Skip to content

Latest commit

 

History

History
34 lines (24 loc) · 1.62 KB

README.md

File metadata and controls

34 lines (24 loc) · 1.62 KB
Raw

Lauren's Roman forced photometry pipeline.

Runs on Troxel et al. RomanDESCSims simulated Roman WFI images. Currently developed and tested at Duke Computing Cluster and Perlmutter at NERSC.

Depends on

phrosty
roman_imsim

Runs with

  1. get_object_instances.py to get all the images by filter, pointing, and SCA that contain the specified transient.
  2. preprocess.py to create references, get PSFs, and do cross-convolutions.
  3. sfftdiff.py to run SFFT subtraction on a GPU.
  4. postprocess.py to generate decorrelation kernels, apply those to images, and make stamps.
  5. mk_lc.py to extract a lightcurve from the subtractions.

sfft_and_animate_*

sfft_and_animate_*.sh as a SLURM array job that has SN ID hardcoded.
This then calls get_object_instances, preprocess, sfftdiff, postprocess, and mk_lc with the SN ID (refered to as oid in the code).
sfft_and_animate_*.py reads the SLURM_ARRAY_JOBID from the environment and uses that to select the band to process. (1-7).

sfft_and_animate_*.py reads input data from input data as defined by phrosty.utils._build_filepath, which uses environment variableSIMS_DIR to look for the data. SIMS_DIR should match where the RomanDESCSims are.

preprocess.py, sfftdiff.py, and postprocess.py write files out to environment variable DIA_OUT_DIR.

mk_lc.py writes files out to environment variable LC_OUT_DIR.

Example usage at the DCC Run on SNID 20172782 (yes, the SN IDs are 8-digit numbers beginning with 20, they are not dates), filter R062, with specified input directory.

python mk_lc.py 20172782 --band R062 --inputdir /work/lna18/