Peter Loch University of Arizona

1. Peter Loch University of Arizona Tucson, Arizona 85721 Introduction to Jet/Missing Et/Tau Reconstruction in ATLAS This is not the Physics Department at the University of Arizona! Peter Loch University of Arizona

2. Peter Loch University of Arizona Tucson, Arizona 85721 This Talk • ATLAS calorimetry - a brief introduction into the prominent hadronic final state detectors; • coverage and precision requirements for particle detection and energy measurements in ATLAS; • brief description of detectors; • Jet reconstruction and calibration – where we are and where we want to be; • guidelines and available algorithms in ATLAS; • calorimeter signals: towers and clusters, data flow; • Strategies on finding taus – “classical” discriminants and energy flow; • input to tau likelihood measurement: tracking and calorimeter variables; • use of energy flow concepts; • Calculating missing transverse energy – ambiguity resolution strategies; • event decomposition and refined calibration; • use object based calibrations; • Conclusions

3. Peter Loch University of Arizona Tucson, Arizona 85721 Physics Requirements for Jet and Missing Et Measurements • hadronic final state reconstruction requirements are stringent and often exceed what has been achieved in running experiments at Tevatron and HERA, for example; • top reconstruction in ttbar events requires jet energy scale error of <1% absolute (immensely challenging!); • jets need to be tagged to highest possible rapidities (~5) to enhance Higgs signal-to-background ratio in WW scattering production (order 10% or all Higgs over expected mass range); • good missing Et resolution also requires • largest possible rapidity coverage; • SUSY final state reconstruction also requires • excellent hadronic calibration at a level of 1% to • reconstruct complex decay chains with steep cut- • offs in mass distributions etc.; • Higgs and SUSY need efficient tau reconstru- • ction in “busy” events; • Interesting: • increasing particle detection from |η|<3 to |η|<5 improves mass resolution for a light MSSM Higgs (MA = 150 GeV) from 8 to 2 GeV; • yet, quality requirements to forward particle measurements are relaxed  most missing Et is produced in the central region!

4. Peter Loch University of Arizona Tucson, Arizona 85721 • The ATLAS Calorimeters • Electromagnetic Calorimeters: • Liquid Argon/Pb accordion structure; • highly granular readout (~170,000 channels); • 0.0025 ≤ Δη ≤ 0.05, 0.025 ≤ Δφ ≤ 0.1; • 2-3 longitudinal samplings; • ~24-26 X0 deep • covers |η|<3.2, presampler up to |η|<1.8; • Central Hadronic Calorimeters • Scintillator/Fe in tiled readout; • Δη x Δφ = 0.1 x 0.1 • 3 longitudinal samplings, • covers |η|<1.7; • EndCap Hadronic Calorimeters • Liquid Argon/Cu parallel plate absorber structure; • Δη x Δφ = 0.1 x 0.1 (1.5<|η|<2.5), Δη x Δφ = 0.2 x 0.2 (2.5<|η|<3.2); • 4 samplings; • Forward Calorimeters • Liquid Argon/Cu or W absorbers with tubular electrodes in non-projective geometry; • Δη x Δφ ≈ 0.2 x 0.2 (3.2<|η|<4.9) • 3 samplings; Electromagnetic Liquid Argon Calorimeters Tile Calorimeters η=1.475 η=1.8 η=3.2 Forward Liquid Argon Calorimeters Hadronic Liquid Argon EndCap Calorimeters

5. Peter Loch University of Arizona Tucson, Arizona 85721 The ATLAS Calorimeters Details Cryostat walls (warm/cold) Electromagnetic Barrel Module Hadronic EndCap (2 wheels) to interaction vertex p from LHC Electromagnetic EndCap FCal1 FCal3 Cu Shielding FCal2 total ~200,000 channels, with hadronic coverage ~10 absorption lengths in full acceptance (|η|<5) and a typical level of non-compensation e/h≈1.3-1.6;

6. Peter Loch University of Arizona Tucson, Arizona 85721 Jet Reconstruction Guidelines in ATLAS • Jets define the hadronic final state of any physics channel -> jet reconstruction and calibration essential for signal and background definition; • But which jet algorithm to use ? Recommendations based on CDF & DØ experience from Tevatron Run I† very helpful; • Some “theoretical” requirements: • infrared safety • collinear safety • invariance under boost • order independence (same jet from partons, particles, detectors) • Some “experimental” requirements: • detector (technology) independence • minimal contribution to spatial and energy signal resolution (beyond effects intrinsic to the detector) • stability with luminosity (‼, control of underlying event and pile-up effects) • “easy” to calibrate, small algorithm bias to signal • identify all physically interesting jets from energetic partons in pQCD (high reco efficiency!) • efficient use of computing resources • fully specified (pre-clustering, energy/direction definition, splitting and merging) †G. Blazey et al., “Run II Jet Physics”, hep-ex/0005012v2, 2000

7. collinear sensitivity (1) (signal split into two towers below threshold) infrared sensitivity (artificial split in absence of soft gluon radiation) collinear sensitivity (2) (sensitive to Et ordering of seeds) Peter Loch University of Arizona Tucson, Arizona 85721 Jet Finding Algorithm Implementations in ATLAS (1) • from guidelines + easy implementation -> implemented Kt clustering (exploits kinematical correlations between particles) and (seeded and seedless) cone algorithm (geometrically motivated); • Seeded cone algorithm (most common and fast) has problems with some theoretical and experimental requirements: • But seeded cone is easy to implement and fast -> • added split/merge step helps with dynamics; • alternatively use Kt, seedless cone (typically slow, though!) and mid-point cone; schematics from G. Blazey et al., “Run II Jet Physics”, hep-ex/0005012v2, 2000

8. P.A.Delsart, (U. Montreal) ATLAS T&P Week March 2006 CPU time (arb. units) Peter Loch University of Arizona Tucson, Arizona 85721 Jet Finding Algorithm Implementations in ATLAS (2) • Kt clustering avoids problems of cone • finders, but can be very slow in standard • implementation (CPU time ~n³) -> use pre- • clustering to reduce number of kinematic • objects on input or new FastKt implemen- • tation; M.Cacciari/G.P.Salam hep-ph/0512210; • Common implementation details for all • algorithms: default 4-momentum recom- • bination in jet clustering procedures, user- • defined pre- and final selections, signal treatment/pre-processing (e.g. calorimeter signals, see later), universal jet finder implementations,… • software design (jet event data model, jet algorithm implementations) provides universal jet finders (order independent?): can take tracks, calorimeter cells, -towers, -clusters, energy flow objects, and MC truth objects on input without code changes or adaptations; • in particular, the Jet data object provides a universal representation for all jets, independent of the signal(data) source and jet algorithm choice, with support to retrieve the concrete jet constituents through object navigation (important for calorimeter jet calibration, truth particle nature studies, etc.);

9. Peter Loch University of Arizona Tucson, Arizona 85721 Jet Finding Algorithm Implementations in ATLAS (3) • jet algorithm parameters not universal – narrow jets generally preferred in busy events (SUSY, ttbar with W->jj,…) to increase resolution power of final state; wider jets used in QCD for jet cross-section measurements to assure most complete/correct reconstruction of parton kinematics; • Standard jet collections provided for each event as in indicated in table below; • Discussions with CMS and theory suggested one common algorithm, likely 2-stage mid-point (c.f. J.Huston); mW N.Godbhane, JetRec Phone Conf. June 2006 P.-A. Delsart, JetRec Phone Conf. June 28, 2006

10. Peter Loch University of Arizona Tucson, Arizona 85721 Jet Reconstruction in the ATLAS Calorimeters longitudinal energy leakage detector signal inefficiencies (dead channels, HV…) pile-up noise from (off-time) bunch crossings electronic noise calo signal definition (clustering, noise suppression ,…) dead material losses (front, cracks, transitions…) detector response characteristics (e/h ≠ 1) jet reconstruction algorithm efficiency jet reconstruction algorithm efficiency added tracks from underlying event lost soft tracks due to magnetic field physics reaction of interest (parton level)

11. calorimeter domain jet reconstruction domain physics analysis domain Peter Loch University of Arizona Tucson, Arizona 85721 Calorimeter Jets Today (-> Release 13.0.0) • all electromagnetic energy scale calorimeter cell signals are collected into projective towers (ideal detector geometry, ); • noise cancellation by re-summation is applied to these towers, meaning towers with E<0 are added to near-by towers with E>0 until the resulting protojet has E>0 (all cells are kept!); • jet finding is run on the protojets, resulting in “uncalibrated” electromagnetic energy scale calorimeter tower jets; • the cells in a jet are retrieved using object navigation, and cell signal weights are applied based on cell energy density and location (default) to compensate and correct signals for e/h ≠ 1 and inactive material losses -> hadronic energy scale jets; • additional corrections for residual Et and ηdependencies are applied -> jets are calibrated to particle level (see next slide); • additional corrections possibly derived from W->jj, photon/Z+jet(s) could be applied -> careful, potential biases due to collision physics environment! -> no default today for good reasons!) -> no general refined physics jet yet!

12. Peter Loch University of Arizona Tucson, Arizona 85721 ATLAS Jet Calibration • several schemes under study, most developed (default) scenario is based on fitting cell signal weights in cone jets (R=0.7) in fully simulated QCD di-jet events -> motivated by H1 signal weighting technique; • clearly only possible in MC -> fitting of weights requires choosing truth reference (particle level jets found with the same algorithm, particles pointing • into direction of calorimeter jet…); • approach folds all corrections into cell signal weights -> very little factorization • wrt to different detector • effects, somewhat instable • numerical procedure, intro- • duces jet context bias into • hadronic signal calibration, • on top of calorimeter sig- • nal bias,… • but it seems to work! S. Padhi, ATLAS Physics Workshop 07/2005 very preliminary! Cluster jets Tower jets F. Paige, ATLAS SW Workshop 09/2004 S. Padhi, ATLAS Physics Workshop 07/2005

13. Peter Loch University of Arizona Tucson, Arizona 85721 Problems with Towerjets • use of (quasi-projective) calorimeter cell towers as input to jet finding includes too many non-signal cells into jets due to longitudinal summation and fixed area coverage (non-discriminating summing of cells in tower); • additional sources of noise -> attempt to cancel noise by tower resummation (add “negative” towers to nearby towers with positive signals); C.Roda Hadronic Calibration Workshop July 2005 CERN Entries

14. Peter Loch University of Arizona Tucson, Arizona 85721 Other Problems: “Uncalibrated Input” Miscalibration c: 0.9,0.8,0.7,0.6,0.5 • tower signals used in jet finding calibrated on electromagnetic energy scale -> relative calibration of jet input can be 30% (or more!) off if hadronic! • not too big a problem for cone jets (tbc), but affects shape of Kt jets; • in particular, Kt jets can get much wider if constituent signal too low; • relative uncalibrated input also problematic for kinematic cuts on this input! Radius up to which two objects are combined by the Kt algorithm, as function of the pt ratio of the constituents, with (objects are always combined for if )

15. calorimeter domain jet reconstruction domain physics analysis domain Peter Loch University of Arizona Tucson, Arizona 85721 Calorimeter Jets Tomorrow (Release 13.0.0->) • general attempt to factorize hadronic calibration, dead material corrections, and jet energy scale corrections; • deal with detector (calorimeter) effects in the detector (calorimeter) domain - > common systematic error contribution for all hadronic final state objects! • in particular, e/h compensation and noise can be addressed by replacing towers with “3-d energy blob” (topological cluster) reconstruction, thus avoiding the jet context biases;

16. Peter Loch University of Arizona Tucson, Arizona 85721 Jet Reconstruction from Clusters • form topological cell clusters • efficient and least biased noise suppression ideally provides calorimeter signals of fixed significance with respect to the underlying electronic and pile-up noise; • calibrate clusters to hadronic energy scale (not jet energy scale!) • topological clusters can be calibrated with this respect (“local calibration”) using deposited energies and signals in (pion) MC at cell level; • calibration includes cluster classification, hadronic signal weighting, and dead material corrections; • this takes care of many detector effects (e/h, energy losses in upstream and internal dead material…); • already very useful as input to missing transverse energy calculations; • note that local cluster calibration derived from MC can be applied to testbeam pions for comparisons!! • use calibrated clusters as input to jet finding • reduce potential 30% relative miscalibration to ~<5% or so -> enhance jet reconstruction stability, especially for Kt; • apply jet energy scale corrections • very specific corrections dependent on jet algorithm and algorithm parameters, physics environment (underlying event contribution, final state multiplicities and particles,…); • one or two step normalization: directly to parton level derived from Z/γ+jet(s) or W->jj (careful about biases!) or first particle level corrections, then particle->parton level corrections;

17. Peter Loch University of Arizona Tucson, Arizona 85721 Calorimeter Hadronic Signal Definition • topological clustering in three dimensions, with • local maxima based splitting, attempts to recon- • struct cell signal correlations created by shower • development; • implements noise suppression based on signal • significance of cell and (!!) its neighbours -> much • less noise cells included in hadronic calorimeter • signal, but some small signal bias unavoidable; • some sensitivity to shower structures allows classification/tagging of clusters -> selective and optimized calibration procedures for individual clusters possible; I.Vivarelli, Calorimeter Calibration Workshop, September 2006

18. cluster candidate #1 #3? cluster candidate #2 Peter Loch University of Arizona Tucson, Arizona 85721 R.Seuster, priv. comm. September 2006 Kt, Et>200 GeV not understood drop in #clusters in central region (finest granularity!) Topological Clusters in Jets I.Vivarelli, CCW 2006, September 2006

19. ParticleJet CaloJet (def. calib) CaloJet (addt’l corr.) Peter Loch University of Arizona Tucson, Arizona 85721 Other Early Observations for Jets from TopoClusters Cone jets with R=0.7 in QCD di-jet events • jet finding efficiency fortopoclusters: jet seeds more efficiently found due to thecombination of cell signals and free location in the calorimeter (especially for low energetic jets); towers can actually reduce thejet finding efficiency due to their fixed grid event view with possible splitting seed energy deposit into signals below threshold (shower spread distorts “perfect” transverse energy flow in cone); • additional cluster features like em tagging can maybe help with jet characterization and calibration; • presently cluster based jets have typically 5-10% less signal than the default calibrated jets -> missing some out-of-cluster corrections, O(1%) for single pions and O(10%) for jets (expected to be fixed for 13.0.0); TowerJets ClusterJets F.Paige, private communications, May 2006

20. 1 charged ID track & calorimeter signals: • standard egamma reconstruction: em calorimeter “sliding window” cluster matched with track (+ cluster shapes, isolation for identification); electron kinematics taken from corrected calorimeter cluster signal; • softe reconstruction: more strongly curved low pt track is extrapolated into em calorimeter, used to seed the cluster direction (η,φ) (+ cluster shapes and isolation); kinematics taken from corrected calorimeter cluster signal; • ambiguity resolution: if same track reconstructs standard and soft electron, standard electron is kept for physics analysis; • 1 charged ID track & calorimeter signals: • calorimeter signal: “hadronic sliding window”, topological clusters (“3-d energy blobs”), jets from cone ΔR=0.4 and Kt, matched with track; use cluster shapes and energy sharing; • 3 charged ID tracks & calorimeter signals: • calorimeter signal: same as in single prong case; • ID track: secondary vertex; Peter Loch University of Arizona Tucson, Arizona 85721 • Tau Reconstruction • leptonic tau decays (~35%) have signatures including two neutrinos and one electron or muon (each ~50% of leptonic modes); • hadronic tau decays (~65%) are mostly single prong (1 charged pion,~75% of all hadronic decays) and less often three-prong (3 charged pions,~25% of all hadronic decays); not discussed here!

21. Peter Loch University of Arizona Tucson, Arizona 85721 Hadronic Tau Identification: Discriminants Plots by M. Heldmann, presented by L.Zivkovic at ATLAS Physics Workshop, Rome, June 2005 EM Cluster Radius Isolation Fraction # ID Tracks Charge # η Strips in Calorimeter Charged Impact Parameter Et/pt of 1st Track Cluster Width in η Strips

22. likelihood pt dependent likelihood cuts (6 var.) Peter Loch University of Arizona Tucson, Arizona 85721 Hadronic Tau Identification: Efficiency and Rejection • efficiency in tth->ttττ~40-50% over dynamic range; • QCD jet rejection between ~10-~100 at the same efficiency; All plots by M. Heldmann

23. Peter Loch University of Arizona Tucson, Arizona 85721 Energy Flow Approach (F. Tarrade, HCP2005) • well established technique (LEP,HERA…): use tracking for kinematics reconstruction of charged particles when preferable from point of detector resolution; • requires to “remove” the charged particle signal from the calorimeter if tracking signal is better -> non-trivial for high occupancy LHC scenario; • but well applicable for tau reconstruction, where object has a well-defined topology and is somewhat isolated, especially in light Higgs or soft SUSY, with Et(tau) ≈ 20-70 GeV in central region; • technique delivers similar performance to previously discussed method, with some improvements in low pt efficiency (E. Richter-Was, D. Froideveau,…); • energy flow approach is also studied in other physics contexts by UK group (Sheffield, D. Tovey, M. Hodgkinson,…); (by O. Oye) red – tau1P black - tauRec

24. Peter Loch University of Arizona Tucson, Arizona 85721 Default Missing Et Calculation

25. Peter Loch University of Arizona Tucson, Arizona 85721 Default Missing Et Performance D.Cavalli, CCW San Feliu 2006

26. Peter Loch University of Arizona Tucson, Arizona 85721 Alternative Missing Et Calculations: Object-Based B.Mellado,S.L.Wu et al. Electrons from Egamma collection IsEm=0 L.Flores sampling based calibration Track from TrackParticles E>2Gev, good quality, isolation. Remove overlap with e, muons, jets Muons from StacoCombined collection Emuon-Eloss. Eloss in calo is standard parametrisation based Mini Topo-jets from low pT Topoclusters separated in 0 (sampling based calib) and charged  (cell based calib) Cone 0.7 Jets from Topoclusters & pseudo-H1 S.Padhi calibration. Jet clusterisation correction remaining cells with a scale factor depending on 3 regions

27. Peter Loch University of Arizona Tucson, Arizona 85721 Missing Et Calculation Evolution: Refined Calibrations • basic idea is to decompose calibra- • ted final state objects back into calo- • rimeter cells, with a weight attached • to each cell signal reflecting the final • calibration (accumulation of geometri- • cal and calibration weights along the • reconstruction sequence); the weigh- • ted cell signal can then be used in the • missing Et calculation; • if a cell has already been found in a • final state object of a different type, • it is not used again; • if a cell has already been found in • an object of the same type, the • weights are added up; • the priority in cell usage is defined • by the tool sequence used to retrieve • the cells and their weights; typically identified particles first: electron/photon->tau->jet->rest; • unused cells are included only if they are in topological clusters (survived noise suppression); • approach provides ambitguity resolution without geometrical assumptions or overlap measures;

28. Peter Loch University of Arizona Tucson, Arizona 85721 Conclusions • hadronic final state reconstruction in ATLAS is still a very dynamic field… • hadronic signal likely based on calibrated topological clusters still needs fine-tuning concerning cluster sizes in the central region; • not all effects of the signal selection in clusters is understood: stability against changes in noise/pile-up fluctuations, sensitivity limit to low energetic hadronic energies,… • recent activities have started to develop strategies to monitor and improve cluster formation and calibration in the pp (low lumi) environment; • …but we have come a long way since Rome: • jet calibrations have been stabilized (there are probably still too many similar ones); • clear improvement in software, with new features like refined and object based missing Et, a common EDM for taus, clearly structured jet reconstruction algorithms and additional data for jets (jet moments); • some work on online and quasi-online monitoring of hadronic final state reconstruction performance has started -> preparations for early intervention; • CSC notes • some topics not yet covered (I will send a mail concerning the overall jet performance note next week); • hope to get cluster calibration to be usable for CSC reconstruction (13.0.0?);

29. Peter Loch University of Arizona Tucson, Arizona 85721 Final Comment When collecting this talk I noticed that the documentation of the jet reconstruction software is not at all satisfactory. With this regard Chiara and I will see if we can start a series of jet reconstruction software tutorials soon (before lots of work for the CSC actually starts), if there is sufficient interest, that is… Besides that we will try to improve the Wikis etc.

30. Peter Loch University of Arizona Tucson, Arizona 85721 Backup Slides

31. Peter Loch University of Arizona Tucson, Arizona 85721 B-jet tagging • basically implemented as calorimeter jet classification; • first, a cone jet with R=0.4 is found in the calorimeter, then the ID tracks pointing into this cone are collected; • the tag information is calculated from this collection of tracks; • performance seems to be ok this way, but I would have expected that finding a jet in charged tracks first, and then collect the calorimeter signal to determine the complete kinematics would have been the more appropriate way; • presently technical problems and software design issues a big issue – representation of tagged (and all other!) jets for physics analysis, how to attach the tag information, etc.

32. Peter Loch University of Arizona Tucson, Arizona 85721 Use of ID Tracks in Calorimeter Jet Calibration • presently calibration normalization from MC truth jets of stable particles -> typical cut-off ~300 MeV; • calibration to particle level is consistent, but corrections to parton level (implicit in in-situ calibration scenarios!) need to be understood as well; • also, number of charged tracks pointing to a jet, and energy carried by those tracks, could give an idea on the hadronic content of the jet -> help with calibration of non-compensating calorimeters (not clear if correlation is good enough or killed by hadronic shower fluctuations anyway); • isolated hadronic tracks can be seen as a in-situ testbeam -> careful look at taus with low pt track(s) -> magnetic field provides enough separation to disconnect the hadronic response from the neutral pion/photon response ? Which cut-off, triggers,…? • low end hadronic calibration is important (most jets have significant number of low energetic particles, even at high total pt) -> can we find isolated hadronic tracks in minimum bias events at intial luminosity? Trigger? Isolation criteria/measure for neutral background?

33. Peter Loch University of Arizona Tucson, Arizona 85721 New Approach in ATLAS: Local Hadronic Cluster Calibration • use the 3-d nearest neighbour topo- • logical cell clusters as a basic calorimeter • signal; • move the cell level calibration, • basically intended to compensate • e/h ≠ 1, from the jet context to the • cluster context  expect more sensitivity • of cluster shapes to nature of energy • deposit; • classify each cluster according to its • shape and location as “electromagnetic” or • “hadronic” (possibly one or two hadronic classifications!); • apply the best calibration function to each cluster, depending on its classification and possibly cluster (shape) and/or signal variables; • apply dead material and crack corrections at cluster level, too  (nearly) all detector effects are calibrated out or corrected before jet finding, or missing Et calculation  common basic hadronic signal for both!!

34. Peter Loch University of Arizona Tucson, Arizona 85721 Final Jet Calibration • using fully calibrated and corrected clusters in jet finding reduces the problem of jet calibration to the understanding of the contributions from the jet algorithm inefficiencies, the underlying events, and the overall event topology possibly including pile-up; • W  jj can help to estimate these final corrections, but are mostly found in a very specific ttbar topology (bias ?), and with special jets (no color link to rest of event); also, there are kinematical limits on the effectiveness of this calibration signal – order(3%) more likely than 1% today; • other channels like Z+jet(s) • or photon+jet(s) can help, but • good understanding of initial and • final state radiation needed; η(W) ~1.8 P. Savard, P. Loch, CALOR97

35. Peter Loch University of Arizona Tucson, Arizona 85721 Missing Transverse Momentum Reconstruction • best missing Et calculation is using calibrated cell signals and all cells with true signal; • in a real detector this calculation is very • sensitive to electronic noise (at least) – • typically 70-90% of all cells in ATLAS • have no true or significant signals; • symmetric or asymmetric cell noise cuts • reduce the fluctuations significantly, but in- • troduce a bias (shift off 0) due to this cut; • topological clustering imposes a noise cut, • but lets cells survive based in the signal in • their neighbours  less bias, yet near • optimal suppression of incoherent • (electronics) noise; K. Cranmer, in talk by S. Menke, ATLAS Physics Workshop 07/2005

36. Peter Loch University of Arizona Tucson, Arizona 85721 Missing Et Reconstruction • calculating missing Et from topological clusters brings a small improvement in the bias; S. Resconi, ATLAS Physics Workshop 07/2005 • calculating missing Et from topological clusters improves the missing Et resolution for many physics channels; SU1 W  VBF h(130 ) Z°  SU2 A°(600) W e A°(300) A°(450) ttH   Z°   A°(800) S. Resconi, ATLAS Physics Workshop 07/2005

37. no pile-up added Et ~ 81 GeV LHC design luminosity pile-up added ~7.5 charged particles/unit rapidity ~8.5 neutral particles/unit rapidity Et ~ 58 GeV Number Charged Tracks/(Δη=1) Number Neutral Particles/(Δη=1) Pseudorapidity η Pseudorapidity η Peter Loch University of Arizona Tucson, Arizona 85721 Pile-Up • high lumi, large -> ~23 min bias events/bunch crossing, with ~75 charged tracks/event within typical detector acceptance |η|<5 at 1034 cm-2s-1; ATLAS Calorimeter Performance Technical Design Report CERN/LHCC 96-40 • effect of pile-up depends on detector technology and readout electronics - long bi-polar shaping functions in ATLAS calorimeters lead to out-of-time contributions with negative signals;

38. Peter Loch University of Arizona Tucson, Arizona 85721 Pile-Up vs Electronic Noise in Calorimeter Cells • cell signal fluctuations introduced by pile-up dominate wrt electronics noise in the ATLAS endcap calorimeters, but are comparable to, or less than, electronic noise in the central calorimeters; • pile-up fluctuations in a given cell are not Gaussian due to the lateral and longitudinal coupling of signals in neighbouring cells introduced by the showers in minimum bias events; • still, pile-up RMS in each cell (a function of the instantaneous luminosity) can be used to define the significance of the cell signal to first order; • still more detailed studies are needed to understand the structure (mini-jets, etc…) in the pile-up events, and to find a measure for the instantaneous lumi for each triggered event (correlation between forward energy flow and pile-up ?); S. Menke, ATLAS Physics Workshop 07/2005

39. Signal Significance Jet Cone Size R 1034 cm-2s-1 P.Loch & P.Savard, CALOR97 Peter Loch University of Arizona Tucson, Arizona 85721 Pile-Up in Jets • typically around 10 GeV Et signal fluctuations in jet cone of R = 0.5 at design luminosity; • this finding is rather independent of η; R = 0.5 P. Savard et al., ATLAS-CAL-NO 084/1996 Forward tag jets in Higgs production have average Et of 30-50 GeV only (descreasing with increasing η) • forward jet detection is non-trivial (low significance of signal even in smallest jet cone); • further studies needed to understand pile-up contributions to other jet finders;

40. Peter Loch University of Arizona Tucson, Arizona 85721 …Longitudinal Energy Leakage… • energy splash into muon system indicates leakage for high energetic jets – how to tag this to suppress fake X-section enhancements at high transverse momentum? Any estimates of the amount of energy lost possible? calorimeter leakage? crack effect? Frank Paige, Jet/EtMiss/τ performance meeting, ATLAS T&P Week February 2006

41. stable operation only for r ≤ 1 J.P.Rutherfoord, Scientific Note SN-ATLAS-2001-003 Peter Loch University of Arizona Tucson, Arizona 85721 Jets in SLHC • several upgrade scenarios have been discussed - increase of luminosity by 10, doubling of energy at present design lumi, doubiling of energy and increase of lumi by 10; • due to technical constraints from the LHC machine, increasing the energy above 7.54 TeV/beam is only possible by replacing (all) beam elements (not really an option); • lumi only upgrade somewhat more obvious – mainly changing cavities only; • even that is challenging for the ATLAS calorimeters (space charge effects etc.); • some ongoing studies to understand • the benefits for physics versus the • impact on the detectors and machine;

42. Peter Loch University of Arizona Tucson, Arizona 85721 Jets Performance Degradation • major problem is increase of pile-up activity (23 events/crossing to 230 events per crossing); • pile-up signal fluctuations in jet cone increase by ~sqrt(10) to more than 30 GeV  performance degradation e/jet separation, b-jet tagging, Higgs tagging… R = 0.4 R = 0.2