This scripts allows to easily design an analysis based on produced flat trees with the Heppy framework.
If you do not attempt to contribute to the FCChhAnalyses repository, simply clone it:
git clone git@github.com:FCC-hh-framework/FlatTreeAnalyzer.git
If you aim at contributing to the heppy repository, you need to fork and then clone the forked repository:
git clone git@github.com:YOURGITUSERNAME/FlatTreeAnalyzer.git
then initialize:
source ./init.sh
Analyses are run the following way:
./bin/analyze.py -n [analysis_name_in_heppy] -c [heppy_cfg] -t [heppy_tree_location] -o [output_dir] -p [analysis_parameters] -j [proc_dict]
./bin/analyze.py -n Zprime_ttbar -c ../heppy/FCChhAnalyses/HELHC/Zprime_ttbar/analysis.py -t /eos/experiment/fcc/helhc/analyses/Zprime_ttbar/heppy_outputs/helhc_v01/ -o /eos/experiment/fcc/helhc/analyses/Zprime_ttbar/FlatTreeAnalyzer_outputs/helhc_v01/ -p templates/HELHC/Zprime_ttbar.py -j /afs/cern.ch/work/h/helsens/public/FCCDicts/FCC_procDict_fcc_v02.json
To run with multi-threads, simply add "-m" to the execution line
It is also possible to send one lxbatch job per signal hypothesis. Each hypothesis will then run on a separate node (multi-threading on that node is allowed). For instance:
python bin/analyze.py \
-n tth_boosted \
-c ../../../FCCSW/heppy/FCChhAnalyses/tth_boosted/analysis.py \
-t /eos/user/s/selvaggi/heppyTrees/tth_boosted/ \
-p templates/FCC/test_lsf.py \
-j /afs/cern.ch/work/h/helsens/public/FCCDicts/FCC_procDict_fcc_v02.json \
-o test_lsf \
--nev 1000 \
--multi_threading \
--lsf \
--queue 8nh \
--force \
In order to collect jobs when running is over re-run the exact same command.
If some jobs has failed, the script will automatically re-submit them.
When all jobs are completed, they will be collected and stored in the output directory specified by the -o option,
and the yield tables and final plots will be produced.
After every plot is done, you can clean LSF junk with the --cleanlsf option.