Measurement of jet properties with the ATLAS detector Hugo Beauchemin Tufts University Put photo of team. DIS2016, DESY, Germany, April 13th 2016

Introduction

The strong interaction intervenes in various ways and at various scales in every single event at the LHC Matrix element of interest Gluon emission (ISR/FSR) Proton structure Fragmentation and hadronization Underlying event Long distance physics Short distance physics Long distance physics We can make robust predictions! Not so sure…

Factorization theorem: The probabilities for short-distance and long-distance processes factorize The long-distance factors are universal and can be empirically obtained from ancillary measurements. Ok, at first order we are fine: even if there is a regime where QCD becomes non-perturbative, and even if every single events will involve such long distance physics effect, factorization theorem allows us to make robust predictions at the LHC for any processes of interest. There are nevertheless what I would call second order effects that could be quite annoying, and which requires a sophisticated measurement program at the LHC. Evolution equations (e.g. DGLAP), analogous to b-functions for aS, account for transition from one scale to the other

These QCD predictions involve assumptions, approximations, and phenomenological modeling impacting final state selections and differential cross section predictions Parton shower accounting for the effect of evolution on final states: Soft and collinear approximations (where QCD radiation is enhanced) Leading order kernel functions Choice of ordering parameter Parton distribution function (PDF): Uncertainties on measurements used to extract structure functions Modeling of structure functions at Q0 Fragmentation function: Gaussian modeling of D(x,s) at small x Supplemented by hadronization model

Predict different number of charged particle within jets and different distribution of the charged-particle energies in jets. These effects implies possibly large uncertainties affecting the modeling or measurement of: Observables likes Mjj, ETmiss-related, jet gap-related Other infrared sensitive observables Precision beyond 1/Qn for infrared safe observables

Measuring observables sensitive to these effects allows to: Determine best model/calculations for predictions Tune fragmentation and FSR parameters or PDF fit. Assuming PDF, measuring observable sensitive to the transition from parton to jet can be used to: Tag jet flavor Distinguish between q-jet and g-jet Distinguish between qq-resonance and QCD background

ATLAS measurements: Jet-charge and charged-particle multiplicity of jets Phys. Rev. D93 (2016) 052003 arXiv:1602.00988 [hep-ex]

Jet charged-particle multiplicity An observable that is sensitive to longer-distance QCD effects To be measured for various PT track threshold, thus regulating the sensitivity to soft particles This observable depends on the parton that initiated the jet Gluon jets contain more particles than quark jets because of larger color charge

Jet-charge Definition: The momentum-weighted charge sum of particles in jets is sensitive to the electric charge of quarks and gluons from jets Definition: Measure mean mQJ and standard deviation sQJ vs PT jets regulates the sensitivity of the jet charge to FSR and fragmentation Low k enhances contribution from low-PT particles k = 0.5 is the most sensitive value to parton charge initiating the jet

The average jet charge mQJ increases with jet-PT due to an increase in up-flavor from PDF at the origin of hard jets Also more visible when the soft radiation is not suppressed Distribution varies with rapidity due to higher up/down than gluon contributions in more forward regions Discrepancies with predictions, meaning that large-distance QCD effects need to be better understood.

Similar observations about data to predictions agreement from measurements of the charged-particle multiplicity <ncgh> in jets

Sensitivity to PDF Each PDF set has its own underlying event tune, factorizing out these effects from interpretation Better agreement Pythia (LO generator) with LO PDF (CTEQ6L1) than with any NLO PDF, but still has wrong shape Standard Deviation of jet charge sQJ well described by all. Decoupled effects from FSR and fragmentation to a large extent

Sensitivity to parton emission (FSR), fragmentation and hadronization Discrepancy between data and MC is not fully explained by PDF Sensitivity to parton emission (FSR), fragmentation and hadronization Compare data to different hadronization and FSR predictions : Compare pythia 6 default tune with RadHi and RadLo tunes to quantify how the observable is sensitive to QCD FSR The scale in as regulating FSR is changed by ½ and 2 for these two tunes Realistic tune with some of the effects being absorbed in tuning Compare Pythia 8 with Herwig++ to assess the sensitivity to fragmentation parameters and hadronization model

Significant sensitivity to fragmentation and hadronization Average jet-charge mQJ distribution The low k mQJ observable is insensitive to FSR FSR effects are drown in the sensitivity of the observable to soft particle from infrared processes Significant sensitivity to fragmentation and hadronization

Average jet-charge mQJ distribution Sensitivity to FSR increases with k Suppression of soft radiation makes the jet charge distribution more sensitive to the modeling of the energy fraction of leading emission

Jet charge standard deviation sQJ Fluctuation decreases as soft radiation is suppressed Because we see less charge Significant sensitivity to FSR at large k, but not to fragmentation or hadronization. Good for FSR tuning Hard to disentangle FSR from fragmentation at low k

Multiplicity of charged particles <nchg> in jets FSR: too much radiation in the main pythia P2012 tune Insensitive to underlying event pythia tune (AU2 vs Monash), but not to Herwig corresponding tunes (EE3 vs EE5) Best description at low jet-PT by Pytha 8 A14 tune, but at high jet-PT by Herwig EE5 tune ATLAS charged-particle multiplicity in jets of PT<50 GeV included in A14 tune, and lower as FSR

mQJ differences between up and down Using flavor-fraction information in PDF and matrix element calculation, we can extract mQJ and <nchg> for various parton flavor initiating the hard jets mQJ differences between up and down <ncharged> differences between quark and gluon

Conclusion

ATLAS performed a comprehensive study of jet-charge and jet charged-particle multiplicity sensitive to the transition between parton and hadrons Sensitive to PDF, FSR, fragmentation and hadronization Jet charge mean mQJ , and standard deviation sQJ, and jet charged-particle multiplicity <nchg> are not well described LO PDFs provide a better description for LO generators, but cannot fully explain data to prediction discrepancies Low k mQJ is good to test/tune fragmentation functions High k and are good to test/tune QCD radiation <nchg> infrared sensitive to underlying event modeling These observables can be used to separate between jets of different parton origins

Back-up slides

Pythia (8.175) vs Herwig++ (2.63) Hadronization: Lund string (linear confinement) Hadronization Cluster (preconfinement) -Shower ordering: determine evolution of resolved QCD emission -Linear confinement: color field stretching between quark-anti-quark pair like a string breaking spontaneously in segment of q-qbar pairs. -Preconfinement: pair of color connected neighbor partons with a universal and Q2-independent asymptotic mass distribution suggesting particles. Check ref 47 and 48 of jet-charge for the discussion on use of NLO PDF on LO ME. pT-ordered showers (natural for a shower partly based on dipole approach) Showers ordering: (nLL resummation) angular-ordered showers (deal with quantum interference)

Pythia (8.175) vs Herwig++ (2.63) PDF: Different can be used in generators, but as default: CT10 (NLO) -Shower ordering: determine evolution of resolved QCD emission -Linear confinement: color field stretching between quark-anti-quark pair like a string breaking spontaneously in segment of q-qbar pairs. -Preconfinement: pair of color connected neighbor partons with a universal and Q2-independent asymptotic mass distribution suggesting particles. Check ref 47 and 48 of jet-charge for the discussion on use of NLO PDF on LO ME. CTEQ6L1 (LO) AU2 & A14 tunes (Many parameters) Multiple interactions: (Each generator is tuned for different PDF on Run-1 ATLAS data) EE3-EE5 tunes (Fewer parameters)

Event topology I Measure <ncgh>, mQJ and sQJ in dijet events with balanced PT More sensitive to hard scatter jets and suppresses MPI Measure impact of large-distance effects on short-distance processes Less experimental and theoretical confusion with close-by jets Hard-scattered quarks and gluons more cleanly matched to jet

Event topology II Separate central from more forward jets because jet flavor depends on rapidity Forward jets are less likely to come from gluon, than central jets Forward jets are limited to |h| < 2.1 to detect all tracks Define jet flavor from the highest energy parton within DR < 0.4 of particle-level jet

Unfolding Detector effects distort the underlying physics distribution The average effect per jet-PT bin is about 10-20% Reconstructed observable distributions are unfolded to produce results directly comparable to theoretical calculations Unfolding accounts for fake, efficiency, and resolution effects Unfold for <ncgh>, mQJ and sQJ and PT Iterative Bayesian technique Dominant source of uncertainty in low jet-PT bins

Variation of jet flavor origin with jet-PT and for different rapidity regions Gluon and sea-quark fractions decrease with jet-PT Up and down contributions increase with jet-PT Up increases faster than down with Jet-PT, but this effect get attenuated in forward regions

Sensitivity to PDF does not depend on the amount of soft particles accounted for in the jet-charge definition: PDFs impact leading jet charge mostly through flavor fraction Decoupled effects from FSR and fragmentation to a large extent Some systematic trend might be observed with Herwig, but not significant