HLTriggerOffline/SUSYBSM Package description

This package is designed to compare the trigger behaviour in CMSSW release by release in order to find bugs or unexpected troubles. We compare two samples built with CMSSW_x and CMSSW_(x+1). A set of benchmark plots is built for both samples. The code checks in an automatic way if the plots for the two samples are consistent within a given tolerance. The package is organized in two different pieces of code:
  • A CMSSW analyzer runs over the two samples to be compared and produces a ROOT file with histograms for each sample
  • A python script takes the histograms and calls a ROOT macro which calculates the efficiencies and produces the comparison plots. It writes a warning in a log file if discrepancies are observed.

A web page is automatically generated, containing some reference plots and those plots for which discrepancies have been observed. A logfile reporting all the warnings is accessible and it is anyway possible to access all the plots, even if they are not visualized in the web page.

The ROOT files contain the L1 and HLT bits and distributions of some variables of main Reco objects, according to a definition useful for SUSY and Exotica analyses:

  • Multiplicity, pt, eta and phi for:
    • Electrons: PixelMatchGSFElectrons, pt>10 GeV
    • Muons: muons, pt > 7 GeV
    • Jets: IC5 Calo, uncorrected, pt > 30 GeV
    • Photons: photons, pt > 30 GeV
  • MET, METphi and Sum(Et) for:
    • MET: met
All the distributions are plotted also for events passing every single bit of L1 and HLT.

The comparisons are not performed on all plots, but only L1 and HLT efficiencies and on multiplicities for Electrons, Muons, Jets and Photons and on MET distribution. The comparison is preformed also on multiplicities and MET distribution for every L1 and HLT path with efficiency greater than 30%. For the L1 and HLT efficiencies, a warning is given if a discrepancy of 5 sigma is found.

The code contains also two selection templates, one based on Reco objects and one based on MC truth. In the root file, two folders called "McSelection" and "RecoSelection" are created, containing the L1 and HLT bits after the selection. The user has to modify the selection templates to adapt them to his specific analysis, in order to get the trigger efficiencies for his signal for a specific final state, selected according to the RECO objects in the event or to the MC information. The MC selection template can be switched off in the configuration file in order to make the code working with true data.

A detailed description of the package can be found in CMS IN-2008/030.

The CMSSW analyzer is in HLTriggerOffline/SUSYBSM. The python script for the plot production is in HLTriggerOffline/SUSYBSM/test/BSMTriggerCheck/

List of Samples

Unlike other PAGs, for SUSY and Exotica there is not a clear definition of interesting events. We identify a minimal set of MC samples representing as much as possible the many New Physics signatures expected at LHC. Need as much samples as possible samples for a wide coverage.

For the moment, we chose six signals:

  • SUSY
    • LM1 (Jets+MET+leptons)
    • LM5 (Jets+MET+leptons, rich in b's and taus)
    • LM9p (low pt spectra, trileptons)
  • Exotica
    • GM1b (photons)
    • RSgrav (750 GeV) (high pt di-leptons and di-jets)
    • Z' (750 GeV) (high pt di-leptons and di-jets)

We would like to add other samples for a wide coverage of signatures:

  • non-pointing objects
  • large lifetime objects
  • 3rd generation dominance (is LM5 enough?)
  • New Physics models with compact spectra (low pT objects + MET in cascade)
  • ...

Integration with RelVal workflow

The RelVal team accepted to put two of our samples in the official RelVal production.

  • LM9p and RSgrav are in the high-stat slow turnaround set
  • LM1 is in the low-stat fast turnaround set (requested by JetMet POG)

The first piece of the code (the CMSSW analyser filling the histograms) is ready for the integration in the RelVal workflow. The code has been translated to make it compliant with the DQM framework. The second piece of the code (the efficiency calculations and the publishing of the final web page containing the plots) is still stand-alone. We plan to translate it step to step, checking what features are available in the DQM gui.


The results of comparisons for past releases can be found in the web page: http://cms-cern.web.cern.ch/cms-cern/mpierini/BSMTrigVal/ and in: HLTTriggerMenuValidation

-- 30 Jan 2009

Responsible: JasonSlaunwhite

Edit | Attach | Watch | Print version | History: r8 < r7 < r6 < r5 < r4 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r8 - 2011-05-30 - MonicaVazquezAcosta
    • 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 & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback