Simulation Tasks Status

1
Simulation TasksStatus

• Qizhong Li
• Fermilab
• Sept. 14, 2001
• D0 Collaboration Meeting

2
Outline

• P10 production release
• D0gstar
• D0Sim
• D0Trigsim
• Plans for p11 production release
• D0gstar/D0Sim/D0TrigSim
• D0Raw2Sim
• Pmcs

3
D0gstar

• D0gstar -- full GEANT simulation (lead by Shuichi
  Kunori)
• In p09 and p10 production releases
• All detectors,
• including LUMI and FPD.
• New 3D magnetic field map.
• New interface to access to the field
  configuration (get map version, get solenoid and
  toroid polarity).
• Default ECUTs for calorimeter plate level
  geometry were lowered to get more accurate
  response.

4
D0gstar(cont.)

• A new option to swim very forward p/pbar to FPD
  in double precision was added. This option is OFF
  as default.
• New access methods on the option of FPD geometry
  and tracking in event information.
• Changed from GEANT 3.21.12 to GEANT 3.21.13
  (Geant bug fix).

5
D0sim

• D0Sim (lead by Serban Qizhong), uses
  D0gstar output to produce simulated raw data
  events.
• It performs following functions
• Merge hard scatter and minbias events
• Add calorimeter pileup from previous events
• Make L1CalTTowerChunk for L1 simulation
• Add calorimeter noise
• Add SMT noise and inefficiencies
• Add CFT noise and inefficiencies
• Add Muon noise and inefficiencies
• Make a RawDataChunk
• Drop all D0GStar chunks

6
D0sim production releases

• P08.12.00
• pileup CalWeight cps, fps, cft, smt
  digitization muon raw data CalDataChunk
  RawDataChunk L1CalTTChunk
• RawDataChunk calmuosmtcpscftfps
• CalDataChunk
• uses calpreshower optimized weights
• Only FPD and L0 still missing.

7
D0sim releases (cont.)

• P09.10.00
• Turned on smt noise simulation
• use cal only optimized weights in L1CalTTChunk.
• Handle both mixture and plate geometry.
• P10.03.00
• CalDataChunk uses cal only optimized weights
• added code to handle pseudo SimChunks.

8
CPU Time

• D0gstar
• (sec/event on d0mino)
• Plate uses a factor of 3 more CPU than
  mixture.

9
CPU Time

• D0sim
• (sec/event on d0mino)
• 2.5mb overlap uses a factor of 2-3 more CPU than
  no minbias.
• Plate uses 15-30 more CPU than mixture.

10
Production release certification procedure

• Generate MC events using D0gstar/D0sim
• Test RawDataChunk by running each subdetector
  examine in offline mode.
• Run Sim_exam, compare histograms.

11

• SMT examine plots
• (from Elizaveta)

12
ttbar 2.5 mb mixure vs plate in calorimeter
ltPt in cellsgt in the different layers of calo
plate mix EM1 0.019 0.018 EM2 0.014 0.015
EM3 0.013 0.013 EM4 0.020 0.021 Fine
hadr 0.025 0.025 Coarse hadr 0.031 0.031
13
pT in cells (ttbar 2.5mb plate)
14
pT in cells (ttbar 2.5mb mixture)
15
ttbar 2.5 mb mixture vs plate in calorimeter
ltEnergy in layersgt in the different layers of
calorimeter (GeV)
plate mix EM1 60.0
49.6 ß 10 GeV difference EM2 25.9
28.1 EM3 62.3 60.0 EM4 46.8
46.2 Fine hadr na na
(overflow) Coarse hadr 19.0 24.4
16
Etot in layer (ttbar 2.5mb plate)
17
Etot in layer (ttbar 2.5mb mixture)
18
ttbar 2.5 mb mixture vs plate in calorimeter
lt of hit cellsgt in the different layers of
calorimeter
plate mix EM1 2475
2114 ß 15 more cells
in plate EM2 2259
2297 EM3 na na
(overflow) EM4 2265 2363 Fine hadr
5900 5416 Coarse hadr 1296 1387
19
D0TrigSim

• The trigger simulation program, D0TrigSim,
  includes L1/L2 and L3 simulation.
• This effort is lead by Serban and Dugan Oneil.
• The main cvs package is d0trigsim. L1/L2 default
  config is in tsim_l1l2. L3 default config is in
  tsim_l3.
• A new script runD0TrigSim is a trigsim wapper to
  be used to run the executable. Features include
• setup up all necessary config files
• allows use of xml trigger list

20
D0TrigSim and the Trigger Database

• The design has always been that users grab a
  trigger list from the trigger database (or make a
  new one), hand that list to trigsim and wait for
  the results.
• How close are we?
• The trigger database is now in production. A 20
  trigger list has been written for MC use.
• The new script will allow users to take any xml
  file generated from the trigger database, and
  send it through coorsim.
• Coorsim creates configuration files which can be
  read by d0trigsim.
• D0trigsim runs, creates an ntuple.
• The chain is now in place! Need more testing, but
  great progress has been made.

21
Release status

• P06.00.01 - The 1st production release... too
  old.
• P08.12.00 - Stable version. It has run on
  Nx100K events without crashing. getting old.
• The above versions didnt use cal trigger tower
  weights to reconstruct energy.
• P09.08.00 - not stable, but use cal weights
  optimized for plate geometry.
• P10.03.00 - The most recent release made this
  week. More functionality, need test.
• For example,
• Single e
• (PT 20 GeV)

Mean RMS
p08 17.1 1.6
P09-mix 18.8 2.5
P09-plate 19.1 2.0
22

• Single e pT plot from p10 mix plate compared
  with old release.

23
D0TrigSim in p10

• P10 contains a lot of big changes for d0trigsim
  infrastructure. For example L1/L2 changes to run
  on real data
• l1l2unpacker learns VRB format.
• more analyze packages read directly from RDC
  (l1cal, l1muon, l2caljet, l2calmet, l2calem,
  l2cps, l2gbl).
• tsim_l1ft completely rewrites L3 output classes
  to better emulate real data. New unpacker to be
  written for l1ft analyze package and examine.
• More functionality added to l2gbl (track match),
  L3 filter/tools, L2errorlogger integration, etc.

24
D0TrigSim P10 certification

• Run d0trigsim on p09.08.00 MC reco certification
  samples
• process of events
• ttbar 1k
• WH tn 1k
• single e 15k
• single m 10k
• QCD (pTgt20) 2k
• Z bb 200
• Memory use and CPU time
• ttbar
  single m Memory max 125Mb 115Mb
• L1/L2 time 2.1s/evt 0.6s/evt
  L3 time 23.7s/evt 7.4s/evt
• Total CPU 27.2s/evt 8.3s/evt

25

• P10 plot for the number of L2 trigger bits passed

26

• P10 plot for the fired L2 trigger bit
• (L2 plots are from Oneil)

27
Plan for p11production release

• P11 production release schedule
• Oct 15, 2001 All major improvements to D0gstar,
  D0SIM, TrigSim and L3 tools MUST be released.
• Oct 29, 2001 All major improvements to D0reco
  MUST be released.
• Nov 26, 2001 Production version is created, and
  formal certification begins.
• Dec. 21, 2001 Certification finishes.

28
Plan for p11(cont.)

• D0gstar
• Tune parameters for GEANT with real data.
• Adjust ECUT values after studying the low energy
  particle effects.
• Remove bugs.
• D0Sim
• get more realistic minbias overlap and noise
  simulation by using real minbias data overlapping
  on MC events. Use new program D0Raw2Sim

29
Plan for p11 (cont.)

• D0Raw2Sim
• Intended to produce minbias files from real data
  for merging with MC events in D0Sim.
• Necessary steps
• 1) Unpack RawDataChunk
• 2) Convert unpacked chunk to digi chunks
• 3) convert digi chunks to pseudo SimXXXHit chunks
• 4) Make L1CalTTChunk
• Step 1) and 2) handled by d0reco packages.
• All packages for step 3) are ready and released.
• For calorimeter need to make
• MC CalDataChunk data CalDataChunk
• RawDataChunk L1CalTTChunk

30
Plan for p11 (cont.)

• D0TrigSim
• all L1 packages pack/unpack like real data
• all L1 packages work in "pass-through" mode
• L1FPS joins simulation
• all L2 analyze packages work directly on real
  data
• L2STT joins simulation
• XML files from trigger database
• d0_analyze works fully (integrates mc_analyze,
  trigsim_analyze and reco_analyze outputs into one
  root tree)

31
Plan for p11 (cont.)

• Pmcs -- Fast Monte Carlo (Sarah Eno Elemer
  Nagy)
• The general framework of PMCS is in place and the
  current work focuses on providing the smearing
  functions for the various physics objects.
• Major effort now to put pmcs output into d0phyobj
  chunk, and to simulate the variables in chunk
  properly.
• A first version of PMCS for general use is
  expected for p11.

32
Summary

• P10 D0gstar/D0Sim/D0TrigSim are ready to be used
  to generate big sets of Monte Carlo events for
  physics use.
• A lot of progress and improvements.
• Working hard on preparing p11 production release.
• Pmcs will be ready for physics use soon.
• Our goal is to make the Monte Carlo simulation as
  close to real data as possible.