Johns Hopkins University and CMS B. Barnett, B. Blumenfeld, A. Gritsan, P. Maksimovic M.Swartz Johns Hopkins University Baltimore, Maryland, USA 1
JHU CMS Group Person Position Experiments (8) Recent CMS Activities B. Barnett Faculty CMS/CDF(7/) FPix sensor/detector test, mat. val. B. Blumenfeld Faculty CMS/CDF(9/1) CMS Computing (FroNTier DB) A. Gritsan Faculty CMS/BaBar(75/5) Tracker Alignment P. Maksimovic Faculty CMS/CDF(75/5) Pixel Offline Software M. Swartz Faculty CMS Pixel Offline Software G. Giurgiu Postdoc CMS/CDF(9/1) Pixel Offline Software Z. Guo Postdoc BaBar/CMS(5/5) Tracker Alignment C.-K. Lae Postdoc CMS DCS, Tracker Alignment S. Rappoccio Postdoc CMS Pixel HLT, b-tag, Offline Software D. Fehling Student CMS Pixel Offline Software N. Tran Student CMS Tracker Alignment C. Eskew Student CMS Join May 8 J. Rovner Student CMS Join May 8 Y. Zhang Student CMS Join May 8 J. Shaev Underg. CMS Pixel Offline Software
Recent CMS-related activities are summarized in the previously submitted document. Highlights of recent work are presented here: Recent Activities Offline Software - Pixel Template Reconstruction: based on successful modeling of pixel sensors before and after irradiation (MS + V.Chiochia, Zurich) Read-Out Chip depleted region Charge (Arbitrary units).5 1.5 1.5 V bias =15 V (a) Charge (Arbitrary units).5 1.5 1.5 V bias = V (b) Charge (Arbitrary units).5.5 1.5 1.5 V bias = V (c) Charge (Arbitrary units) 4.5.5 1.5 1.5 V bias =45 V (d) 5 1 Position (µm) 5 1 Position (µm) 5 1 Position (µm) 5 1 Position (µm)
- Pixel Template Reconstruction (Swartz, Giurgiu, Maksimovic): scale and translate simulated shapes to fit observed cluster projections! tricks to control delta rays! improves resolution (incl tails) before and after irradiation! goodness-of-fit probabilities test cluster shape hypotheses! CMS Note 7/ (public) Before Irradiation After Irradiation (a) (b) (a) (b) 4
- Pixel Template Reconstruction: use technique in second pass track fit (angles from 1st pass) - becomes baseline in next CMSSW release! Study with sample of simulated muon tracks (GG) " template reconstruction kills tails! " biggest improvements are in d,! pulls in the regions > " " expect to see significant S/N improvements in b/#-tagging pull of d parameter -1 1-1 - 1 pulld Entries 1461 Mean.1581 RMS 1.85! / ndf 48. / 159 Constant.711!.77 Mean.11!.91 Sigma 1.87!.7 d 1 GeV $ s + standard alg + template alg pulld Entries 1461 Mean.157 RMS 1.16! / ndf 18 / 74 Constant.749!.8 Mean.185!.88 Sigma 1.58!.8-4 1-5 - -15-1 -5 5 1 15 5 pull pull of of " dz parameter parameter 1 1 1 1 1 1-1 -1 - - - - pulldz pullphi Entries 1461 1461 Mean.5677 Mean.185 RMS 1.649 RMS 1.56! / ndf 496.8 / 15 Constant / ndf.6856! 47.4.77 / 18 Mean Constant -.198.7174!.948!.8 Sigma Mean 1.14.175!.8!.9 Sigma 1.81!.8! pullphi pulldz Entries 1461 Entries 1461 Mean.188 Mean.66 RMS 1.16! RMS / ndf 1.51 185 / 7 Constant! / ndf.7489 71.5!.79 / 95 Mean Constant.78.178!!.8.88 Sigma Mean.1615 1.5!.8889!.7 Sigma 1.67!.8 1 1-4 -4-5 - -15-1 -5 1 15 5-5 - -15-1 -5 5 1 15 5 5
- Pixel Template Reconstruction: D. Fehling, P. Maksimovic have created a template-based seed cleaner that uses the angle-sensitive probabilities. Tested with a sample of 75 simulated t-tbar events: Seed Generator.1 s/event 185k seeds Kalman Filter 1.8 s/event 7.6k tracks 1.9 s/event 185k seeds Seed Cleaner.6 s/event 476k seeds Kalman Filter.96 s/event 7.k tracks 1.15 s/event! Reduces number of seeds and tracking time by factor of ~! Loses 1.6% of tracks! quality of lost tracks is unknown as yet! May gain more from cut optimization and use low-q cut 6
- Pixel Template Reconstruction: D. Fehling has studied the effect of the nd-pass template reco and templated-based seed cleaning+nd-pass reco on b-tagging:! Use 8-1 GeV P T QCD events! Track counting doesn t need re-calibration! track probability also improves /wo calib! Improvement in mistag rate is in range -!! Shows that the seedcleaner improves the track quality udsg-efficiency FlavEffVsBEff_DUSG_discr_trackCountingHighEffJetTags_GLOBAL - 1 Standard Reco Template Reco Only Template Seeding+Reco.5.4.45.5.55.6.65 b-efficiency 7
- FPix-material budget verification (Barnett + V.Cuplov Purdue, Calumet): material as measured in CMSSW simulation agrees with the measured weights at the % level! - tau-tau HLT (Rappoccio): implement new isolation criteria at the.5 level using triplet tracks and loose hits Tau Track Isolation Jet Hit Isolation Track Hits Loose Hits Track 8
- Physics Analysis Tools Group: co-lead by P. Maksimovic and charged with providing/maintaining all software used in physics analyses! Responsible for: high level objects (electrons, b-jets, candidates), composite candidates, event hypothesis, collections, filters, generic histogramming/ntupling, trigger confirmation, efficiency determination, MC matching, fitting data distributions, visualization (event displays like Iguana)! Organization: actual development is done by the Physics Toolkit Taskforce (also co-lead by PM, SR is a contributor):! Official set of " algorithms (for cleaning and selection) " high-level objects (PAT objects)! Crucial: defines the scope of early physics! " skimming will be done centrally " most users will use PAT objects directly! Starter Kit: a set of always-functioning examples of simple analyses that new (and old) users can use to get started 9
Tracker Alignment (Strips + Pixels): A. Gritsan, C.-K. Lae, N. Tran - Tracker Alignment Group now co-lead by A. Gritsan - Build framework to include hierarchical survey information from a database in the alignment procedure Tracks Combine Survey χ (R,Ω)= N hits i=1 ε T i V 1 i ε i + 1 N str. j=1 ε T j,surveyv 1 j,survey ε j,survey
- Forward Pixel Survey (NT): used CMM, photogrammetry, touch probes! Survey goals! sensor positions/orientations! hierarchical errors Typical errors: ( R, Ω) halfcylinder (σ > 5µm) ( R, Ω) halfdisk (σ > 1µm) ( R, Ω) blade (σ 1µm) ( R, Ω) panel (σ 1µm) ( R, Ω) sensor (σ few µm) 11
- Realistic MC Studies: track + survey alignment, systematics! Align over 15, modules! 6 position/orient. params! Alignment of the full tracker! framework! strategy! implementation Tracks only Tracks + Survey x y z x y z!x (!m) "1!y (!m) "1!z (!m) "1!x (!m) "1!y (!m) "1!z (!m) "1 1 1 1 1 1 1-1 -1-1 -1-1 -1 - - - - - -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration (mrad)!" x (mrad)!" y (mrad)!" z (mrad)!" x (mrad)!" y (mrad)!" z 1 1 1 1 1 1-1 -1-1 -1-1 -1 - - - - - - -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration 1 -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration -.5 1 1.5.5.5 4 4.5 5 iteration
- Analysis of Alignment Data:! AG lead the cosmic alignment performed in the Tracker Integration Facility during summer 7 - winter 8! public CMS Note in preparation! Performed Alignment for the Computing/Software/Analysis 7 Challenge (CSA 7) [CKL]! summer-fall 7, test of complete workflow 1
Detector Control System (C.-K. Lae) - build database and graphical interface for the configuration of the DCS including configuration of the Finite State Machines - design of the graphical panels for reading the temperatures, humidities, currents and voltages to monitor the state of the detector - provide DCS support to the commissioning of the Forward Pixel Detector at the Tracker Integration Facility. 14
CMS Computing (B. Blumenfeld) - Frontier Database Access: efficient delivery of all non-event data to all T, T1, T, T CMS jobs and the HLT farm! essential part of all CMS computing! developed by JHU-FNAL Collaboration! deployed at 5 sites in 18 countries! worked flawlessly during CSA7! BB developed special configuration for High Level Trigger Requests served by local cache Requests served by remote source 15
! implemented, maintained, supported by only people! Blumenfeld is the only physicist involved - Created/operates Monitor Alert for Frontier and expanded to CMS Global Data Bookkeeping System 16 15
Forward Pixel Detector Testing (B. Barnett) - ~ 5% of sensors tested at JHU (BB, MM, + undergrads) - testing of assembled plaquettes at Fermilab SiDet 17
Outreach (Barnett) - Quarknet: 6th annual summer workshop for Baltimore area teachers - Physics Fair: 4th fair held in April 7 - EPPOG MasterClasses: teachers and 6 students in 7, teachers and 1 students in 8 18
Future Activities Offline Software - Pixel Template Simulation (MS, GG): need to be able to quickly simulate radiation damaged pixel detector in CMSSW! detailed simulation: s/cluster; CMSSW simulation (no damage): 1 ms/cluster! use templates to re-weight CMSSW simulated clusters (solve 1-D projections to -D weights):. ms/cluster (a) (b)! basic re-weighting procedure is written+tested, implementation in CMSSW needs infrastructural work to handle edge clusters 19 18
- Pixel Template Reconstruction Calibration Suite (MS, DF, PM,?): we know how to scale our detector model with n-equiv fluence: N A (Φ)=R A (Φ)N A (Φ ), N D (Φ)=R D (Φ)N D (Φ ), Γ e/h (Φ)=R Γ (Φ)Γ e/h (Φ )! Need to vary ~1 parameter to tune model! Repeat beam test charge collection profile measurements in-situ in CMS! many inclined tracks even near barrel center! must live with zero-suppressed readout! need fine voltage scan to compensate! Useful in commissioning the pixel system 6 5 4 14! eq ="1 n/cm (d) V bias =5 V Data PIXELAV Best Fit V bias =5 V (e) V bias =1 V (f) V bias =15 V (g) Charge (A.U.) 1 invisible 5 1 Position (!m) 5 1 Position (!m) 5 1 Position (!m) 5 1 Position (!m)
-Pixel Template Cluster Splitter (MS, GG, PM,SR): use template reconstruction to develop an algorithm to split merged clusters in high-pt jets (interest driven by search for t-tbar resonances)! Step 1: first pass tracking with loose cuts on %! Step : examine template probabilities of tracked pixel hits! if small, try fitting two hit hypotheses in both projections! take the angles to be the same for both hits! should improve template probabilities! produces 4 new hits w/ - fold ambiguity (-x X -y coords)! Step : re-track event w/ tighter cuts 1 hit 1 hit
- b-tagging (PM, GG, DF, SR): expand work on tracking/b-tagging! finish study of pixel-template improvements on b-tagging! evaluate more sophisticated track-based and secondary vertexing tags! port new secondary vertexing (SR) and topological vertexing (PM) techniques from CDF to CMS - FPix-material budget verification (B.Barnett): use photon conversions to map material in tracking volume - tau-tau HLT (S.Rappoccio):! implement template improvements in level.5 (pixel triplet tracks)! test and commission all improvements (isolation + templates) - Physics Analysis Tools (PM, SR):! expect to switch from development to maintenance by summer 8! monitoring of skimming operations for all data analysis
Tracker Alignment (AG, NT, ZG, new postdoc): activities for 8 are well-defined - Complete analysis of TIF alignment data - Complete analysis of pixel survey data - CSA 8: complete workflow checkout before real data - Cosmic alignment in LHC P5: use cosmic rays after installation at point 5, summer 8 - Alignment with collision data: first full alignment with complete system, fall 8? Frontier Database System (B. Blumenfeld): - Constant consultation with Tier-N sites as they expand - Complete configuration of Tier- sites around the world - Expand deployment to Tier- sites - Continue work to improve efficiency of system - Test, tune and possibly reconfigure the system after acquiring many months worth of real data - Implement critical service for High Level Trigger
Physics Currently have two separate efforts: Angular Analysis of new states decaying into leptons: Gritsan and Tran have studied the angular analysis of new states (Higgs, KK Gravitons) decaying into ZZ and then to 4 leptons. 7 6 5 Count 4 1-4 - 4 Significance - use this analysis to study tracker alignment effects on real analyses - build framework for angular analyses of many possible final states 4
Third Generation Window on New Physics (DF, GG, PM, SR, MS): search for new physics with t, b, # final states - start with t-tbar final states at low luminosity! develop machinery for exotica search! use as calibration of efficiencies, mistag rates, etc (SR) Fractional Error..18.16.14.1.1.8.6-1 Uncertainty Vs Background Shape Uncertainty, F =.4,, Lum = 1 fb BG Cross Section b-efficiency c-efficiency p-efficiency.4..5.1.15..5..5.4.45.5 Background Uncertainty (%) - beginning development of a template-based pixel cluster splitting + retracking algorithm for tagging/reconstruction of highly boosted jets 5
JHU Group is currently leading several crucial efforts:! Pixel Reconstruction, Tracking, Simulation! Tracker Alignment! Physics Analysis Tools Summary! Frontier DBA for CMS Computing Unique innovations have put CMS well ahead of Atlas! better pixel resolution => better track resolution => better b-tagging! pixel shape information => faster tracking, better suppression of spurious hits, ability to split clusters in high-multiplicity environments! unique understanding of radiation damage => calibrate-able algorithms, fast simulation of irradiated pixel detectors Propose to continue this program and to:! restore NSF support to BB for CMS Computing (Frontier)! increase modestly student manpower to help busy faculty+postdocs 6