This PR adds new code (tools/analysis_slurm.py) to generate a slurm script (named slurm.sub by default) to run lightcurve-analysis on an HPC resource. The code parses all current arguments in em/analysis.py along with slurm-specific arguments in order to allow maximum customizability.
The idea is that this code could be run one time on an HPC resource before being queued (e.g. via API calls) many times for different models/sources. Thus, a few of the defaults for analysis.py arguments are changed. These include --model, --label, --trigger-time, and --data, which all take environment variables by default ($MODEL, $LABEL, $TT, $DATA). These arguments can be customized upon script submission by exporting the appropriate environment variables in the call to sbatch, e.g:
This PR adds new code (
tools/analysis_slurm.py
) to generate a slurm script (namedslurm.sub
by default) to runlightcurve-analysis
on an HPC resource. The code parses all current arguments inem/analysis.py
along with slurm-specific arguments in order to allow maximum customizability.The idea is that this code could be run one time on an HPC resource before being queued (e.g. via API calls) many times for different models/sources. Thus, a few of the defaults for
analysis.py
arguments are changed. These include--model
,--label
,--trigger-time
, and--data
, which all take environment variables by default ($MODEL
,$LABEL
,$TT
,$DATA
). These arguments can be customized upon script submission by exporting the appropriate environment variables in the call tosbatch
, e.g:Also, the default prior is
priors/$MODEL.prior
.