12 Dec

December 2009 posts

2009.12.01 BEMC 1x1, 2x1, 2x2, 3x3 clustering

Monte-Carlo setup:

  • Throwing one photon per event
  • Full y2009 STAR geometry configurations with/without LOW_EM option
  • Throw particles flat in eta (-1,1), phi (0, 2pi),
    with energy: 30GeV with flat (+/-0.5 GeV) spread
  • bfc.C options:
    trs,fss,y2009a,Idst,IAna,l0,tpcI,fcf,ftpc,Tree,logger,ITTF,Sti,MakeEvent,McEvent,
    geant,evout,IdTruth,tags,bbcSim,tofsim,emcY2,EEfs,
    GeantOut,big,-dstout,fzin,-MiniMcMk,beamLine,clearmem,eemcDB,VFPPVnoCTB
  • Use fixed (7%) sampling fraction in StEmcSimpleSimulator.cxx
    mSF[0] = 1/0.07;
    mSF[1] = 0.;
    mSF[2] = 0.;
  • Vertex z=0
  • 50K/per particle type

data base settings (same settings in bfc.C (Jan's trick) and in my MuDst reader):
dbMk->SetFlavor("sim","bemcPed");
dbMk->SetFlavor("Wbose","bemcCalib");
dbMk->SetFlavor("sim","bemcGain");
dbMk->SetFlavor("sim","bemcStatus");

Note: for BEMC ideal pedSigma set to 0, so effectively
there is no effect when I apply 3-sigma threshold above pedestal.

Figure 1: Energy sampling of various cluster in the Barrel EMC
E_reco is the total energy in the BEMC towers from mMuDstMaker->muDst()->muEmcCollection()
eta_thrown - rapidity of the thrown photon from the Geant record
Cuts: |eta| < 0.97 && |eta|>0.01 && total energy s.f. > 0.055

Figure 2: Various cluster energy ratios

 

2009.12.07 Low EM study: LOW_EM option, 100KeV cuts, and DCUTE=100KeV

Conclusions/dicsussion at the emc2 hypernew
http://www.star.bnl.gov/HyperNews-star/get/emc2/3369.html
http://www.star.bnl.gov/HyperNews-star/get/emc2/3375.html

Monte-Carlo setup:

  • Throwing one photon per event
  • Full STAR y2006h (latest EEMC, v6.1 and TPC, v04 geometries)
  • Throw particles flat in eta (1.08, 2.0), phi (0, 2pi), and energy (5-35 GeV)
  • Using A2Emaker to get reconstructed Tower/SMD energy (no EEMC SlowSimulator in chain)
  • Vertex z=0
  • ~50K/per particle type
  • Non-zero energy: 3 sigma above pedestal

GEANT EM cuts list (default values in GeV)

  • CUTGAM - cut for gammas (GEANT default = 0.001)
  • CUTELE - cut for electrons (GEANT default = 0.001)
  • CUTHAD - cut for charged hadrons (GEANT default = 0.01)
  • CUTNEU - cut for neutral hadrons (GEANT default = 0.01)
  • CUTMUO - cut for muons (GEANT default = 0.01)
  • BCUTE - cut for electron brems (GEANT default = CUTGAM)
  • BCUTM - cut for muon brems (GEANT default = CUTGAM)
  • DCUTE - cut for electron delta-rays (GEANT default = 10^4)
  • DCUTM - cut for muon delta-rays (GEANT default = 10^4)
  • LOSS - energy loss
  • STRA - energy fluctuation model
  • Birks law parameters (Tracking Parameters)
    MODEL BIRK1; RKB BIRK2; C BIRK3

Low EM cut configurations (values in GeV)

  1. NoCuts: Default STAR geometry EM cuts

    Endcap EMC setup is quite non-uniform
    (all cuts are set via "Call GSTPAR (ag_imed,'CutName', Value)":

    • Block EMGT: 30 degree megatile

      CUTGAM = 0.00001
      CUTELE = 0.00001

    • Block ESCI: active scintillator (polystyrene) layer

      CUTGAM = 0.00008
      CUTELE = 0.001
      BCUTE = 0.0001
      CUTNEU = 0.001
      CUTHAD = 0.001
      CUTMUO = 0.001
      c-- Define Birks law parameters
      BIRK1 = 1.
      BIRK2 = 0.013
      BIRK3 = 9.6E-6

    • Block ELED : lead absorber plate

      CUTGAM = 0.00008
      CUTELE = 0.001
      BCUTE = 0.0001
      CUTNEU = 0.001
      CUTHAD = 0.001
      CUTMUO = 0.001

    • Block EALP: thin aluminium plate in calorimeter cell

      CUTGAM = 0.00001
      CUTELE = 0.00001
      LOSS = 1.
      STRA = 1.

    • Block EHMS: defines the triangular SMD strips

      CUTGAM = 0.00008
      CUTELE = 0.001
      BCUTE = 0.0001
      c-- Define Birks law parameters
      BIRK1 = 1.
      BIRK2 = 0.0130
      BIRK3 = 9.6E-6

  2. 100KeV: All cuts are set to 100KeV

    CUTGAM = 0.0001
    CUTELE = 0.0001
    BCUTE = 0.0001
    BCUTM = 0.0001
    DCUTE = 0.0001
    DCUTM = 0.0001

  3. DCUTE: All cuts are set to 10KeV, except electron delta-rays (DCUTE = 100KeV)

    CUTGAM = 0.00001
    CUTELE = 0.00001
    BCUTE = 0.00001
    BCUTM = 0.00001
    DCUTE = 0.0001
    DCUTM = 0.00001

  4. LOW_EM: All cuts are set to 10KeV

    CUTGAM = 0.00001
    CUTELE = 0.00001
    BCUTE = 0.00001
    BCUTM = 0.00001
    DCUTE = 0.00001
    DCUTM = 0.00001

Figure 1: Endcap EMC sampling fraction for different cluster sizes:
1x1, 2x1, 3x3, and total energy in the EEMC
Lower right plot shows total s.f. vs. photon thrown energy

Figure 2: Endcap EMC shower shapes

Figure 3: Endcap EMC shower shape ratios

2009.12.08 Low EM timing study: 10KeV vs. 100KeV cut settings

Conclusions/dicsussion at the emc2 hypernew:
http://www.star.bnl.gov/HyperNews-star/get/emc2/3374.html

List of LOW_EM cuts and defaults

Low EM cut configurations (values in GeV)

  1. NoCuts: Default STAR geometry EM cuts
  2. LOW_EM:100KeV: All LOW_EM cuts are set to 100KeV
  3. LOW_EM:10KeV: (default) LOW_EM cuts (10KeV)
  4. DCUTE: All cuts are set to 10KeV, except for electron delta-rays DCUTE = 100KeV

QCD hard processes timing

Pythia QCD Monte-Carlo:

  • Pythia pp@500GeV 2->2 hard QCD processes for parton pt>15GeV
  • Full STAR y2009a (latest EEMC, v6.1 and TPC, v04 geometries)
  • 50 events per file, 100 jobs
  • BFC options and kumac details are here

Figure 1: QCD Total (GEANT/GSTAR+bfc) timing (seconds/event)

Figure 2: QCD GEANT/GSTAR timing (seconds/event)

Figure 3: QCD bfc.C timing (seconds/event)

EEMC single photon timing

EEMC single photons Monte-Carlo

  • One photon per event
  • Full STAR y2006h (latest EEMC, v6.1 and TPC, v04 geometries)
  • flat in eta (1.08, 2.0), phi (0, 2pi), and energy (5-35 GeV)
  • 250 events per file, 200 jobs

Figure 4: EEMC single photon Total (GEANT/GSTAR+bfc) timing (seconds/event)

Figure 5: EEMC single photon GEANT/GSTAR timing (seconds/event)

Figure 6: EEMC single photon bfc.C timing (seconds/event)

2009.12.17 Ecalgeo-v6.2: embedded LOW_EM cuts in the calorimeter geometry

Monte-Carlo setup:

  • Throwing one photon per event
  • Full STAR y2006h (latest EEMC-v6.2 and BEMC with LOW_EM cuts, rest of geometry from CVS)
  • Throw particles flat in eta (1.08, 2.0), phi (0, 2pi), and energy (5-35 GeV)
  • Using A2Emaker to get reconstructed Tower/SMD energy (no EEMC SlowSimulator in chain)
  • Vertex z=0
  • ~50K/per particle type
  • Non-zero energy: 3 sigma above pedestal

Figure 1: (left) Endcap EMC sampling fraction (total calorimeter energy), (right) SMD-u sampling fraction
Red: (previous) ecalgeo-v6.1 with global LOW_EM option
(Note: same points as in this post, Fig. 1 lower left, label y6:LOW_EM)
Black: (new) ecalgeo-v6.2 (embedded LOW_EM cuts), no global LOW_EM option

Figure 2: Pre-shower migrations
There is only a few events with pre1>4MeV with new simulations: potential problem with TPC geometry?

2009.12.20 Ecalgeo-v6.2: embedded LOW_EM cuts after TPC/EEMC overlap fix

Monte-Carlo setup:

  • Throwing one photon per event
  • Full STAR y2006h (latest EEMC-v6.2 and BEMC with LOW_EM cuts, rest of geometry from CVS)
  • Throw particles flat in eta (1.08, 2.0), phi (0, 2pi), and energy (5-35 GeV)
  • Using A2Emaker to get reconstructed Tower/SMD energy (no EEMC SlowSimulator in chain)
  • Vertex z=0
  • ~50K/per particle type
  • Non-zero energy: 3 sigma above pedestal

Results: Update for the previous tests of EMC v6.2 geometry after fixing TPC/EEMC overlap

Figure 1: Endcap EMC sampling fraction: total calorimeter energy, pre1-, pre2-, post- shower layers, and SMD-u energy
Red: (previous) ecalgeo-v6.1 with global LOW_EM option
(Note: same points as in this post, Fig. 1 lower left, label y6:LOW_EM)
Black: (new) ecalgeo-v6.2 (embedded LOW_EM cuts), no global LOW_EM option

Figure 2: Pre-shower migrations
Change in TPC geometry seems to introduce a reasonable (small) change in pre-shower migration

Photon-jet simulation request

Simulation needs with y2006/y2009 geometry
specific to the photon-jet analysis

* Update version of the previous simulation
request from December 18, 2008 (see Ref. [1])

Understanding effects of trigger, material budget differences,
and throughout comparison between 2006 and 2009 data
requires to have dedicated Monte-Carlo
data samples with both y2006 and y2009 geometries.

Requested samples

We request to produce the following set of
Monte-Carlo samples for the photon-jet analysis:

  • S1: 1st priority

    Dedicated (gamma filtered, Refs. [2-5]) data sample
    for Pythia pp@200GeV prompt photon processes
    with y2006 STAR geometry configuration
    and partonic pt range 2-25GeV.

    Simulations configured with:

    • LOW_EM option in starsim (Ref. [6]).
      Low cuts on electromagnetic processes in GSTAR

    • y2006h geometry tag, which includes
      latest Endcap EMC (v6.1) and TPC (v4) geometry fixes.

    • Pythia 6.4 CDF Tune A or Perugia tunes (6.4.22)?

  • S2: 1st priority

    Dedicated (gamma filtered, Refs. [2-5]) data sample
    for Pythia pp@200GeV hard QCD processes
    with y2006 STAR geometry configuration
    and partonic pt range 2-25GeV.

    Same simulation setup as for the sample S1.

  • S3: 2nd priority

    Pythia pp@200GeV prompt photon and hard QCD
    processes with y2009 STAR geometry configuration
    and partonic pt range 2-25GeV.

    Same simulation setup as for the sample S1
    but with y2009a geometry tag.

  • S4: 3rd priority

    Pythia pp@500GeV prompt photon and hard QCD
    processes with y2009 STAR geometry configuration
    and partonic pt range 2-25GeV.

    Same simulation setup as for the sample S1
    but with y2009a geometry tag.

Event number, CPU time, and disk space estimates

Below I provide some estimates of CPU and disk space
which are required to produce the data samples listed above.
These estimates are based on the previous (private)
production of the MC gamma-filtered events with y2006
geometry which was done at MIT computer cluster
by Michael Betancourt (Ref. [2,4-5]):

  • E1 (prompt photons)

    Pythia pp@200GeV prompt photon simulations
    with ~7 pb^-1 luminosity:

    • ~60 days running time on a single CPU

    • ~17Gb of disk space to store MuDst/geant files

    • Number of (filtered) events:
      ~ 30K for pt range 6-9GeV
      ~ 15K for pt range 9-15GeV

  • E2 (QCD hard process)

    Pythia pp@200 QCD hard process simulations
    with (at least) 1 pb^-1 luminosity:

    • ~ 620 days running on a single CPU
      (less than a week on a cluster with 100 CPUs)

    • ~ 150Gb of disk space to store MuDst/geant files

    • Number of (filtered) events:
      ~ 650K for pt range 6-9GeV
      ~ 300K for pt range 9-15GeV

Notes on the estimates:

  • N1

    Enabling LOW_EM option in GSTAR increases
    the time estimates by ~40% (Ref. [7]).

  • N2

    Additional production of jet trees will
    require a disk space on the order of < 2%
    of the total size of the MuDst/geant files.

  • N3

    Additional production of gamma trees will also
    require a disk space on the order of a few percents
    of the total size of the MuDst/geant files.

References

  1. Previous simulation request (Date: 2008, Dec 18):
    http://www.star.bnl.gov/HyperNews-star/protected/get/starspin/3596.html

  2. Michael's document on
    "Targeted MC procedure for the gamma-jet program at STAR":
    http://drupal.star.bnl.gov/STAR/system/files/20080729_gammaFilter_by_MichaelBetancourt.pdf

  3. simulations with filtering readiness:
    http://www.star.bnl.gov/HyperNews-star/protected/get/starsimu/387/1/1/2/1/1/2/3/1.html

  4. Filtered photon production with y2006 geometry:
    http://www.star.bnl.gov/HyperNews-star/protected/get/phana/256.html

  5. More details on statistics needed and disk space estimates:
    http://www.star.bnl.gov/HyperNews-star/protected/get/phana/297.html

  6. LOW_EM option in GSTAR:
    http://www.star.bnl.gov/HyperNews-star/protected/get/phana/371.html

  7. Time estimates with and without LOW_EM option: