Skip to content

Latest commit

 

History

History
115 lines (77 loc) · 3.83 KB

README.md

File metadata and controls

115 lines (77 loc) · 3.83 KB

Coffea4bees analysis

To run the analysis, remember first to set the coffea environment and your grid certificate. If you followed the instructions in the README.md, the set_shell.sh file must be located right after the package, and then:

voms-proxy-init -rfc -voms cms --valid 168:00
cd coffea4bees/ ## Or go to this directory
source set_shell.sh

In this folder

Here you find the code to run the analysis from picoAODs and to make some plots. Each folder contains:

  • helpers: python files with funcions/classes generic to the analyses
  • metadata: yml files with the metadata for each analysis. In these files you can find input files, datasets, cross sections, etc.
  • processors: python files with the processors for each analysis.
  • pytorchModels: training models
  • weights: JCM txt files with weights
  • tests: python scripts for testing the code.
  • hists (optional): if you run the runner.py without a name of the output folder, this folder will be created to store the pickle files.

Then, the run-all script is called runner.py and it is one directory below (in python/). This script will run local or condor depending on the flag used. To learn all the options of the script, just run:

# (inside /coffea4bees/python/)
python runner.py --help

Run Analysis

Example to run the analysis

For example, to run a processor you can do:

#  (inside /coffea4bees/python/)
python runner.py -t -o test.coffea -d data TTToHadronic TTToSemiLeptonic TTTo2L2Nu  HH4b  -p analysis/processors/processor_example.py -y UL18  -op analysis/hists/ -c analysis/metadata/example.yml

The output file of this process will be test.coffea (a coffea output file), which contains many histograms and cutflows.

To produce some plots

Assuming that the file with your histograms is called hists/hists.coffea, you can run:

python analysis/makePlots.py hists/hists.coffea  -o testPlotsNew

To produce some plots interactively

python -i analysis/iPlot.py  hists/hists.coffea  -o testPlotsNew

Examples

>>> examples()

1D Examples

>>> plot("SvB_MA.ps_hh",doRatio=1, debug=True, region="SR",cut="passPreSel",rebin=1,rlim=[0,2],norm=1)
>>> plot("SvB_MA_ps_zh",cut="passPreSel",region="SB",doRatio=True,debug=True,ylabel="Entries",norm=False,legend=True,rebin=5,yscale='log')

2D Examples

>>> plot2d("quadJet_min_dr.lead_vs_subl_m",process="TTToHadronic",region=sum,cut="passPreSel")
>>> plot2d("quadJet_min_dr.lead_vs_subl_m",process="TTToHadronic",region=sum,cut="passPreSel",full=3)

To plot the same process from two different cuts

>>> plot("canJet0.pt", region="SR", cut=["passSvB","failSvB"],process="data")
>>> plot("canJet0.pt", region=["SB","SR"],cut="passSvB",process="data")

To plot different processes

>>> plot("v4j.mass", region="SR", cut="passPreSel",process="data",norm=1)
>>> plot("v4j.mass", region="SR", cut="passPreSel",process=["TTTo2L2Nu","data"],norm=1)

To plot the same process from two different inputs

> py  -i analysis/iPlot.py hists/histAll_file1.coffea hists/histAll_file1.coffea -l file1 file2

>>> plot("canJet0.pt",region="SR",cut="passPreSel",process="data")

To debug the code

If you want to debug small portions of the code, you can run it interactively in python by using some commands like:

from coffea.nanoevents import NanoEventsFactory, NanoAODSchema
fname = "root://xrootd-cms.infn.it//store/mc/RunIISummer20UL18NanoAODv9/DYJetsToLL_M-50_TuneCP5_13TeV-madgraphMLM-pythia8/NANOAODSIM/20UL18JMENano_106X_upgrade2018_realistic_v16_L1v1-v1/230000/9EEE27FD-7337-424F-9D7C-A5427A991D07.root"   #### or any nanoaod file
events = NanoEventsFactory.from_root( fname, schemaclass=NanoAODSchema.v6).events()