Searching for SUSY on di-stop production with semi-leptonic final states using matrix elements

Analysis Description

This analysis consist on the search for Supersimetry trought di-stop production resulting on semi-leptonic final states containing b-jets, jets produced from hadronization processes of one of the Ws and a soft lepton (electron or muon) from the other associated W boson plus missing energy from the neutrino produced in the leptonic decay of the W.

The main background of our channel is the production of a pair of top quarks with the same final state. Therefore, the key component of this analysis is to reduce to the maximum this copious background in order to improve the signal signifincance. We accomplish this by using a matrix elements technique that allow us to identify the best possible permutations of jets comming from the top decays. Based on those permutations we find the best top-pair and reject those events from our signal selection.The matrix elements method is carriedout using the program MadWeight. For each event, we determine a specific weight that is the convolution of the squared matrix element and the resolution function. All the relevant information on this program can be found here:

https://cp3.irmp.ucl.ac.be/projects/madgraph/wiki/MadWeight


Getting the code from GIT

  1. Create an account in GIT: https://github.com/
  2. In the bottom right of the page select "ssh" in order to get your GIT address e.g. git@githubNOSPAMPLEASE.com:user/Code.git
  3. Go to lxplus and do:
    • $ git config --global user.name <YOUR USER NAME>
    • $ git config --global user.email user@cernNOSPAMPLEASE.ch
    • $ git clone ssh://lxplus.cern.ch/afs/cern.ch/cms/DB/rep/cmsDbWebServices.git
    • cd ~/.ssh/
    • $ ssh-keygen -t rsa -C "your_email@example.com" (same e-mail and password used when you oppend the GIT account)
  4. Go to https://github.com/settings/ssh
  5. In "Add ssh key" place a any title you like (e.g. laveLXPLUS) an copy the text from ".ssh/id_rsa.pub" on lxplus to the key-box.
  6. In order to get the code, go to the GIT page and on the search window tipe "djp28/ThesisCode.git" and then click on the "Fork" widget.
  7. Go back to lxplus and "cd" into "cmsDbWebServices"
  8. Then type: "git remote add origin git@githubNOSPAMPLEASE.com:djp28/ThesisCode.git"
  9. git branch <NewBranch> (you choose the name of the "NewBranch" e.g Analysis)
  10. git checkout <NewBranch>
  11. git clone git@githubNOSPAMPLEASE.com:djp28/ThesisCode.git
  12. Some useful commands:
    • git diff
    • git push --all <project> (in our example project would be "origin")
    • git push --all <project>
    • git remote -v show
  13. Some documentation can be found here: http://net.tutsplus.com/tutorials/other/the-perfect-workflow-with-git-github-and-ssh/

Installing the code

Currently, we are using the "CMSSW_5_3_9" release. Be aware that this release is only available on SLC5 architecture machines on lxplus. For convenience, try to work on your "work" area on lxplus. if you do not have a work area, please request one.

  • First, setup the release on lxplus: cmsrel CMSSW_5_3_9
  • from the "cmsDbWebServices/Code" directory, copy the whole content in the "src" directory to the "src" directory to "CMSSW_5_3_9/src"
  • cd to "CMSSW_5_3_9/src"
  • cmsenv
  • ./setup_for_pat.csh (to setup all the external packages)
  • scram b -j8

Structure of the Code

Currently, the analysis is ran in several steps.

  1. Patuple skimming and slimming
    • In this step we select only the collections of interest for our analysis. Furhtermore, we require a few preselection criteria on the different objets (muons, jets, electrons etc) in order to skim our samples, to reduce them to a more manageble size.
    • The script used to submit the crab jobs is called "makingPattuplesWithCrab.sh". In the " src " directory you'll find a text file called "pruebaData.txt". In this text file you'll see three parameters: data set name, output index name and data type (data or MC). When running the "makingPattuplesWithCrab.sh" you must provide a similar text file (pruebaData.txt) with the relevant information for your run.
    • In order to run the file do: $ ./makingPattuplesWithCrab.sh pruebaData.txt.
    • The python file with all the sliming and slimming criteria is called "pat_tupleWithoutTriggerInfo_data.py" and can be found in the "src" directory.
  2. Preselection (best possible combinations)
    • Create a directory called "resultsCrab" inside "CMSSW_5_3_9"
    • The script used to run the jobs is called "makingPreselection.sh".
    • In order to run the "makingPreselection.sh" script you need once again a text file as the "pruebaData.txt" decribed above.
    • The "makingPreselection.sh" script calls another script called "Preselection.sh" which in turn call a "preselectionStudy_template.py" python configuration file if running on MC or a "preselection_template.py" config if running on data.
    • The "preselection*_template.py" scripts configure the code that performs the preselection of the best possible combinations of jet candidates resulting from the top decay. This preselection is done using a discriminant. The discriminant is determined using a anti-b-tag discriminant, a ΔR between one of the b-jets and the lepton from the W, the mass resolution obtained for the hadronic decay of the W (wrt the mass value fromt he PDG), and the mass resolution for the reconstructed top wrt to its PDG value. The anti-b-tag discriminant results from the fit to a polinomial function, obtained from plotting the number of events vs the b-tagging dicriminator (for b-jets and non b-jets).
    • You can find the "PreselectionStudy.cc" and the "Preselection.cc" codes at the "CMSSW_5_3_9/src/Analyzers/" directory.
    • As a results from running the preselection step, you'll get a set of text files inside the e.g. "resultsLocal/LHCO_data2/" directory. The text files contain information of the kinematics of the best combinations of jets that passed the preselection step. There could be upto 100 combinations per event and several events can be stored on each file.
  3. Splitting (per event number)
    • The splitting step aims to organize the information optained in the preselection step on an event basis. This is accomplished by storing the information in an individual file per event.
    • The script that does the splitting is called "makingSplittedLHCOFiles.sh".

Data and MC

  1. Data
    • We are using the collected data by the CMS experiment during the 2012 run at 8 TeV.
    • In order to select only good quaility events, with all the CMS subsistems in optimal run conditions, we are using the JSON file: Cert_190456-208686_8TeV_PromptReco_Collisions12_JSON.txt
    • The data set we are using is: /SingleMu/Run2012D-PromptReco-v1/AOD
  2. MC
    • An oficial CMS MC sample (centraly produced) is being used to conduct our studies on the ttbar background.
    • The sample can be found as: /TTJets_SemiLeptMGDecays_8TeV-madgraph/Summer12_DR53X-PU_S10_START53_V7A_ext-v1/

Useful Information

Preliminary Results

Presentations

Analysis Note

Important links

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2013-08-30 - AndresCarlosFlorezBustos
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    Sandbox All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback