DeconvolvePSF is an attempt at high-level PSF modeling for DES and LSST. The key idea for this software is that the PSF can be modeled as a convolution of an optics piece and an atmospheric piece. The optics piece in this work comes from the WavefrontPSF module by Chris Davis. This module finds the model of WavefrontPSF, deconvolves it from the observed stars with the Richardson-Lucy deconvolution algorithm, and then models the residuals with the PSFEx PSF modeler.
The module can be run locally on an exposure expid and stored in output_dir via
python afterburner.py <expid> <output_dir>
Alternatively it can be submitted to the SLAC batch cluster (if you are a SLAC employee logged into ki-ls) via:
python do_call.py <expid1> <expid2> ... <expidN>
or run on 40 random exposures simply via
python do_call.py
Please note the output directory in do_call is hardcoded in.
This software is distibuted open source under the MIT license.