SUSY VBF

Python package for analyzing compressed mass spectrum SUSY via electroweak VBF with 0-, 1-, and 2-lepton final states. The package uses a columnar framework to process input tree-based NanoAODv12 files using Coffea and scikit-hep Python libraries.

• Processors
• Generate input datasets
• Submit Condor jobs
• Postprocessing

Processors

• ztojets: Select events for a Z+jets 0-lepton channel

The processors are defined in analysis/processors/<processor>.py. The selections, variables, output histograms, triggers, among other features, are defined through a configuration file located in analysis/configs/processor/<processor>/<year>.yaml (see here for a detailed description).

Generate input datasets

Connect to lxplus and clone the repository (if you have not done it yet)

# connect to lxplus 
ssh <your_username>@lxplus.cern.ch

# clone the repository 
git clone https://github.com/deoache/susy_vbf.git
cd susy_vbf

You need to have a valid grid proxy in the CMS VO. (see here for details on how to register in the CMS VO). The needed grid proxy is obtained via the usual command

voms-proxy-init --voms cms

Use the make_filesets.py script to build the input filesets with xrootd endpoints:

# get the singularity shell 
singularity shell -B /afs -B /eos -B /cvmfs /cvmfs/unpacked.cern.ch/registry.hub.docker.com/coffeateam/coffea-dask:latest-py3.10

# move to the fileset directory
cd analysis/filesets/

# run the 'make_filesets' script
python make_filesets.py --year <year>

# exit the singularity
exit

Submit Condor jobs

Jobs are submitted via the submit_condor.py script:

usage: submit_condor.py [-h] [--processor PROCESSOR] [--dataset DATASET] [--year YEAR] [--flow FLOW] [--submit] [--label LABEL] [--eos] [--nfiles NFILES]
                        [--do_systematics]

optional arguments:
  -h, --help            show this help message and exit
  --processor PROCESSOR
                        processor to be used (default ztojets)
  --dataset DATASET     sample key to be processed
  --year YEAR           year of the data {2016preVFP, 2016postVFP, 2017, 2018} (default 2017)
  --flow FLOW           whether to include underflow/overflow to first/last bin {True, False} (default True)
  --submit              Enable Condor job submission. If not provided, it just builds condor files
  --label LABEL         Tag to label the run (default ztojets_CR)
  --eos                 Enable saving outputs to /eos
  --nfiles NFILES       number of root files to include in each dataset partition (default 20)
  --do_systematics      Enable applying systematics

Example:

python3 submit_condor.py --processor ztojets --dataset <sample> --year 2017 --label test --eos

Note: It's recommended to add the --eos flag so that the outputs are save in your /eos area, so that postprocessing can be done from SWAN. In this case, you will need to clone the repo also in SWAN before submitting jobs in order to be able to run the postprocess step afterwards.

The runner.py script is built on top of submit_condor.py and can be used to submit all jobs (MC + Data) for certain processor/year

usage: runner.py [-h] [--processor PROCESSOR] [--year YEAR] [--nfiles NFILES] [--label LABEL] [--submit] [--eos] [--do_systematics]

optional arguments:
  -h, --help            show this help message and exit
  --processor PROCESSOR
                        processor to be used {ztojets} (default ztojets)
  --year YEAR           dataset year {2016preVFP, 2016postVFP, 2017, 2018} (default 2017)
  --nfiles NFILES       number of root files to include in each dataset partition (default 20)
  --label LABEL         Tag to label the run (default ztojets_CR)
  --submit              Enable Condor job submission. If not provided, it just builds condor files
  --eos                 Enable saving outputs to /eos
  --do_systematics      Enable applying systematics

Example:

python3 runner.py --processor ztojets --year 2017 --label test --submit --eos 

After submitting the jobs you can watch their status by typing:

watch condor_q

You can use the resubmitter.py script to see which jobs have not yet been completed

usage: resubmitter.py [-h] [--processor PROCESSOR] [--year YEAR] [--label LABEL] [--resubmit] [--eos]

optional arguments:
  -h, --help            show this help message and exit
  --processor PROCESSOR
                        processor to be used
  --year YEAR           year of the data {2016preVFP, 2016postVFP, 2017, 2018} (default 2017)
  --label LABEL         label of the run
  --resubmit            if True resubmit the jobs. if False only print the missing jobs
  --eos                 Enable reading outputs from /eos

Example:

python3 resubmitter.py --processor ztojets --year 2017 --label test --eos 

Some jobs might crash due to some site being down. In this case, identify and remove the problematic site from the sites list, generate the datasets again with make_filesets.py, create new condor files with runner.py or submit_condor.py (without the --submit flag), and resubmit the missing jobs adding the --resubmit flag:

python3 resubmitter.py --processor ztojets --year 2017 --label test --eos --resubmit

Postprocessing

Once you have run the corresponding datasets for the processor, you can get the results using the run_postprocess.py script:

usage: run_postprocess.py [-h] [--processor PROCESSOR] [--year YEAR] [--label LABEL] [--eos] [--log_scale] [--yratio_limits YRATIO_LIMITS YRATIO_LIMITS]
                          [--output_dir OUTPUT_DIR] [--savefig] [--extension EXTENSION]

optional arguments:
  -h, --help            show this help message and exit
  --processor PROCESSOR
                        processor to be used {ztojets}
  --year YEAR           year of the data {2016preVFP, 2016postVFP, 2017, 2018} (default 2017)
  --label LABEL         Label of the output directory
  --eos                 Enable reading outputs from /eos
  --log_scale           Enable log scale for y-axis
  --yratio_limits YRATIO_LIMITS YRATIO_LIMITS
                        Set y-axis ratio limits as a tuple (e.g., --yratio_limits 0.5 1.5) (default 0 2)
  --output_dir OUTPUT_DIR
                        Path to the outputs directory (optional)
  --savefig             Enable plot saving
  --extension EXTENSION
                        extension to be used for plotting {png, pdf}

Example:

# from the susy_vbf folder in SWAN (105a release)
python3 run_postprocess.py --processor ztojets --year 2017 --label test --eos --log_scale --savefig

Results will be saved to the same directory as the output files