Heppy (High Energy Physics with PYthon) is a modular python framework for the analysis of collision events.
If you're not very familiar with python yet, you will probably find the Python Tutorial useful before you get started with heppy.
Table of contents: master
- Installation
- Introduction
- A very simple example
- Parallel processing: running jobs
- Full analysis workflows
- Reference guide
- Generic analyses: working in several experiments
- Papas, the parametrized particle simulation
Support & feedback: https://github.com/cbernet
submit example :
heppy_batch.py -o Outdir FCChhAnalyses/FCChh/tttt/analysis.py -b 'run_condor.sh --bulk Outdir -f microcentury' --nevent 1000
In this example, CONDOR will look at all directories (could be Chunk too) in Outdir
(--bulk Outdir
) and run jobs for all of them into a single job. For example here, 10 jobs are coming from FCChhAnalyses/FCChh/tttt/analysis.py
. And each job will be run on 1000 evenmts.
[djamin@lxplus037 heppy]$ condor_q
-- Schedd: bigbird09.cern.ch : <188.185.71.142:9618?... @ 03/05/19 15:13:52
OWNER BATCH_NAME SUBMITTED DONE RUN IDLE TOTAL JOB_IDS
djamin CMD: batchScri 3/5 15:05 4 6 _ 10 594302.1-9
run_condor.sh has been added in the new script/ directory
Instead of flavour (-f
), it is possible to use maxruntime
(unit = minute) : -t 60
Predefined timing jobs are done from flavour :
20 mins -> espresso 1h -> microcentury 2h -> longlunch 8h -> workday 1d -> tomorrow 3d -> testmatch 1w -> nextweek
If job fails, can resubmit each failed job with :
heppy_check.py Outdir/*Chunk* -b 'run_condor.sh -f microcentury'
FCC actually have their own quota. To use it, you need to get yourself added to the egroup:
fcc-experiments-comp
Then you can add the following to your submit file:
+AccountingGroup = "group_u_FCC.local_gen"