HIP CMS PROGRAMME

1
HIP CMS PROGRAMME

• HIGHLIGHTS 2007-2008
• PLANS 2008-
• FUNDING

2
HIP CMS PROGRAMME

• CMS PHYSICS ANALYSIS PROJECT
• Project Leader V.Karimäki
• CMS TRACKER PROJECT
• Project Leader Eija Tuominen

3
1. PHYSICS ANALYSIS PROJECT

• Physics simulation, study of the discovery
  potential of CMS, preparation for event
  reconstruction and physics analysis in CMS
• CMS computing at HIP
• CMS user support coordination
• Software alignment of the CMS tracker
• Test beam data analysis
• GEANT4 simulation tools

4
Project personnel

• Veikko Karimäki, PhD Project Leader Helsinki
• Jorma Tuominiemi, Prof. Programme
  Director CERN/Hki
• Ritva Kinnunen, PhD Senior Scientist Helsinki
• Kati Lassila-Perini, PhD Senior Scientist CERN
• Sami Lehti, PhD Senior Scientist Helsinki
• Tomas Lindén, PhD Senior Scientist Helsinki
• Tapio Lampén, PhD PostDoc Scientist Helsinki
• Mikko Voutilainen, PhD PostDoc Scientist CERN
• Aatos Heikkinen, MSc PhD Student Helsinki
• Matti Kortelainen, MSc(Tech) PhD
  Student Helsinki
• Lauri Wendland, PhD Student Helsinki
• Pekka Kaitaniemi PhD Student Saclay

5
R. Kinnunen, M. Kortelainen S. Lehti, L.Wendland

• HIP responsability Analysis of the H -gt ??, ?
  -gt hadrons? in fully hadronic channel from gg -gt
  tbH
• Trigger development and tests
• ? identification for one- and three prong ts
• Veto on leptons and hadronic ts from associated
  W
• b-tagging, top and W mass reconstruction
• One of the most difficult channels to observe,
  and maybe the only possibility to discover the
  charged Higgs boson
• 2. Tau energy corrections
• With tracks and calorimeter information
• Energy resolution excellently improved!

6

• 3. Preparation for real data studies for
  background measurements from data for 0.1 fb-1
• Measurement of tt and W3/4jet backgrounds from
  data with W-gt?? decay modes
• Method developed to measure the background due to
  MET mis-measurement from data (isolated muon with
  pTgt100 GeV/c selected instead of t jet)
• Separation of events to tt and W3/4jet samples
  with double b-tagging, top and W mass cuts (for
  tt events) and b-jet veto (for W3/4jet events)
• Measurement of QCD background from data
• Method developed to measure the QCD background
  due to MET mis-measurement from data in the
  hadronic multi-jet events exploiting the
  determination of fake t probability in the gjet
  or Zjet, Z-gtmm events

7

• 4. Continuation of the H/A-gt??-gt 2 jets analysis
  for a PhD work (LW)
• 5. Higgs boson searches in SUSY cascades
  withnon-universal gaugino masses
• Joint project with HIPphenomenologists

SUSY reach with non-universal gaugino masses and
H/A/h -gt bb decays in CMS
8
Energy correction for t jets with tracks

• Developped for the collimated energetic one-prong
  t jets (ETgt100 GeV) from heavy
• charged Higgs boson in gg -gt tbH, H -gt ??, ? -gt
  hadrons?
• Principle of the method
• Separation of the t -gt p n and t -gt p np0
  n decay modes and separation of
• the interacting charged pions testing matching
  between the track(s)/CALO cluster
• and track(s)/HCAL cluster
• Replacing the CALO jet with the track(s) for
  non-interacting charged pions and adding
• the ECAL cluster for t -gt p np0 n
• Keeping the CALO jet for the interacting charged
  pions

Energy resolution, comparison with CALO t jet and
Particle Flow t jet
9
Identification of t jetfor H -gt ??, ? -gt
hadrons?
Event rate before and after t identification for
1- and 3-prong final states. Suppression of 105
achieved against QCD background!
QCD background
Filtered event rate (fb)
Signal
10
Backup slides
11
Identification of t jet for H -gt ??, ? -gt
hadrons?

• Kinematical cuts ETjet gt 100 GeV, hjetlt2
• Tight tracker isolation track pT cut0.5 GeV,
• signal cone 0.04, isolation cone 0.45
• Electromagnetic isolation
• 1-prong selection
• - one track in the signal cone
• - pldg.trk./Ejet gt 0.8
• - neutral hadron and electron rejection
• with HCAL/track matching
• 3-prong t-gt 3p n selection
• - 3 tracks in the signal cone
• - suppression of t-gt3p n np0
  decays
• with CALO cluster/tracks matching
• - p(p p ) / Et-jet gt 0.75
• - cut on t invariant mass
• - cut on t flight path significance

1-prong t jets
Distribution , pldg.trk./Ejet sensitive to t
polarization
3-prong t jets

Distribution of t invariant mass
12
Measurement of tt and W3/4jet backgrounds from
data with W-gt?? decay modes
Signal for gg -gt tbH, H -gt ??, ? -gt hadrons?
is visible in mT(t jet,MET), provided the
backgrounds from tt and W3/4jet are limited to
mT(t jet,MET) ?100 GeV
Method to measure the background due to MET
mis-measurement from data with muonic multi-jet
events

• Selection of events with one isolated muon with
  pTm gt 100 GeV, instead the t jet
• Other selections METgt100 GeV, veto on leptons
  and associated hadronic W-gttn decays
• Separation of events to tt and W3/4jet samples
  with double b-tagging, top and
• W mass cuts (for tt events) and b-jet veto
  (for W3/4jet events)

mT(m,MET) for the selected tt sample events for
mT(m,MET) ?100 GeV mainly from recidual
associated hadronic W-gttn decays
Top and W mass reconstruction with kinematic fit
with respect to jet energies Ei , one jet b
tagged
13
Measurement of QCD background from data

QCD multi-jet events can lead to background in
the signal area mT(t jet,MET) ? 100 GeV through
MET mis-measurement and through hadronic jets
identified as t jets
The background due to MET mis-measurement can be
measured from data in the hadronic multi-jet
events, measuring the fake t probability in the
gjet or Zjet, Z-gtmm events
mT(t candidate,MET)

• Method
• Selection of hadronic multi-jet events
• with at least one jet with pTjet gt 100 GeV
• Assignment of one jet randomly as t candidate
• large MET gt 100 GeV
• b tagging, top and W mass reconstruction
• Propagation to signal selection multiplying
• with the fake t probability

Events found in the signal area, more efficient
MET correction methods under development
14
T.Lindén

• CMS Software and services
• Resource usage
• CSA08
• Plans
• Summary

15
Service Status CMSSW OK local manual
installation https//twiki.cern.ch/twiki/bin/vi
ew/CMS/CMSSW_aptinstaller dCache OK
performance testing, xrootd versus gsidcap
http//www.dcache.org/ PhEDEx OK Physics
Experiment Data Exchange http//cmsdoc.cern.ch/
cms/aprom/phedex/ Frontier OK Open source
squid web cache http//frontier.cern.ch/squidst
ats/mrtgcms/hip/proxy-hit.html WLCG SAM OK WLCG
Site Availabilty Monitoring CMS SAM SRM OK CMS
SRM Site Availabilty Monitoring https//twiki.c
ern.ch/twiki/bin/view/CMS/Dashboard ProdAgent OK
Monte Carlo Production ARC plugin is production
ready https//twiki.cern.ch/twiki/bin/view/CMS/
ProdAgent CRAB in use CMS Remote Analysis
Builder http//cmsdoc.cern.ch/cms/ccs/wm/www/Crab/
in use gLite WMS to ARC submission, working
to fix scalability in use glideinWMS to ARC
submission in progress direct CRAB to ARC
submission JobRobot pending for site testing,
needs VOMS roles CMS SAM CE pending CMS CE Site
Availabilty Monitoring, needs VOMS roles CMSSW
pending grid software installation jobs, needs
VOMS roles WLCG accounting preparing Extract CMS
SGAS information, inject into APEL (CSC)
16
The CMS Tier-1 and Tier-2 sites.
17
2. Resource usage

• Most of the physics analysis presented here was
  done using HIP Tier-2 resources
• The CMS disk usage at CSC is about 70 of the
  available CMS quota of 110 TB
• Data samples have been transferred for local
  analysis use and for PhEDEx LoadTest bandwidth
  monitoring use
• The summed CMS CPU wall time usage during April,
  May and June 2008 was 58800 CPU hours

18
3. Combined Computing Readiness Challenge

• The Common Computing Readiness Challenge 2008
  (CCRC) or CMS CSA08 was a Data Challenge
  simultaneous to all LHC experiments to test the
  grid infrastructure before datataking at the LHC.
• Phase 1 of CCRC was in February and Phase 2 of
  CCRC was during May.
• CSA08 Phase 1
• Data transferred using PhEDEx (Physics
  Experiment Data Export) to Finland
• Data stored in dCache (siberie at HIP and
  madhatter at CSC)
• HIP Transfer targets were met

19
3. Combined Computing Readiness Challenge

• CSA08 Phase 2
• Needed CMSSW versions installed and visible in
  BDII
• Needed datasample(s) transferred with PhEDEx
• HIP participated in the CMS CSA08 Analysis
  comissioning activities
• CRAB jobs were submitted from CERN as part of the
  chaotic submission exercise
• CRAB jobs were submitted to HIP from the US using
  the glideinWMS, see http//tinyurl.com/58fugq
• During CSA08 CMS could run up to 200000 jobs/day
  in total
• Tier-1 to HIP Tier-2 transfer tests were repeated
  with met targets.
• In total CMS transferred 3,6 PB of data in May
  with a total maximum rate of 1,7 GB/s out of CERN

20
3. Combined Computing Readiness Challenge
CRAB glideinWMS submitted jobs from the 30th of
May until the 2nd of June. Sepeli is in the
middle of the list

21
4. Plans

• The Nordic Optical Private Network is being set
  up (ping works to Denmark)
• A new Linux cluster shared between Physics,
  Chemistry and HIP is being purchased to Kumpula
• Plan 2009 aquisitions with HIP Technology
  programme and CSC
• Study dCache disk performance scalability
• Compare CRAB submission methods (CERN WMS,
  glideinWMS, native ARC plugin)
• Study ARC performance with disk caching enabled
• Study PROOF on the clusters for parallell data
  analysis
• Study CRAB ARC to gLite submission possibility
• Work on obtaining funding from 2011 onwards

22
5. Summary

• HIP contributed to CSA08 activities with ARC
  resources
• CRAB submission works with new CERN WMS and
  glideinWMS
• Jobsubmission scaling problem with new CERN WMS
  is being worked on
• VOMS roles will be implemented to enable
  remaining services
• WLCG Accounting will be implemented
• HIP is ready for LHC startup

23
Aknowledgements

• Antti Pirinen, HIP, project leader
• Dan Still, CSC, CSC project leader
• Tomas Lindén, HIP, grid coordinator
• Jukka Klem, HIP/CERN, PhEDEx, Frontier
• Jonas Dahlbom, HIP, dCache support
• Chris Hanke, CSC, dCache support
• Arto Teräs, CSC, dCache support
• Vera Hansper, NDGF/CSC, Finnish Node coordinator
• Erik Edelmann, NDGF/CSC, ProdAgent ARC plugin
• DongJo Kim, JYFL, ALICE computing
• Mikko Närjänen, HIP, Alice support
• Jesper Koivumäki, HIP, CRAB ARC port
• Kalle Happonen, HIP/CERN, CSC, middleware support
• Pekko Metsä, HY, cluster administration in Kumpula

24
CMS User Support by HIP
Coordinated by Kati Lassila-Perini

• Documentation
• CMSSW WorkBook
• Reference Manual
• CMS Offline Guide
• Help desk
• Tutorials
• Savannah portal for user questions
• e-mail contact
• hypernews forums

25
Track based alignment for CMS Tracker modules
T.Lampén, V.Karimäki

• Related task update of misalignment simulation
  scenarios of CMS. An essential tool for realistic
  physics analyses for early data-taking.
• Application activities of the H.I.P. alignment
  algorithm lately
• Cosmic muons from TIF
• CSA08 exercise, application of all three
  alignment algorithms
• Many CMS groups actively using the H.I.P.
  alignment algorithm
• Johns Hopkins (3 people, CSA08 CRUZET tests)
• Fermilab (J.Pivarski, muon chambers alignment)
• INFN (R.Covarelli et al., alignment at TIF,
  cosmics)
• Strong cooperation between different groups!
• Other two algorithms used in CMS are Millepede
  and Kalman Filter algorithm. The H.I.P. algorithm
  is computationally simple, and it has the most
  robust approach. It has so far been used in most
  of the alignment studies in CMS, although
  Millepede, developed at Hamburg, has lately
  become at least equally popular.

26
CSA08, CRUZET, CRAFT (future) exercises
T.Lampén, V.Karimäki

• Dress Rehearsals of all three alignment
  algorithms and whole procedure of their
  application to real data
• Learning experience for startup of CMS
• Matters of practical use now topical (track hit
  selection, constraining unphysical deformations,
  validation process etc.)
• Also alignment-related issues taking shape
  (triggers, data flow, validation etc.)
• HIP participation in CSA08, CRAFT foreseen

TOB residuals
TIB residuals
RMS 0.041 0.036 0.039
RMS 0.099 0.056 0.052
Design geometry HIP Millepede
27
Near future work

• Studies/application of HIP algorithm (and
  comparisons with other algorithms) with
• real data from Cosmic Rack setup (CERN and
  Helsinki C-Racks)
• test beam studies (SiBT)
• other real (cosmic) data?
• Refinement of CSA08 alignment exercise
• Alignment tasks for CMS Tracker

28
Test beam data analysis

• Alignment and full offline data analysis of SiBT
  2007 data
• SiBT 2008 data taking took place in July, data
  analysis has just begun, in cooperation with the
  Tracker project
• Provided software for data taking of 2008 and
  monitoring plots with a quick 5-minute response
  time (essential ingredient of success for the
  4-day test period, during which CID, MCz and
  3D-detectors were tested)

M.Kortelainen, T.Lampén
29
CSA08, CRUZET, CRAFT (future) exercises
T.Lampén, V.Karimäki

• Dress Rehearsals of all three alignment
  algorithms and whole procedure of their
  application to real data
• Learning experience for startup of CMS
• Matters of practical use now topical (track hit
  selection, constraining unphysical deformations,
  validation process etc.)
• Also alignment-related issues taking shape
  (triggers, data flow, validation etc.)
• HIP participation in CSA08, CRAFT foreseen

30
Geant4 activities (I)

• Geant4 Bertini intra-nuclear cascade models
  developed by HIP group are now
• widely accepted when choosing optimal
  physics models for LHC simulations.
• Use of Bertini model largely solve the problem
  seen in shower shape year 2007.
• Models developed play a major role in precision
  studies of LHC experiments. Typically Bertini
  cascade is used in combination of high energy
  model QGSP in physics list QGS BERT
• Latest improvements include optimization of speed
  in the CMS production runs, and more detailed
  Coulomb scattering model.
• Our emphasis has been in developing new Geant4
  models based on INCL cascade and ABLA
  fission-evaporation codes.
• First release of INCL 4.2 and ABLA V2 was made in
  Dec'07 release of Geant4 9.1.
• Currently we are working in collaboration with
  Commissariat à lÉnergie Atomique (CEA), Saclay,
  to include INCL5 extensions, such as Carbon
  projectiles into Geant4.
• A.Heikkinen and P.Kaitaniemi with G.Folger et
  al. Progress in Hadrornic Physics Modeling in
  Geant4 (Submitted to Journal of Physics, July
  2008)

31
Geant 4 activities (II)
CMS HCAL response (linearity) to protons. Test
beam data from year 2006 brass (50mm)
scintillator (3.7mm) sampling is compared with
Geant4 QGSP_BERT simulation.
HCAL resolution (Courtesy of S. Kunori)
32
Geant 4 activities (III)
Unprecedented accuracy made available in Geant4
using INCL and ABLA models 1.2 GeV proton on Al
and Zr targets is compared with data from SATURNE
experiment.