Plots configuration for mkShapes, mkPlot, mkDatacards
First, setup the LatinoAnalysis framework:
cmsrel CMSSW_10_6_4
cd CMSSW_10_6_4/src/
cmsenv
git clone --branch 13TeV git@github.com:latinos/setup.git LatinosSetup
source LatinosSetup/SetupShapeOnly.sh
scram b -j4
Download the PlotsConfigurations package anywhere, but remember to do 'cmsenv' of the CMSSW release you are using:
git clone git@github.com:latinos/PlotsConfigurations.git
Make a copy and edit the following python file (userConfig.py) to specify your base directory, i.e. the directory in which your job related information will be stored:
cd LatinoAnalysis/Tools/python/
cp userConfig_TEMPLATE.py userConfig.py
cd -
Prepare your configuration, you can use the following configuration as a template (this is based on 2018 data and MC):
https://github.com/latinos/PlotsConfigurations/tree/master/Configurations/ggH/Full2018
Produce the histograms submitting batch jobs using HTCondor:
mkShapesMulti.py --pycfg=configuration.py --doBatch=1 --batchSplit=Samples,Files --batchQueue=longlunch
You can choose one of the following queues according to your job expected running time:
espresso = 20 minutes
microcentury = 1 hour
longlunch = 2 hours
workday = 8 hours
tomorrow = 1 day
testmatch = 3 days
nextweek = 1 week
If some of your jobs have failed because the wall clock time have been exceeded, you can resubmit the failed ones by going into the jobs directory (the one set in userConfig.py), and changing the queue using the following command:
for i in *jid; do sed -i "s/longlunch/workday/g" ${i/jid/jds}; condor_submit ${i/jid/jds}; done
For quick tests you can run interactively by just typing mkShapesMulti.py --pycfg=configuration.py. Use mkShapesMulty.py --help for more options.
You can also run interactively but submitting jobs in parallel with the command mkShapesMulti.py --pycfg=configuration.py --doThreads=True.
Once all your jobs are done (you can check job status with condor_q), you will find a rootFile directory in your area containing all the histograms specified in your configuration.
You can proceed by h-adding all the files to get a single one containing everything:
mkShapesMulti.py --pycfg=configuration.py --doHadd=1 --batchSplit=Samples,Files --doNotCleanup --nThreads=10
The --nThreads=10 option allows for running the Hadd step in multithreading mode (with 10 threads in this case), and is especially useful when your configuration contains many cuts and variables.
The --doNotCleanup option is used to keep the input root files. Without this option the input files will be deleted after the hadd step and only the final root file will be kept.
You can now proceed making plots (mkPlot.py --help to see all available options):
mkPlot.py --pycfg configuration.py --inputFile rootFile/plots_TAG.root --showIntegralLegend 1
and datacards (mkDatacards.py --help to see all available options):
mkDatacards.py --pycfg configuration.py --inputFile rootFile/plots_TAG.root
If you need yield tables in either .tex or .csv format, after running the FitDiagnostics method of Combine on a workspace obtained from the datacards:
combine -M FitDiagnostics -d WORKSPACE.root --saveNormalizations --saveWithUncertainties
you can feed the output to mkTable (mkTable.py --help to see all available options):
mkTable.py fitDiagnostics.root
in case you want to merge a set of categories and/or processes in the table, you can define a merging scheme. A template, as well as more detailed instructions on mkTable, can be found in LatinoAnalysis/ShapeAnalysis/data/.
Congratulations! You have done the analysis, or at least a very first step...
-
For fits and combine general info check the following:
-
The current Latinos framework is based on the NanoAOD format. The list of available
nanoLatinotrees can be found here: -
The NanoAOD-nanoLatino MC sample mapping is defined in the following python dictionaries: