CMSPublic Web>SWGuide>SWGuideFwdPhysics>SWGuideFwdSingleDiffractiveDijet (2012-01-02, DavidDEnterria)

EditAttachPDF

Single diffractive dijet production software page

Contents:

Code and tags
Analysis software

Code and tags

The single diffractive dijet analysis software resides in DiffractiveForwardAnalysis/SingleDiffractiveDijetsAnalysis

To use it, follow the procedure reported here, tested in CMSSW_2_2_9 on the Winter09 Fast Sim Production.

In the CMSSW_2_2_9/src directory check out and compile the following packages:

cvs co -r v02-fastsim DiffractiveForwardAnalysis/SingleDiffractiveDijetsAnalysis

cvs co -r V00-00-01 DataFormats/CMS.CastorReco
cvs co -r V00-01-00 FastSimulation/ForwardDetectors
cvs co -r V00-01-00 RecoLocalCalo/Castor
cvs co SimDataFormats/GeneratorProducts

(needed to run the FastSimulation of the CASTOR detector)

cvs co -r V01-07-02 CondFormats/JetMETObjects
cvs co -r V01-08-13 JetMETCorrections/Configuration
cvs co -r V02-09-02 JetMETCorrections/Modules
cvs co -r V01-08-00 JetMETCorrections/Algorithms

(Winter09 jet corrections in CMSSW_2_2_X)

cvs co -r CMSSW_2_2_9 PhysicsTools/HepMCCandAlgos
rm PhysicsTools/HepMCCandAlgos/plugins/GenEventKTValueProducer.cc
cvs co PhysicsTools/HepMCCandAlgos/plugins/GenEventKTValueProducer.cc
scram b

This version of the code is meant to be run on all Winter09FastSim signal and background samples and produces results for different HF and CASTOR noise threshold setting.

Analysis software

The analysis is organized in two steps:

sample reduction (skimming)
analysis of the skimmed samples

Skimming code

The script used for the skim process is crabCallerSkimmer.py

. It prepares the CMSSW configuration file (pset.py) and crab.cfg file and submits jobs to the Grid using CRAB.

The skimmed sample contains:

CMS.CaloTowers
CorrectedCaloJets (needs to include uncorrected CaloJets)
GenJets (sisCone5)
genEventKTValue (pthat for Herwig)
genEventScale (pthat for Pythia)
recoGenParticles
CASTORTowers
recoTracks
TriggerResults

The skimming code also applies a selection at trigger level: only events that pass the HLT_DiJetAve30 conditions are saved.

Set your configuration by editing the script crabCallerSkimmer.py (see next section) and run it:

chmod +x crabCallerSkimmer.py
./crabCallerSkimmer.py

How to configure skim jobs

The following parameters need to be configured by the user before running the script crabCallerSkimmer.py:

storage_element: storage element where skimmed samples are going to be stored; UERJ (default) is a good option
njobs: the number of jobs to be submitted
correction (°): the correction service to be used
cross_section (°): the cross section relative to the analyzed sample; it is not used by the present code
dataset (°): the dataset to be skimmed
uiworkingdir (°): the crab working directory to be created; it will also be the base of the skimmed dataset name
remotedir (°): the directory where skimmed samples will be stored in the chosen storage element . Note: *every time* a job is submitted, this parameter has to be changed to something never used
sewhitelist (°): list of storage elements (SE); only jobs accessing data at SE in the chosen sewhitelist will be created and then submitted
generator (°): the generator used for MC production

The parameters marked by (°) need to be configured for each sample one wants to skim. More information about CRAB settings can be found here.

For each dataset to be skimmed, set the corresponding parameters and uncomment the line

ConfigHandler(correction,cross_section,dataset,njobs,uiworkingdir,remotedir,sewhitelist,generator)

placed the end of the parameter list; when the script is run, this will create the pset.py and crag.cfg and submit the job.

Skimmed dataset publication

To publish the skimmed data on DBS type:

crab -publish -c UI_WORKING_DIR

where UI_WORKING_DIR has to be the same directory configured in the script script _crabCallerSkimmer.py
By default, data will be published on dbs_ph_analysis_02

. To change it, follow instructions given in the CRAB documentation page.

Data analysis code

The directory SingleDiffractiveDijetsAnalysis/src

contains the following files, each dedicated to a particular step of the analysis:

DijetsFilter.cc selects dijet events
GapSideProducer.cc determines the rapidity gap side
GenScaleFilter.cc filters events with pthat lower than 15 (background) or 30 (signal), not used for the analysis
KinPlotter.cc produces kinematical plots for signal and background
TrackMultiplicity.cc calculates the track multiplicity
WeightProducer.cc associates a weight to the event, based on pthat; weights calculated here assume that all events of each signal and background Winter09 Fast Sim dataset are run for the analysis.
SdDijetsPlotter.cc produces multiplicity plots

The code runs on skimmed data using the script crabCallerAna_Skim_Fast_v1.py which creates the CMSSW configuration file (pset.py) and crab.cfg file and submits jobs to the Grid using CRAB.

Set your configuration by editing the script crabCallerAna_Skim_Fast_v1.py (as described below) and run it:

chmod +x crabCallerAna_Skim_Fast_v1.py
./crabCallerAna_Skim_Fast_v1.py

How to Configure analysis jobs

The following parameters can be configured in crabCallerAna_Skim_Fast_v1.py:

HF and CASTOR noise thresholds. Lines like

Hf1Castor100 = {"name":"Hf1Castor100","hfnoise":"0.1","castornoise":"10.0"}
parametersList.append(Hf1Castor100)

are used to set different HF and CASTOR noise thresholds. The command parametersList.append(...) creates a new path which corresponds to the chosen HF and CASTOR threshold configuration (0.1 GeV for HF and 10 GeV for CASTOR, in the example). One can add as many paths as one wants, but has to take into account the increasing size of the output file.
GenLevel CASTOR threshold are fixed to 10.0 GeV.

Job Configuration. Lines like

##PomWigDiJetsMinusPt80
generator = 'pomwig'
dataset = '/PomWigDiJetsMinusPt80/fsilva-PomWigDiJetsMinusPt80_Skim_FastSim_v1-d9106ce61bbaa279a796fd9486526c79/USER'
#njobs = '2000'
uiworkingdir = 'PomWigDiJetsMinusPt80_Ana_UIDIR'
##options: uerj,kfki,pic
##use '' for no se_white_list
sewhitelist = 'uerj'
ConfigHandler(generator, lumi, rgsp, eff, parametersList, dataset, njobs, uiworkingdir, sewhitelist)

configure and submit jobs for different datasets. For each of them, the user has to set the following parameters:

generator: generator used for MC production; it makes difference on the weight
dataset: skimmed dataset to be analyzed
uiworkingdir: name of the crab working directory. It will be crated in the directory where the script crabCallerAna_Skim_Fast_v1.py is run; in its subdirectory crabCallerBackup one can find the pset.py and crab.cfg used.

As for the skimming, the script will submit jobs only for datasets for which the line

 
ConfigHandler(generator, lumi, rgsp, eff, parametersList, dataset, njobs, uiworkingdir, sewhitelist)

is uncommented.

Output

The code will dump out several root files (one for each job). They have to be merged in the following way (hadd is a good tool for this):

ALL.root: all files
PomWigDiJets.root: Signal files
QCD.root: Background files

plotProducer.C is a ROOT macro which will create multiplicity and kinematical plots. There are a few things to be configured before running the macro:

Lines like

   kinPlotsFolders.push_back("OfflineKinPlotsHf1Castor100");

define directories (inside the root files, check them) to be opened to produce kinematical plots. They need to exist inside the ROOT file and correspond to different threshold levels configured in crabCallerAna_Skim_Fast_v1.py.

Lines like

   data_[0] = "sdDijetsPlotterInfHf2Castor100";
   data_[1] = "sdDijetsPlotterInfRandomHf2Castor100";
   //data_[2] = "N^{ max}_{ track} = Inf";
   data_[2] = "No N^{ max}_{ track} cut";
   multPlotsFolders.push_back(data_[0]);
   multPlotsFoldersRandom.push_back(data_[1]);
   multPlotsTitles.push_back(data_[2]);

   data_[0] = "sdDijetsPlotter5Hf33Castor100";
   data_[1] = "sdDijetsPlotter5RandomHf33Castor100";
   data_[2] = "N^{ max}_{ track} = 5";
   //data_[2] = "No N^{ max}_{ track} cut";
   multPlotsFolders.push_back(data_[0]);
   multPlotsFoldersRandom.push_back(data_[1]);
   multPlotsTitles.push_back(data_[2]);

do the same for multiplicity plots. Each block corresponds to a different HF - CASTOR threshold configuration, but with the additional parameter n_track cut. In the examples above, the first block corresponds to an HF threshold of 0.2 GeV, a CASTOR threshold of 10 GeV and no track cut (Inf); the second block corresponds to an HF threshold of 3.3 GeV, a CASTOR threshold of 10.0 GeV and an n_track cut = 5.

To execute the macro type

root -b -q plotProducer.C

In the end there will be a set of directories, each corresponding to a different type of plot (kinematical of multiplicity), HF and CASTOR threshold level and n_track cut.