relavant twiki



W helicity measurement plan, aiming at 2010 and 2011 data

Man power and responsibilities

  • Liming Chen:
    • Lepton selection: implement the lepton selection in code, and update it.
    • KLFitter reco method,
    • Systematics from PDF, top mass,
    • Data skiming
  • Bo Li:
    • DD Wjets background estimation
    • Systematics from b-tagging efficiency
    • Electrion isEM efficiencies estimation from data
  • Lianyou Shan:
    • Jets selection: implement the lepton selection in code, and update it.
    • Final parameters extraction
    • Correction function for each helicity state
    • Systematics from pileup, ISR/FSR, JES,
  • Chengguang Zhu:
    • Etmiss selection: implement the lepton selection in code, and update it.
    • Etmiss performance: connection with Etmiss liaison gruop
    • Background shape estimations
    • Systematics from shape of background met scale
  • Wang Jin:
    • DD QCD estimation,

time scale

  • conference ahead:
    • June 6: Physics at LHC,
    • July 22: EPS 2011
    • Aug 22: lepton-Photon 2011
    • Sep 25: Top2011
  • Luminosity schedule:
    • March 18 -- March 25: 11pb-1
    • April 17 -- May 4: 100pb-1
    • May 14 -- June 15: 600pb-1
  • Note aproval:
    • INT note posting to TOPWG approval 3 days
    • TOPWG approval to CIRCULATION 10 days *Includes writing of CONF note, EB approval of conf note, PhysCoord signoff
    • ATLAS circulation takes 7 days
      • Includes EB final signoff and final reading

  • Proposed EPS schedule working backwards
    • Thu July 21- Conference starts
    • Mon July 18 Final reader approval
    • Tue July 12 ATLAS approval meeting
    • Mon July 4 CONF note circulation
    • Tue June 28 top-wg approval on full data
    • Mon June 27 INT/CONF on full data posted
    • Wed June 22 full dataset available for analysis
    • Wed June 15 End of LHC run
    • Tue June 14 top-wg analysis pre-approval
    • Fri June 10 - INT note posted
    • Fri June 3 Cut dataset for preapproval
    • Thu May 19 Signoff on TopPhys production cache
    • Sat May 14 Start of LHC run
    • Thu May 12 Finalization of object definitions

  • Aiming at top2011, inverse time direction:
    • Sep 25: Conference starts
    • Sep 22: Final reader approval
    • Sep 16: ATLAS approval meeting
    • Sep 8 : CONF note circulation
    • Sep 2: top-wg approval on full data
    • Sep 1: INT/CONF on full data posted
    • Aug 25 : full dataset available for analysis
    • Aug 17: top-wg analysis pre-approval
    • Aug 13: INT note posted
    • Aug 6: Cut dataset for preapproval

current task

  • include KLFitter into D3PD processing code
  • data driven method to estimate QCD events number and cospsi shape
  • data driven method to estimate Wjets events number and cospsi shape
  • Maximum Likelyhood method to draw the FL-FR 2D fit
  • data slimming and skimming
  • check all needed data is there, otherwise to produce.

analysis procedure

  • Objects selection: topinputs selection, any methods to improve the selection to get better S/B
  • Reconstruction method: KLFitter and CSC method.
  • Background estimation: data driven for QCD/Wjets and MC
  • Bacground to be subtrcted from the all observables
  • Correction function from MC to correct the cospsi distribtuion. should verify the correction function for FL, FR and F0 part are same or not
  • Histogram fit or LH fit
  • Systematics Estimation

Full plan

2010 data ref page

  • 2010 data: top group central produced D3PD
    • real data: period E5 - I, 35.3 pb-1 (+/-3.2%)
    • MC:
      • signal: ref MC@NLO v3.41 [7], assumed top-quark mass of 172.5 GeV, NLO PDF set CTEQ66
      • backgroup:
        • Alpgen v2.13 implementing the exact LO matrix elements and LO PDF set CTEQ6L1, ‘MLM’ matching.
          • W+jets up to 5 partons, W+c+jets, Wbb, Wcc
          • Z+jets up to 5 partons and dilepton invariant mass mll > 40 GeV, Zbb, Zcc
          • QCD multi-jet up to 6 partons, bb, cc
          • diboson WW+jets, WZ+jets and ZZ+jets events.
        • Pythia: Z boson covering 10 GeV < m\x{2113}\x{2113} < 40 GeV.
        • MC@NLO: single-top, removing overlaps with tt events.

2011 data

  • Collision data:
    • Collected in period B and D, please use the GRL: Top_GRL_D.xml
    • The list of the datasets is datasets_2011_realdata
      • These data is slimed to 2 datasets, if it satisfied SlimCuts_leptonMoreThan0_jetsMoreThan3:
        • user.zhuchengguang.Slim_physics_Muons_201106080525/, has transfered to CERN-PROD_SCRATCHDISK,
        • user.zhuchengguang.Slim_physics_Egamma_201106080525/, except one sub dataset, has been transfered to CERN-PROD_SCRATCHDISK

  • MC data.
    • MC10a: the list of the MC10a data used is datasets_2011_MC10a.
      • these data has been slimed to user.zhuchengguang.Slim_MC_201106080525/, most of them copied to CERN-PROD_SCRATCHDISK, some of them still in the site where it was slimed.
    • MC10b: we can now switch to these data: datasets_2011_MC10b
      • MC10 vs MC10a vs MC10b: use MC10 for 2010 data analysis, use MC10a,b for 2011 data analysis.
      • it will be slimed soon.
      • If you want to use more data than you can taken from this list
      • 50ns bunch spacing in 3 trains of 36 bunches with 225ns train separation

* MC11

        • planned for release 17, used for beyond summer conference.

Background estimation

  • 从data选择特定事例的方法
    • TTbar选择方案为top组标准方案
    • Wjets选择方案和TTbar选择方案除了Bjets选择外,其他相同。Wjets选择方案要求没有Bjets。
      • 对Wjets事例的选择效率:假设light jet的fake rate为f,一般f是个很大的数,其他cut选择了4个jet时,再要求其中没有bjet,通过率为~1-4*1/f, 5个jet时的通过率:1-5*1/f,远大于TTBar选择方案对Wjets事例的选择。
      • TTbar事例的混入:根据Bjet tag效率50%,可知25%的事例中将没有Bjets,因此可以混入25%的TTbar事例(忽略轻jet的fake),和TTbar选择方案选择TTbar事例的效率相当。
      • 为了比较wjets事例的重建效果,从light jet中以50%的概率,任意指定为bjets,随后再通过top pair重建。
    • QCD选择方案:要求事例中没有轻子(任何可疑轻子),没有b tag的jets,其他和TTbar选择方案相同。
      • 通过选择的事例很少,是否应该放松一下(没有可疑轻子)的选择。
      • 没有b tag的jet的选择是否必要?
      • 为了比较QCD事例的重建效果,从jet中任意指定一个jet为轻子,从其余的light jet中以50%的概率,任意指定为bjets,随后再通过top pair重建。
  • Background estiomation
    • Data driven estimate QCD background.
      • method 1 (obsolete): Take control sample: small met & leading pt electron or muon without quality cut as the lepton (remove jets which overlap with the lepton). these should be dominated by QCD events. Pass them through ttbar reconstruction to determine the cospsi shape of QCD. Scale the shape to the estimated number of QCD events. This should be validated whether the shape of cospsi of different met bin are same: it is proved NOT, because the met is used in the W reconstruction and cospsi calculation.
        • method1 patch (obsolete): on the "top" hadronic side, omitting Pz of one light jet (randomly) to mimic met, the other light jet as lepton. Then reconstruct the W as if W decay leptonically and then "leptonic top". If the cospsi distributution of both "lep and had" top side are same, then we can use low met to take pure QCD events and use the 3 jets to mimic lepton and met to calculate the cospsi. the met pt distribution and the jet pt distribution are not same, which makes the cospsi shape NOT same.
      • method 2:
        • MC or control data: cuts (e.g. no medium electron above ) select un-biased dominant QCD events, other backround and signal should be supressed
        • electron in QCD comes mainly from jet-fake, so take random jet as electron. Cut "electron" and met >20GeV.
        • Should prove that electron + met =W and jet+met=W are same in 4m?
        • take one bjet or random jet as bjet (if there is). Shoud test here W+bjet=top has same cospsi shape for jet-electron and electron.
        • Should test extra jets change the cospsi distribution or not?, if yes, need 3 extra jets in the event selection.
        • need the top mass is around 172.5
        • calculate the cospsi, as the background shape of QCD.
        • muon channel, muon in QCD comes mainly from jet decay, hard muon takes most energy of the closest jet? muon is little softer than electron and jets?
        • check 4m of muon, take jets as muon, scale event weight if 4m not same
        • if no much statistical, use looser cuts to test this distribution is same as the standard reco method.
    • Data driven estimate WJets background.
      • method 1:
        • MC or data: top group standard cut for version 16 .
        • 使用W/Z ratio方法估计Wjets事例的数量
        • 使用Wjets选择方案选择Wjets事例,估计事例的cospsi形状
          • 比较Wjets选择方案和TTbar选择方案选择的Wjets事例的所有变量具有相当的运动学特征
          • 比较Wjets选择方案和TTbar选择方案选择的Wjets事例的cospsi shape is same.
    • MC estimate of other non ttbar background
    • MC estimate conbinatorial background
  • Extract measurement
    • Reconstruct ttbar events with KLFitter and CSC method when statistics increase(?)
    • Remove background of above estimate
    • Correction function take from signal MC method to correct the signal distortion
    • Function fitting (LH fit for the 2D FL-FR)

Systematics estimation

  • docu:
  • relative x-sections of background to signals
    • in the fit, changing the relative background ratio to evaluate the errors.
  • background shape uncertainties, with psudo-experiment to estimate the errors
    • data-driven estimated Wjets shape:
    • data-driven estimated QCD shape:
    • other small background
  • ISR/FSR, PDF, top mass, JES, The b-tagging efficiency and mistag fraction, met scale makes the correction function different, which makes the fit results different, as the systematic error
    • ISR/FRS: more samples varying the parameters controlling ISR and FSR in a range consistent with experimental data. Conventionally using ISR/FSR on/off samples.
    • top mass: difference top mass sample.
    • PDF: PDF reweight method
    • JES: varies in the range 6–10% as a function of jet pT and Eta;, the varies are used to reproduce the correction function.
      • 2011.02.17: a recommended preliminary jet energy scale uncertainty for b-tagged jets (using SV0 tagger) of 2.5% for pt>25 GeV.
    • The b-tagging efficiency and mistag fraction: the varies are used to reproduce the correction function.
    • met scale: he varies are used to reproduce the correction function.
  • pileup

Ensemble test

  • compare the final objects' kinematics of MC ttbar events with data (ttbar control sample)
    • to make sure whether the combinetorial bagd is same for data and MC


  • Version_1_0: cut flow OK for version 16
    • svn co svn+ssh:// cutflow
  • Version_2_0: add Wjets and QCD selection procedure for version 16
    • svn co svn+ssh:// topAna
  • develeping version ( which is updated regularly
    • svn co svn+ssh:// MyD3PD_Ana
  • 2011-08-12, Whel note proval meeting, corresponding to our int note version 4 at The code to get this note is
    • D3PD process code: svn co svn+ssh://
    • Analysis code: svn co -r 233 svn+ssh://
    • The note tex: svn co -r 232 svn+ssh://

Test run

run small example

  • data file\x{ff1a}

  • run with parameter MC\x{ff1a}
    • total events\x{ff1a}4995: 1874 el triggered, 1694 mu triggered, 1760 not triggered , 333 both triggered.
    • selected 334 = e 143 + mu 191\x{ff0c} 32 with e+mu trigger

  • run as muon stream:
    • 1694 mu triggered 3300 not mu triggered
    • selected 191, 178 Topl recoed.

  • run as electron stream:
    • 1874 e triggered 3020 not mu triggered
    • 143 e triggerd, 89 recoed

needed material:



trigger search:

atlas trigger menu:


ATLAS Official GRL:

GRL: and how to use:

object selection:


Data period:

top group D3PD variable list

top crosssection area:

top group dAOD

Top MC and data

Topinput D3PD maker

lowest un-prescaled trigger chain:

remove double event:


e_mu overlap removal

lumi calculation

cut flow:


pileup reweighting



plot style:

relative result

top cross section plot:

Skim and Slim

No. select condition Data type Data name DaTRI Site
1.1 topEl_n+ topMu_n > 0 data10 Egamma(E4-I) user.limingchen.slimD3PD_electron_201102202133/ CERN-PROD_SCRATCHDISK
1.2 topEl_n+ topMu_n > 0 data10 Muons(E4-I) user.limingchen.slimD3PD_muon_201102202133/ CERN-PROD_SCRATCHDISK
1.3 topEl_n+ topMu_n > 0 MC user.limingchen.slimD3PD_MC_201102202133/ CERN-PROD_SCRATCHDISK


Topic attachments
I Attachment History Action Size Date Who Comment
Texttxt SlimCuts_leptonMoreThan0_jetsMoreThan3.txt r1 manage 2.6 K 2011-06-11 - 18:39 ChengguangZhu  
XMLxml Top_GRL_D.xml r1 manage 34.8 K 2011-06-11 - 18:29 ChengguangZhu  
Unknown file formatext datasets_2011_MC10a r1 manage 7.2 K 2011-06-11 - 18:31 ChengguangZhu  
Unknown file formatext datasets_2011_MC10b r1 manage 7.1 K 2011-06-11 - 18:30 ChengguangZhu  
Unknown file formatext datasets_2011_realdata r1 manage 8.5 K 2011-06-11 - 18:30 ChengguangZhu  
PDFpdf main.pdf r1 manage 544.0 K 2011-08-14 - 00:32 ChengguangZhu  
Edit | Attach | Watch | Print version | History: r46 < r45 < r44 < r43 < r42 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r46 - 2011-08-14 - ChengguangZhu
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Main All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2023 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback