force there to be at least three elements in time reconstruction

This should hopefully resolve issues like

...bilby/gw/likelihood.py", line 403, in generate_time_sample_from_marginalized_likelihood
    "Time posterior reconstruction failed, at least two samples "               
bilby.core.likelihood.MarginalizedLikelihoodReconstructionError: Time posterior reconstruction failed, at least two samples are required.

This should be tested more.

added 1 commit

• 63a12bb1 - remove unused import

Compare with previous version

@colm.talbot I did a quick run for 190425 and it managed to reconstruct the posterior. Is there any reason to not trust the posterior? If not, I'd suggest we merge this and start testing it.

Only the issue with the reference time which we have discussed before.

But that issue is already in current bilby right? So this just means it will produce samples, but they may be off?

Yeah

I did a run for GW190425 using pbilby and this branch. It successfully reconstructed a time posterior that is at least consistent. Compare the IMRPhenomDNRT_LS_PB with the others (ignore the SEOB result) here:

approved this merge request

changed milestone to %0.6.5

added 1 deleted label

approved this merge request

mentioned in commit 6ee08a83

merged

mentioned in issue #450 (closed)