Randall-Sundrum Analysis at CMS Experiment

A Log for Flavia's work

How to Run on Data - LHC 2010

Useful Links

• Collisions 2010 Analysis - https://twiki.cern.ch/twiki/bin/view/CMS/Collisions2010Analysis
• Collisions 2010 Datasets - https://twiki.cern.ch/twiki/bin/view/CMS/Collisions2010Datasets
• Collisions 2010 Recipes - https://twiki.cern.ch/twiki/bin/view/CMS/Collisions2010Recipes
• Hcal Anomalous Signals - https://twiki.cern.ch/twiki/bin/view/CMS/HcalDPGAnomalousSignals
• Ecal Anomalous Signals - https://twiki.cern.ch/twiki/bin/viewauth/CMS/EcalFirstBeam09Anomalous
• DBS Discovery - https://cmsweb.cern.ch/dbs_discovery/

• CRAB Guide - https://twiki.cern.ch/twiki/bin/view/CMS/SWGuideCrab
• Thiago's Example for CRAB - http://cmssw.cvs.cern.ch/cgi-bin/cmssw.cgi/UserCode/ThiagoTomei/RSGraviton/RSAnalyzer/analysis_Summer2010/hadronicChannel/rsanalyzer_JetMET_forCRAB_cfg.py?view=log

Data Considerations - November 10th, 2010

Run A = 35x, 36x, (13508 - )
Run B = 38x (145762 and greater)

A joint decision (Flavia, Thiago, Franciole): Wait for reprocessing in Run A with 38x, and start to look at Run B data.

Dataset

/Jet/Run2010B-PromptReco-v2/RECO
To look for it on DBS: find dataset where dataset like /Jet/.../RECO

Run in Selected Runs and LS - JSON File

The JSON file specifies, for a given run range(s), the luminosity of each section. There is a python script that allows, given a bigger official JSON file, to make it shorter, only with the run range you wish to run on.

We filtered the JSON file from this directory to run only in events greater than run 145762 (Run 2010B according to this twiki).

To run on CRAB using the JSON file, we have to change some parameters of the submission script:

-- FlaviaDias - 14 Nov 2010

Z invisible background - Data Driven Estimation

Estimation from the W->mu nu sample, because it would be needed too much luminosity (~1fb^-1) of collected data for a good estimation from Z-> mu mu

Get a control region with maximal purity of W (mu nu) events. Make the same selection one use for the signal (MET and jet), but with different lepton selection cuts:

• Use the VBTF muon criteria, as in this reference:
  • Inside my background MC tree, this is: muonVBTF

This procedure allows to reproduce the same kinematic region we have for the signal, but with one hard isolated muon, coming from the W with high probability. The systematic effects are those related to the muon selection. Then the number of Z background events can be obtained from the control sample after correcting them for the cross section ratio, and muon isolation and reconstruction efficiency.

We also have to apply the cuts to other sources of background, and estimate the contamination it will have in the sample.

Don't cut on MET directly

The Z (-> nu nu) events have a different MET than the W (-> mu nu) events: The MET fot the Z is the vectorial sum of the 4vectors of both neutrinos, while in the Z, the MET is the 4vector of the neutrino. So, to compare the same thing in both cases, I must cut on the MET (for the Z) and for the W, sqrt( ( METcos(phi) + mupTcos(phi) )^2 + ( METsin(phi) + mupTsin(phi) )^2 ).

I have to take care - does root recognizes such a big expression on TCut? So, check using the Scan function on an especific tree to see if it's doing something correct.

LOG - HOW TO USE SCAN:

flavia-dias-computer:~ fladias$ root -l ~/Desktop/RandallSundrum/Backgrounds/Zinvisible/Zinvisible.root
root [0]
Attaching file /Users/fladias/Desktop/RandallSundrum/Backgrounds/Zinvisible/Zinvisible.root as _file0...
Warning in <TFile::ReadStreamerInfo>: /Users/fladias/Desktop/RandallSundrum/Backgrounds/Zinvisible/Zinvisible.root: not a TStreamerInfo object

root [1] TTree* t = _file0->Get("eventAnalyzer/mytree")
root [2] t
(class TTree*)0x10345f700
root [3] t->Print()
[...]
root [4] t->Scan("jet1pt")
[...]

Scan to see if TCut understands the big expression:

sqrt( ( METcosphi + muoncosphi )^2 +( METsinphi + muonsinphy)^2 )

-- FlaviaDias - 13 Dec 2010