Z Reconstruction for Analysis Skims

Complete: 2


This document is a proposal for a AOD extension for analysis studies using the Z. AOD extensions will be produced as output of central skim production during CSA07, and can be configured for specific analysis requirements.

We propose to add to the standard AOD reconstructed Z candidates, Monte Carlo match maps and other possible event contents useful for analysis.

Z reconstruction

Z reconstruction in decays Z→μ+μ- and Z→e+e- can be performed with the common combiner modules provided with the PhysicsTools toolkit using as input the standard collections globalMuons and pixelMatchGsfElectrons respectively. Mild invariant mass cuts can be set in order to leave a sufficient amount of side-band events. We set for all Z collection the cut:

  • mass > 20 GeV/c2.

Configuration fragments could be the following:

Loose Z reconstruction

Z reconstructed from a track plus a lepton or a super-cluster plus an electron could also be useful as "loose" Z sample. An isolation cut on the track would help reducing the combinatorial background, and should be considered (it is not applied at the moment).

The following scripts can be added, and optionally included in the actual skim process:

MC truth

Monte Carlo truth matching in form of one-to-one association map can be stored together with reconstructed candidate collections. The matching is based on the ΔR distance, and a cut ΔR < 0.15 is used in most of the analysis studies presented up to now.

Lepton MC truth matching

Separate match maps for electrons and muons can be stored. Configuration fragments may be the following:

Match map definitions are located in the common package:

Z MC truth matching

Separate match maps for Z→μ+μ- and Z→e+e- can be stored, if we want to keep separate Z candidate collections. Configuration fragments may be the following:

Lepton isolation

generic isolation modules are been defined in the packages:

Isolation information could also be used to preselect tracks to be used for "loose" Z reconstruction.

For the moment, the following isolation modules, based on pt sum, are added to ZReco:

We could probably save some CPU time not computing isolation for all tracks, but for all tracks above a specified pt threshold.

Isolation definitions are located in the common package:

More possible extensions

Further analysis information can be computed on AOD plus the Analysis Collections mentioned above.

Z common vertex fit

Kalman or Gsf common vertex fit can be applied to reconstructed Z with an available module. Whether the Z vertex fit is actually needed, and/or should run at the level of skimming is matter of discussion within the EWK group.

Skim Event Content

We should store as skim output the standard AOD plus all the collections produced (lepton, Z candidates, isolations, MC truth match maps).

At the moment, the test script store a subset of the AOD information.

The actual skim code should be modified in order to save the full AOD.

Package organization

A new package containing .cfi and in the future possibly algorithms (if any) has been created:

We don't need new data formats definition, at the moment, since all output types dictionaries are defined in standard packages.

Event filtering

It is possible to filter only events where a reconstructed Z is present if the input data set is a mixture ("soup") of different processes. Filtering can be managed with the use of generic filter modules. The actual event filtering criteria have to be defined within the EWK group.

For the moment, all events with one Z or a "loose" Z are written to disk.

Running test skims on 1_5_X

A few tags are needed on top of release 1_5_2 The following recipe can be used:

  scramv1 project CMSSW CMSSW_1_5_2
  cd CMSSW_1_5_2/src
  project CMSSW
  cvs co -r EWK_1_5_0_pre6 PhysicsTools ElectroWeakAnalysis
  eval `scramv1 run -csh`
  scramv1 b
  cd ElectroWeakAnalysis/ZReco/test

Run the test script:

  cmsRun testZtoEEReco.cfg

The .cfg scripts under the test directory will run on a standard Z→e+e- release validation sample (that is what is available in the official 1_5_0_preX production up to now...). So, the script testZtoMuMuReco.cfg will run, but will likely not reconstruct any Z candidate.

Running test skims on 1_3_5

Releases 1_5_0 and 1_6_0 can't read data produced with release 1_3_X. Release 1_3_5 contains a back-porting of tools available in release 1_5_0 and can be used with Spring production data.

Release 1_3_5, available on SLC3 only. The following recipe can be used on SLC3 machines, like lxslc3.cern.ch:

  scramv1 project CMSSW CMSSW_1_3_5
  cd CMSSW_1_3_5/src
  project CMSSW
  cvs co -r  V01-02-01-01 ElectroWeakAnalysis/ZReco
  eval `scramv1 run -csh`
  cd ElectroWeakAnalysis/ZReco/test

The following recipe can be used on SLC4, but requires recompiling about 15 packages:

  scramv1 project CMSSW CMSSW_1_3_4
  cd CMSSW_1_3_4/src
  project CMSSW
  cvs co -r  AnalysisTools_1_3_5_prelim2 DataFormats SimGeneral PhysicsTools ElectroWeakAnalysis
  cvs update -r  V01-02-01-01 ElectroWeakAnalysis/ZReco
  eval `scramv1 run -csh`
  cd ElectroWeakAnalysis/ZReco/test

The default test scripts run on samples of Drell-Yan processes from the Spring07 production.

To run skim tests use:

  cmsRun testZToMuMuReco.cfg


  cmsRun testZToEEReco.cfg

Related documents

Review Status

Editor/Reviewer and date Comments
LucaLista - 25 Jun 2007 Page last edited

Responsible: LucaLista
Last reviewed by:

Contact: Michele de Gruttola, Francesco Fabozzi, Luca Lista

Edit | Attach | Watch | Print version | History: r26 < r25 < r24 < r23 < r22 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r24 - 2007-08-05 - NicolaDeFilippisSecondary

    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    CMSPublic All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2021 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