ssh lxplus
scramv1 project CMSSW CMSSW_5_3_22
cd CMSSW_5_3_22/src/
cmsenv
wget -q -O - --no-check-certificate https://raw.github.com/stiegerb/TopMassSecVtx/master/TAGS.txt | sh
git clone git@github.com:stiegerb/TopMassSecVtx.git UserCode/TopMassSecVtx
scram b -j 9
Note: the ntuplizer is running without pileup jet id and IVF-related b-tagging variables
sh test/topss2014/submitNtupleProduction.sh [sample=presel,syst,mass,control,pdf]
Will create the base ntuple summary
./scripts/runPlotter.py --rereadXsecWeights /store/cmst3/group/top/summer2015/bbbcb36/ -j test/topss2014/samples.json,test/topss2014/syst_samples.json,test/topss2014/mass_scan_samples.json,test/topss2014/qcd_samples.json,test/topss2014/z_samples.json,test/topss2014/z_syst_samples.json,test/topss2014/photon_samples.json
Will create a pickle file summarizing all the cross section-based normalization to be used
./scripts/runLxyTreeAnalysis.py -o treedir -j 8 /store/cmst3/group/top/summer2015/bbbcb36/
./scripts/runLxyTreeAnalysis.py -o treedir/syst/ -j 8 /store/cmst3/group/top/summer2015/bbbcb36/syst/
./scripts/runLxyTreeAnalysis.py -o treedir/mass_scan/ -j 8 /store/cmst3/group/top/summer2015/bbbcb36/mass_scan/
./scripts/runLxyTreeAnalysis.py -o treedir/qcd_control/ -j 8 /store/cmst3/group/top/summer2015/bbbcb36/qcd_control/
./scripts/runLxyTreeAnalysis.py -o treedir/z_control/ -j 8 /store/cmst3/group/top/summer2015/bbbcb36/z_control/
./scripts/runLxyTreeAnalysis.py -o treedir/photon_control/ -j 8 /store/cmst3/group/top/summer2015/bbbcb36/photon_control/
Creates the SVLInfo/CharmInfo trees with the condensed summary info for the final analysis
./scripts/mergeSVLInfoFiles.py treedir/
./scripts/mergeSVLInfoFiles.py treedir/syst/
./scripts/mergeSVLInfoFiles.py treedir/mass_scan/
./scripts/mergeSVLInfoFiles.py treedir/qcd_control/
./scripts/mergeSVLInfoFiles.py treedir/z_control/
./scripts/mergeSVLInfoFiles.py treedir/photon_control/
Will merge all the chunks and then move them into treedir/Chunks/ There is a version of the trees from Feb10 that is still current in:
/afs/cern.ch/work/s/stiegerb/TopSecVtx/SVLInfo/Feb10
You can draw the plots with runPlotter.py -l 19701 -j test/topss2014/samples.json -o treedir/plots treedir
./scripts/makeSVLMassHistos.py treedir/ -j 8 -o outDir/
./scripts/makeSVLSystPlots.py treedir/ -j 8 -o outDir/
Will run the plots and put them in svlplots/ by default. Also saves the histograms in a cachefile. To rerun only the plotting part, run with -c option. (makeSVLMassHistos takes about 25 min to run with -j 8, makeSVLSystPlots takes about 50 min. Run both with -c to just produce the plots from pre-cached files.)
In order to have correct xsection weights and number of generated events, one needs to run on the eos/ directory first to produce a cache file with the weights, like so:
./scripts/runPlotter.py --rereadXsecWeights /store/cmst3/group/top/summer2015/bbbcb36/ -j test/topss2014/samples.json,test/topss2014/syst_samples.json,test/topss2014/mass_scan_samples.json
Then use makeSVLDataMCPlots.py to produce data/MC comparison plots:
./scripts/makeSVLDataMCPlots.py treedir/ -j test/topss2014/samples.json -o outDir/
This produces a number of data/MC comparison plots (both from the SVLInfo trees, and from the histograms produced in the LxyTreeAnalysis). Default output directory is plots/. This is based on the runPlotter.py script.
The script also produces the DY scale factors and applies them directly. It puts the control plots in a sub-directory called dy_control. Note that they are read from the cachefile (.svldyscalefactors.pck) by default. To reproduced them from scratch, delete the cachfile first. To produce only the scale factors, use the extractDYScaleFactor.py script, e.g.
./scripts/extractDYScaleFactor.py outDir/plotter.root --verbose 5
Notice it is not needed to run it explicitely after makeSVLDataMCPlots.py has been run.
To print event yields, run printSVLEventYields.py on the output of the script above:
./scripts/printSVLEventYields.py outDir/plotter.root
To extract the parametrization to reweight the number of tracks in the SVtx to data
python scripts/extractNtrkWeights.py outDir/plotter.root -v 5
makeSVLQCDTemplates.py produces MET and m(l,SV) distributions from the non-isolated data control region and stores them in qcd_templates.root. (Takes about 5 min to run. Run with --cached to run from cached files.)
qcdFitter.py fits the MET distribution in the l+jets signal region, using the shapes from the non-isolated region produced by makeSVLQCDTemplates.py and produces m(l,SV) templates scaled according to the outcome of the fit.
It takes as input a runPlotter output file produced by makeSVLQCDTemplates.py called scaled_met_inputs.root with properly scaled MET distributions in the signal region, and a file with the final qcd templates, also produced by makeSVLQCDTemplates.py called qcd_templates.root.
./scripts/makeSVLQCDTemplates.py treedir/ -o svlplots/qcd/
./scripts/qcdFitter.py svlplots/qcd/scaled_met_inputs.root svlplots/qcd/qcd_templates.root -s
Prepare the pseudoexperiment inputs, which also prepares the background templates for the fitting model.
./scripts/prepareSVLPEInputs.py treedir -o svlplots/ -v 5
Reads the mass scans, systematics histos, and qcd templates from the corresponding cache files, and produces a pe_inputs.root file with the input histograms for running pseudo experiments. Also produces the correctly scaled and added background template shapes and writes them to a cache file to be used in the fitting.
Produce summary plots of the pseudoexperiment inputs:
./scripts/summarizeSVLresults.py --pe svlplots/pe_inputs.root --rebin 2
By default the plots are put in the input directory, under plots/.
./scripts/runSVLFits.py -o svlfits/
Prepares the workspace for the fits and put the RooFit workspace and plots in svlfits/ by default. By default, takes .svlmasshistos.pck and .svlbgtemplates.pck as input for the shapes. (Takes about 45 min to run.)
./scripts/runSVLPseudoExperiments.py svlfits/SVLWorkspace.root svlplots/pe_inputs.root nominal_172v5
Will run the pseudoexperiments for a single variation (e.g. nominal_172v5). Verbose level above 5 prints the result of each channel for each single experiment.
./scripts/runSVLPseudoExperiments.py svlfits/SVLWorkspace.root svlplots/pe_inputs.root -s optmrank
Will run all the pseudoexperiments for all variation on batch. The output (and individual scripts) will be stored under svlPEJobs/. Use -f --filter option to process only selected variations.
./scripts/summarizeSVLresults.py --calib svlPEJobs/optmrank/Apr25/
Will parse the summaries of the pseudo-experiments and produce a calibration file.
./scripts/runSVLPseudoExperiments.py svlfits/SVLWorkspace.root svlplots/pe_inputs.root -c svlPEJobs/optmrank/Apr25/.svlcalib.pck -s optmrank
Will run the calibrated pseudo-experiments
./scripts/summarizeSVLresults.py --syst svlPEJobs_calib/optmrank/Apr25/.svlcalib.pck
Will printout the "final" systematics table
Run the fits on the data to produce the fitting plots and print the final mass results. This also produces a data_results.root file that contains the numbers in PE format for the other scripts to read.
./scripts/runSVLDataFits.py svlfits/Oct13/SVLWorkspace.root svlplots/Oct13/pe_inputs.root -v 5 -o svlplots/Feb18/ -c svlPEJobs/inclusive/Oct13/.svlcalib.pck
Add the --dataresults option to summarizeSVLresults.py to make the final summary plot. Note that this also includes the correct statistical uncertainty from the data.
./scripts/summarizeSVLresults.py --syst svlPEJobs_calib/inclusive/Oct13/.svlcalib.pck -o svlplots/Feb18/ --dataresults svlplots/Feb18/
./scripts/fitSecVtxProperties.py -i treedir/z_control -o outDir/z_control/plots/ --doScales --weightPt --vetoCentral --filter 23
./scripts/fitSecVtxProperties.py -i treedir/ -o outDir/control/plots/ --filter -143,-121,-169
After creating the workspace you can use it directly by using -w workspace.root instead of -i ntuple_dir
deprecated (not sensitive enough)
./scripts/fitSecVtxProperties.py -i treedir/photon_control -o outDir/photon_control/plots/ --weightPt --rebin 2
./scripts/fitSecVtxProperties.py -i treedir/qcd_control -o outDir/qcd_control/plots/ --weightPt --onlyCentral --minLxySig 10
sh runCharmPeaks.sh TREES
sh runCharmPeaks.sh MERGE
sh runCharmPeaks.sh PLOT
sh runCharmPeaks.sh UNFOLD
sh runCharmPeaks.sh DIFF
sh runCharmPeaks.sh DIFFZ
Uses runCharmPeaks.sh to steer the steps of the analysis (from ntuple creation to unfolding and plotting)