tttt - Four Top Analysis

Run 2 Repo

https://github.com/cmstas/FTAnalysis

Setup

First, run the check out script to get all dependencies, i.e.,

wget https://raw.githubusercontent.com/cmstas/tttt/[YOUR FAVORITE BRANCH]/checkout.sh
./checkout.sh

Then, you need to go into the src subdirectory of the CMSSW directory that is set up based on the checkout script, and run

cmsenv
cd tttt
./setup.sh -j [Ncores] # Ncores is optional. Unless you are running on a Condor node, you can keep it blank.
eval $(./setup.sh env)
# ./setup.sh -j is only run to compile, so it is only run once.
# The eval statement is for environment variables,
# so it should be run at every log in.

How to submit jobs using Metis to produce skims

In a screen, run the following:

cd ../ProjectMetis
. setup.sh
cd -
cd test
createFTAnalysisTarball.sh # Creates ftanalysis.tar in the current directory

This will make a tarball to be uploaded to Condor. You can then run the skim submission script as follows:

python submit_skims.py --tarfile ftanalysis.tar --tag [TAG] [CSV FILES]

Example executable to test and try

Assuming you have skims available (after running skims using the instructions above), you can run the test looper by typing a command that looks like

test_looper dataset=/TTTT_TuneCP5_13TeV-amcatnlo-pythia8/RunIISummer20UL18NanoAODv9-106X_upgrade2018_realistic_v16_L1v1-v2/NANOAODSIM short_name=TTTT period=2018 xsec=0.009103 BR=1 input_tag=[INPUT TAG] output_tag=[OUTPUT TAG]

Interactive runs can be exited gracefully using Ctrl+C with some form of output. Condor jobs disable this graceful exit functionality.

How to submit jobs to run on the skims

Standalone scripts are used to submit these jobs. The first step is to run configureFTAnalysisJobs.py, e.g.,

configureFTAnalysisJobs.py --exe analyze_cutbased --nthreads=4 --nevts 500000 \
--tag 220920_TopLeptonMVA \
output_tag=220920_TopLeptonMVA input_tag=220705 muon_id=TopMVA_Run2 electron_id=TopMVA_Run2 \
test/samples_MC.csv

for the MC, or

configureFTAnalysisJobs.py --exe analyze_cutbased --nthreads=4 --nevts 500000 \
--tag 220920_TopLeptonMVA --group_eras \
output_tag=220920_TopLeptonMVA input_tag=220705 muon_id=TopMVA_Run2 electron_id=TopMVA_Run2 \
test/samples_Data.csv

for the data (type --help to see which arguments are optional).

Notice that arguments that are preceded by '--' are arguments to modify how the python script runs, and the other arguments control how the executable ("analyze_cutbased" in this example) should run. The last argument in this example (but it doesn't have to be the last) is the csv file to acquire the data sets. The arguments to the executable also play the purpose of filtering the data sets if you specify any entry already in the csv (e.g., 'short_name=TT_2l2nu'), or 'output_file', which is different from 'short_name' only if real data processing is grouped into eras.

After you run this first step, all submission folders are created, but the jobs are not yet submitted. In order to submit them, following the examples above, you need to run

watchFTAnalysisJobs.sh test/output/tasks/220920_TopLeptonMVA
# Add sleep=N in seconds to modify sleep duration, email=[EMAIL] for watch to notify you when all jobs are done.

This script will trigger the checking and resubmission of jobs until all jobs in the folder you specified are completed. Internally, this script runs checkFTAnalysisJobs.sh and resubmitFTAnalysisJobs.sh, and sleeps for a while before checking again. When all jobs are done, the watch will terminate. Make sure to run the watch in a screen! Otherwise, you terminal will be kept busy.