Performance issue with ROQ
I ran some tests on the lastest master branch of bilby using the ROQ. Something has changes and per likelihood evaluations are taking something like 2 orders of magnitude more than they did previously. Does anyone have any thoughts? I think they two culprits in my mind are !552 (merged) and !536 (merged).
Here is an example log file: https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/bilby_review/test/outdir_pp_test_128s_ROQ/log_data_analysis/pp_test_128s_ROQ_190706_2359_H1L1_dynesty_injection0_0_31630192_0.err Note the single likelihood evaluation is 2.037e-01 s (when it should typically be ~1e-3s)