Efficient PESummary implementation
We should use PESummary for the post processing of parameter estimation results. On the other hand, running full PESummary takes ~hours, and could delay the circulation of parameter estimation updates. There are two options I can think of.
- Run PESummary with options to accelerate it as a part of parameter estimation condor jobs, upload the outputs to GraceDB, and generate skymaps and EM-bright probabilities from them.
- Implement PESummary task into GWCelery, and run it on uploaded samples in parallel with ligo-skymap and embright.
There are no latencies from PESummary in the second option, but the file format of uploaded samples could depend on the PE pipeline.
cc: @charlie.hoy