or a t or i e s.

tion also reported limits on νμ ↔ ν e oscilla- tant issues on the theoretical side, and to iments at colliding beams, passed away on These include the development and produc­ tions. Both talks were relegated to a session select the appropriate technical methods on 9 Januar y in Novosibirsk, Russia. tion of multiwire proportional chambers for on exotic phenomena. The ILL experiment was the experimental side. He was always ready to Onuchin was born in 1934 in a small village the MD­1, various counters based on Cheren­ continued at the Gösgen reactor in Switzer- innovate. In particular, he realised very early in the Gork y (Nizhny Novgorod) region. A fter kov radiation and the creation of a calorimeter land with a longer baseline. No evidence of on the importance of selecting radio-pure graduating from high school with honours, he based on liquid krypton, among many others. oscillations was found and stringent limits in materials in low-count-rate, low-background decided to try his hand at science and in 1953 he Cherenkov counters held a special place a given parameter space were derived, con- Felix Boehm focused on neutrino masses and experiments. All those who worked with him entered the physics department of Moscow State in Alexei’s heart from the very beginning of tradicting positive claims made by others. A mixings long before it was fashionable. appreciated his open mind, his determination, University. In 1959 he graduated with honours his career as a student in the laboratory of larger detector was later built at the Palo Verde his great culture and his kindness. and was invited by Gersh Budker to work at the No b e l-l a u r e a t e P a v e l C h e r e n k o v. St a r t i n g f r o m nuclear power station in Arizona, where again w a s t h e K a m L A N D e x p e r i m e nt i n Ja p a n, w h ic h newly organised Institute of Nuclear Physics pioneering water­threshold counters in the no evidence for oscillations was found. A logi- was exposed to several reactors and eventually, Jean-Luc Vuilleumier University of Bern; at Novosibirsk (INP). e x per iment at V EPP–2, he later developed t he cal continuation of the effort initiated by Felix i n 2002, obser ve d neut r i no osc i l l at ion s i n t he Felix’s friends and colleagues. At INP, Onuchin enjoyed many important M D –1 C h e r e n k o v c o u nt e r s fi l l e d w it h e t h y l e ne roles. He took part in experiments at the world’s pressurised to 25 bar, and finally suggested A nAtoly VAsilieVich efremoV 1933–2021 fi r s t e l e c t r o n– e l e c t r o n c ol l i d e r ( V E P-1), a c t i v ely the aerogel counters with wavelength shifters worked on the preparation of a detector for the (ASHIPH) now operating in the KEDR detec­ electron–positron collider VEPP-2, supervised tor. For this work, in 2008 Alexei Onuchin was An expert in QCD and spin physics the construction of the MD­1 detector for the awarded the Cherenkov Prize of the Russian VEPP-4 collider, was one of the leaders of the Academy of Sciences. O n 1 Ja nu a r y, a f t e r a l o n g s t r u g g l e w it h a s e r i- T A Potapova are his 1979 articles (written together with KEDR detector experiment at the VEPP-4M Alexei was a great teacher. Among his o u s i l l n e s s, A n a t ol y Va s i l i e v i c h E f r e m o v o f t h e his PhD student A V Radyushkin) about the collider and was a great enthusiast of the detec­ former students are professors, group lead­ Bogoliubov Laboratory of Theoretical Physics asymptotic behaviour of the pion form factor tor project for the proposed Super Charm-Tau ers and members of the Russian Academy (BLTP) at JINR, Dubna, Russia, passed away. in QCD, and the evolution equation for hard Factory. He was also an organiser and for many of Sciences. He was also a caring father and He was an outstanding physicist, and a world exclusive processes, which became known as years the leader of the Budker INP group work­ loving husband, who raised a large family with e x p e r t i n q u a nt u m fie l d t h e or y a n d e l e m e nt ary the ERBL (Efremov–Radyushkin–Brodsky– ing at the BaBar experiment at SLAC. four children, five grandchildren and three particle physics. Lepage) equation. Proving the factorisation During his career, Alexei made a great great­grandchildren. He will always be remem­ A n a t ol y E f r e m o v w a s b or n i n K e r c h, C r i m e a, o f h a rd pr o c e s s e s e n a ble d m a n y s u b t le e ffe c t s contribution to the development of experimen­ bered by his family, friends and colleagues. to the family of a naval officer. Since child- in QCD to be described, in particular parton tal techniques in particle physics. It was this hood, he retained his love for the sea and was correlations, which became known as the ETQS that determined the high level of experiments Alexei Onuchin was an expert in Cherenkov counters. His colleagues and friends. an excellent yachtsman. After graduating (Efremov–Teryaev–Qiu–Sterman) mechanism. from Moscow Engineering Physics Institute During the past three decades, Efremov, Tord riem ann 1951–2021 in 1958, where among his teachers were Isaak together with his students and colleagues, Pomeranchuk and his master’s thesis advi- devoted his attention to several problems: sor Yakov Smorodinsky, he started working at the proton spin; the role of the axial anomaly Setting the highest standards BLT P J IN R . A t t h e t i m e, D m it r ij Bl o k h i nt s e v w a s and spin of gluons in the spin structure of a JINR direc tor. A natoly a lw ays considered h im nucleon; correlations of the spin of partons; and Theorist Tord Riemann, who made key contrib­ Tord’s fields of interest In 1989/1990 Tord visited the L3 experiment a s h i s t e a c h e r, a s h e d id D m it r y Sh i r k o v u n d e r momenta of particles in jets (“handedness”). utions to e+e– collider phenomenology, left us at CERN, starting a fruitful collaboration on w h o s e s u p e r v i s i o n h e d e fe n d e d h i s P h D t h e s i s These effects served as the theoretical basis on 2 April. were broad, and aimed at the application of the ZFITTER project at the “Dispersion theory of low-energy scattering for polarised particle experiments at RHIC at Tord was born in 1951 in East Berlin, educated Large Electron–Positron (LEP) collider. In of pions” in 1962. Br o o k h a v e n, t h e SP S a t C E R N a n d t h e n e w N IC A at the Heinrich­Hertz­Gymnasium specialist predicting observables 1991–1992 he was a research associate in the In 1971, A natoly defended his DSc disser ta- facility at JINR. Anatoly was a member of the mathematics school in Berlin and studied phys­ C E R N t h e or y d i v i s i o n, w or k i n g o ut t h e s o - c a l l e d tion “High-energy asymptotics of Feynman COMPASS collaboration at the SPS, where he ics at Humboldt University in Berlin from 1970. measured at accelerators S­matrix approach to the Z resonance. This diagrams”. The underlying work immedi- h e l p e d t o m e a s u r e t h e e ffe c t s h e h a d pr e d i c ted. He graduated in 1977 with a doctorate devoted was a profound contribution to the field, and ately found application in the factorisation of In 1976 he suggested t he first model for t he to studies of the lattice approach to quantum a breakthrough for the interpretation of LEP hard processes in quantum chromodynamics production of cumulative particles at x > 1 o ff field theory. He obtained a research position t he GDR i n Zeut hen (later DESY Zeut hen), a nd data. Tord was one of the first to realise the (Q C D), w h ic h i s n o w t h e t h e or e t ic a l b a s i s o f a l l Anatoly Efremov was one of the key staff at JINR’s nuclei. Within QCD, Efremov was the first to in the theory group of the Institute of High in 1983–1987 worked at JINR, then in the Soviet great potential of a new e+e– “Tera­Z” factory s hard-hadronic processes. Of particular note Bogoliubov Laboratory of Theoretical Physics. develop the concept of nuclear quark–parton s Energy Physics of the Academy of Sciences of Union, in the group of Dmitry Bardin. at the proposed Future Circular Collider,

66 CERN COURIER SEPTEMBER/OCTOBER 2021 CERN COURIER SEPTEMBER/OCTOBER 2021 67

CCSepOct21_Obits_v4.indd 66 04/08/2021 10:51 CCSepOct21_Obits_v4.indd 67 04/08/2021 10:52 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM

PEOPLE OBITUARIES

FCC-ee, and led the charge reviving precision and Nuclei. In 2015 Tord was awarded an league and family man. We feel a great loss, calculations for it. Alexander von Humboldt Polish Honorary personally and as a scientific community, and Tor d’s s c i e nt i fic fie l d s o f i nt e r e s t w e r e br oad, Research Fellowship. remain thankful for his insights, dedication and aimed at predicting observables measured Tord Riemann cared about high stand- a n d a l l t h e pr e c i o u s m o m e nt s w e h a v e s h a r e d. at accelerators. His research topics included ards in scientific research, including ethical linear-collider physics; Higgs, WW, ZZ, 2f and issues. He w as a t r ue professiona l of t he field. Arif Akhundov, Andrej Arbuzov, 4f production in e+e– scattering; physics at Despite illness, he continued working until his Alain Blondel, Ayres Freitas, Janusz Gluza, LEP and FCC-ee; methods in the calculation of last day. Stanisław Jadach, Lida Kalinovskaya multi-loop massive Feynman integrals; NNLO Tord was an outstanding scientist, a just and Sabine Riemann on behalf of Bhabha scattering in QED; higher-order cor- p e r s on o f g re at hone s t y, a re l i a ble f r ie nd, c ol- h i s c ol le ag u e s a n d f r ie n d s. rections in the electroweak Standard Model and some extensions; and electroweak cor- rections for deep inelastic scattering at HERA. Apart from ZFITTER, he co-authored several programmes, including topfit, GENTLE/4fan, HECTOR, SMATASY, TERAD91, DISEPNC, DISEPCC, DIZET, polHeCTOR and AMBRE. T Riemann T

Tord Riemann promoted the application of the ZFITTER project at LEP.

While being an active research scien- tist throughout his career, Tord will also be warmly remembered as a great mentor to many of us. He w as a t hesis adv isor for t wo d iploma and seven PhD students, and was actively engaged in supporting many postdoctoral researchers. He was co-founder and organ- iser of the bi-annual workshop series Loops and Legs in Quantum Field Theory and of the biannual DESY school Computer Algebra and Particle Physics. I n 2000, Tor d a n d t h e Z F I T T E R c ol l a b or a t i o n were awarded the First Prize of JINR, and in 2014 the article “The ZFITTER Project” was awarded the JINR prize for the best publica- tion of the year in Physics of Elementary Particles

68 CERN COURIER SEPTEMBER/OCTOBER 2021

CCSepOct21_Obits_v4.indd 68 04/08/2021 10:52 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 CERNCOURIER.COM virtual 2021 BACKGROUND IEEE NUCLEAR SCIENCE SYMPOSIUM AND MEDICAL IMAGING CONFERENCE Notes and observations from the high-energy physics community 2021 IEEE NSS MIC 28th International Symposium on Room-Temperature Semiconductor Detectors O telescopes, o mores! From the archive: October 1981 In days of yore, an experiment might choose A lot can happen in 40 years a lovable cartoon elephant for its logo first, “Using gauge ideas, the basic questions – CERN and worry about copyright infringement what are the elementary constituents of later. In a sign those times are long since SKAO matter and the forces among them? – past, the SKA Observatory (SKAO) has launched a 49-page “brand become interrelated through the concept book” to police their swanky new style. Meanwhile, on 29 June, SKAO’s of charges, gravitational, electrical and member states gave the green light for construction in Australia and nuclear, carried by elementary particles, South Africa of the world’s largest radio-telescope arrays. and gauge forces proportional to these charges. A postulated symmetry among arXiv:2106.02048 t he c harges leads to a possible unificat ion of the elementary forces. “But we are still far from the elucidation of the nature of these charges or the problems posed by the mass scales. In the Abdus Salam painting next decade, one may optimistically a bleak picture for the envisage a superconducting pp collider experimental prospects installed in the LEP tunnel reaching 10 TeV of particle physics at the in the centre of mass. But what will UK Royal Society happen twenty-five years from now? conference on gauge We desperately need new design ideas on theories earlier this year. accelerators or they may become as extinct as dinosaurs. “But I am continually amazed how rapidly our experimental Shoot for the moon James Beacham (Duke) and Frank Zimmermann colleagues succeed in demolishing (sometimes (CERN) propose this schematic trajectory (black line) for a 14 PeV circular demonstrating) seemingly inaccessible and often outrageous collider on the moon (CCM) that intersects its poles while minimising elevation theoretical speculations.” Online Conference  Based on text from pp347–349 of CERN Courier October 1981. changes (arXiv:2106.02048). The pictured elevation varies from about –8 km (purple) to +8 km (red). “A CCM would serve as an important stepping stone towards a Planck-scale collider sited in our Solar System,” they write. Compiler’s note 16 - 23 October 2021 In 1979, Sheldon Glashow, Abdus Salam and Steven Weinberg shared the Nobel Prize for the unification of electromagnetic and weak forces. In Energy stored in the LHC’s 1983, the W and Z bosons were discovered at LEP, cementing electroweak magnetic circuit, including 8.8 GJ unification and gaining the 1984 Nobel Prize for Carlo Rubbia and Simon in dipoles and 3.5 GJ in detector van der Meer. In 2010, the first high-energy pp collisions were achieved in the LHC, the collider Salam dreamed of, and in 2012 it delivered the 12.3 GJ magnets. ITER’s magnetic Higgs boson, postulated to explain electroweak symmetry breaking, system, now under construction earning the 2013 Nobel Prize for François Englert and Peter Higgs. And (p13), should break this record, now, experiments at CERN and Fermilab hint at the possible existence of exceeding 50 GJ in the 2030s. gauge leptoquarks and a 5th force. There’s life in the dinosaurs yet.

Media corner “Just a few years ago, such an idea was squarely in “Turning to the heavens, special detectors would “We need a commitment and vision with the General Chair: Ikuo Kanno the realm of science fiction, but now, because there analyse rays from astrophysical sources, and magnitude of CERN if we are to build climate is such a strong interest in humans returning to Moon-based particle accelerators would give models that can accurately simulate extremes Deputy General Chair: Ralf Engels the Moon, a CCM is a distinct possibility.” new insight into the nature of matter.” of climate like the Canadian heatwave.” NSS Co-Chair: Hiroyuki Takahashi India Today’s Sibu Tripathi reports on Isaac Asimov in Popular Mechanics Tim Palmer (Oxford) speaks to BBC News NSS Co-Chair: Craig Woody speculations (see above) about building a (March 1988). (16 July, see also CERN Courier July/August circular collider on the moon (5 July). 2021 p49). MIC Co-Chair: Taiga Yamaya “There’s a quality you shouldn’t have – “How many posters of Van Gogh’s ‘The Starry a need for instant gratification.” “You should know, there are Nobel laureates MIC Co-Chair: Jae Sung Lee Night’ are there out there? They do not devalue 2017 Nobel laureate Barry Barish gives and then there are Nobel laureates.” RTSD Co-Chair: Ralph James the painting at all.” advice to young researchers via the Computer-science-pioneer John Cocke, RTSD Co-Chair: Tadayuki Takahashi Cassandra Hatton (Sot heby’s) just ifies t he v a lue organisation’s social-media channels on quoted in the New York Times obituary for of non-fungible tokens (p10) (CNN, 1 July). World Youth Skills Day (15 July). Steven Weinberg (25 July).

70 CERN COURIER SEPTEMBER/OCTOBER 2021 www.nss-mic.org/2021/ [email protected]

CCSepOct21_Background_v2.indd 70 04/08/2021 11:01 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1 Electronics for science, made by people

That’s CAEN The dream of the founders has now become an international reality

More than 180 professionals are working together to provide the best solutions for Nuclear, Particle, and Astroparticle physics experiments all around the Globe thanks to a vast distribution network.

Master High School

50% 30%

PhD Bachelor

10% 10%

More than forty years satisfying your “electronics” needs

International Offices CAEN S.p.A. CAEN GmbH CAEN Technologies, Inc. Viareggio Solingen Staten Island, NY 10305 Italy Germany USA worldwide www.caen.it www.caen-de.com www.caentechnologies.com presence

Small details… Great differences

CAernCourier CAEN Team ADV_OK.indd 1 03/08/21 14:20 CERNCOURIER www. V o l u m e 6 1 N u m b e r 5 S e p t e m b e r / O c t o b e r 2 0 2 1