You are here: Foswiki>ATLAS Web>LiverpoolZBosonAnalysis (22 Jul 2008, PeterVankov)Edit Attach

LiverpoolZBosonAnalysis

Introduction

The properties of Z boson decays (such as Z → ee and Z → μμ) are known to high precision from the LEP experiments. At LHC the Z bosons will be produced at very high rate, thus giving the opportunity their leptonic decays to be used as physics processes for calibration and alignment of the ATLAS detector. In addition, the Z bosons are expected to be the dominant background source for many of the physics channels that ATLAS aims to study. It is therefore necessary that the Z production and reconstruction at LHC and ATLAS are understood and performed as well as possible.

Z-analysis framework

To be ready for the Z's at ATLAS, Liverpool HEP group has started work on a general analysis for offline reconstruction of Z bosons at ATLAS, called LiverpoolZBosonAnalysis. The current version of the analysis code, running under the Athena framework, can be found at the http://atlas-sw.cern.ch/cgi-bin/viewcvs-atlas.cgi/groups/Liverpool/LiverpoolZBosonAnalysis/ ATLAS CVS repository.

Features

- The code is compatible with Athena 14.1.0.

- The code accomplishes Z reconstruction based on electron and muon preselections.

- A possibility for taking into account the trigger decisions is included.

- The code provides the basis for the development of more complicated analyses, such as H → ZZ(*) → 4l (see http://hep.ph.liv.ac.uk/twiki/bin/view/ATLAS/HiggsToFourLeptons LivH4lAnalysis).

Current Algorithms

(Pre-) Selection algorithms

These implement common selection criteria, outputting a filtered container that can be used as input for subsequent algorithms. The algorithm name is ParticlePreselection, it simply acts as a placeholder for various selection tools, which do the real work. The choice of which tools to use (and therefore which selections are applied) is completely up to the user, depending on what they put into their job options file. The actual selection tools so far are:
  • ElectronPreselection provides a completely configurable selection for electrons, including author (Electron or Softe), kinematics (pT, eta, crack), and IsEM.
  • MuonPreselection so far only provides basic kinematic selection (pT, eta) - work is in progress.
  • TruthFilter allows separation of the generator truth from non-generator truth, for ease of processing downstream.

"Calculational" algorithms

These are algorithms and tools that do more specialised, sometimes more complicated, things. So far, we have:
  • Combine2Particles, which makes composites from either one or two input particle containers. This is a generic algorithm, and can be used to construct Z->ee, Z->mumu, H->ZZ, and potentially any signature of the form A->BC. The algorithm writes a CompositeParticleContainer to StoreGate, which can be accessed by subsequent algorithms just like any other container.
  • RecalibrateElectrons, a trial algorithm which performs 4-momentum rescaling and smearing for electrons. It could be used as a template for (as-yet unwritten) RecalibrateMuons, RecalibrateComposite, ... . As with ParticlePreselection, the real work is done using tools, of which there are currently two:
    • BasicChange4MomTool can be used to apply systematic shifts in the 4-momentum of a particle as an arbitrary function of its coordinates (the functions use ROOT's TFormula string-parsing mechanism).
    • GaussSmear4MomTool can be used for simple shifts, but can also apply Gaussian smearing to the coordinates. Again, the use of TFormula means that the mean and width of the Gaussians can be arbitrary functions of the particle's 4-momentum.

NTuple dumping algorithms

These can be used at various points in an analysis, eg before preselection and again afterwards. Most can be configured to dump more or less information as required. The tree structure and branch naming scheme can again be configured completely within the job options.
  • DumpEventInfo Event data (lumi block, time stamps, MC event weight, ...). NB: The event and run numbers appear anyway, even if this algorithm is not called.
  • TriggerResult Dumps the trigger result (L1, L2, EF) for a single, named signature.
  • DumpElectrons Electron AOD variables.
  • DumpMuons Muon AOD variables.
  • DumpCombinedParticle Any CompositeParticleContainer can be used as input. Output: 4-momentum, pdgId and charge of the composite and each constituent.

Job options

There are a number of example job options (LivZBosonExample_topOptions.py, LivH4lAnalysis_topOptions.py, LivZBosonSystExample_topOptions.py) that each demonstrate and test various parts of this framework. The file LivZBoson_jobUtils.py also contains python functions of more general interest, to automate things like dataset finding and adding output ntuple streams to a job.

To be done

- Including MC truth information.

- Improving the electron preselection. (in process, Mike Flowerdew)

- Improving the muon preselection. (in process, Peter Vankov)

- Improving the trigger (in process, Mike Flowerdew)

- Implementing cut flow tables

- Adjust the code to perform studies of possible systematic effects.

- hadronic tau preselection and dumper

- missing Et dumper (needed for tau)

Useful

Some CVS settings and commands

To set access to CVS (after the athena environment has been set up):

export CVSROOT:gserver:isscvs.cern.ch:/local/reps/atlas

kinit user@CERN.CH

To update files in CVS Full instructions:

cvs update -A

cvs ci -m "some description" file1 file2 ... fileN

To add new files in CVS:

cvs update -A

cvs add file1 file2 ... fileN

cvs ci -m "some description" file1 file2 ... fileN

Datasets and name conventions

Athena dataset naming conventions (AMI)

ID geometry version names

DataSetNameProcess NumFiles Location Produced by Notes
valid1.005300.PythiaH130zz4l.recon.AOD.e322_s401_r446_tid022720 40 /hepstore/store3/vossebeld/v14/ Joost
valid1.005145.PythiaZmumu.recon.AOD.e322_s443_r455_tid022885 40 /hepstore/store3/vossebeld/v14/ Joost

Grid submission

See instructions on how to set up ganga+athena consistently here (Provisional ganga 5 instructions are here.

Talks and Presentations

-- PeterVankov - 16 Jun 2008
Topic revision: r11 - 22 Jul 2008, PeterVankov
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback