Ntuple Tools

Ntuple Tools

Overview

The final state analysis package is built around the PATFinalState Data Format. This abstract object corresponds to a "final state candidate," like μ+τ, and encapsulates all of the interesting information needed by the analyst. The advantage of the single object is that is both lightweight and that it holds references to all of the interesting information in the event. You can compose complex observables using only a single object, enabling many tasks to use the string cut parser. This allows new cuts to be implemented quickly, and without writing any C++ code.

Building Ntuples

First, ensure that all samples you need have been PAT tuplized. Then, make sure the nice sample names (found in MetaData/python/dataXTeV.py) that you need are mapped correctly to the PAT tuple directories in MetaData/tuples/PATTuple-XTeV.json. Tools for finding the locations of appropriate PAT tuples are at The PAT Tuple.

Then, you can submit jobs ntuple production from NtupleTools/test by running:

submit_job.py JOBID make_ntuples_cfg.py makeQuad=1 makeTNP=1 makeH2Tau=0 makeTrilepton=1 make4L=1 rerunFSA=1 --tuple-dirs=$fsa/MetaData/tuples/PATTuples-8TeV.json --apply-cmsRun-lumimask --input-files-per-job=1 --shared-fs  --samples "VH*" "*WZ*" "*ZZ*" "data_DoubleMu*" "data_DoubleEl*" "data_MuEG*" "Wpl*" "TT*" "Zjets*" "WW*" "*WH*" > do_higgs8TeV.txt
bash < do_higgs8TeV.txt

For this specific case, the ntuple configuration is named higgs_ntuples.py and makeQuad=1 makeTNP=1 makeH2Tau=0 makeTrilepton=1 make4L=1 rerunFSA=1 are VarParsing arguments. The wildcards after --samples describe which samples we want to run over.

Note that you need to run the 7TeV ntuples from a CMSSW 4_2_8_patch7 release, and 8 TeV from a CMSSW 5_3 release.