Change the default bilby_pipe 128s pp-tests to highSpin

changed milestone to %0.5.3

added 1 deleted label

I ran a quick test with the current master version and increasing the max spin to 0.8. Unfortunately, the pp tests are not passing

Here is the run with a standard settings:

https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/bilby_review/test/outdir_pp_test_128s_ROQ/pp.png

Here I increased the nlive to 2000:

https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/bilby_review/test/outdir_pp_test_128s_ROQ_HACK/pp.png

If I recall, the L.I BNS pp tests are run on the reduced spin settings which is why I originally implemented it this way. @cjhaster is this correct or did I misunderstand something?

One obvious thing to try is doing the pp tests on combined runs (by default we combine 4 runs, but for pp tests we only do a single run). I could try the combined runs or cranking the other settings til it passes.

@colm.talbot @matthew-pitkin @simon-stevenson

• Try increasing the number of steps

changed milestone to %0.5.4

Upping the number of walks to 200 seems to have improved things with a corresponding increase in the run time. This is a comparison of the 128s ROQ PP test and samping time distributions for 100 and 200 walks, there where other minor code changes as well, but I'm confident these don't change the result.

Nlive=1000, walks=100	Nlive=1000, walks=200

Roughly speaking, this doubles the run times.

On the review call, @colm.talbot suggested halving the number of walks might recoup the time cost. Unfortunately this is not the case, with 500 live points it fails the PP test

Nlive=500, walks=200

mentioned in merge request !564 (merged)

Closing as this is now resolved by !564 (merged)

closed

unassigned @gregory.ashton