Calibration tests

Demonstrate that the calibration output of GW150914 for bilby and L.I. are consistent

changed milestone to %0.5.3

added 1 deleted label

@sylvia.biscoveanu @colm.talbot @cjhaster the nodes for the calibration are defined here: https://git.ligo.org/lscsoft/bilby/blob/master/bilby/gw/detector/calibration.py#L86

I think the choice of spline points is independent of the way bilby defined frequencies. To make this the same as L.I. can't we just change this line?

changed milestone to %0.5.4

changed milestone to %0.5.5

changed milestone to %1.0.0

Closing this as all updates are now on https://git.ligo.org/lscsoft/bilby_pipe/wikis/O3-review/calibration

closed

reopened

There are two outstanding questions for the calibration:

1. Should we change the default spline nodes in bilby_pipe to match those in LALInference?
2. Is the difference between the bilby and LALInference posteriors passable or is there more investigation needed before this can pass a review? See the results linked in the second half of the "GW150914: Sylvia's Work" heading here https://git.ligo.org/lscsoft/bilby_pipe/wikis/O3-review/calibration @cjhaster @gregory.ashton

Noting again that the difference seen in the latest bilby posterior is consistent with using neither reflective nor periodic boundaries for the calibration parameters.

I am happy to change the default priors to use one of those boundary conditions and try running again before sending this off to review if people think that's the way forward.

I would recommend trying that, this will involve changing this line https://git.ligo.org/lscsoft/bilby/blob/master/bilby/gw/prior.py#L556, and similar.

added 1 deleted label

Okay, so the action item is to run with "reflective" boundary conditions and see what the difference in posteriors is?

If they end up being consistent, then we can arrange a quick call with Matt and Simon to discuss.

If they end up being different still.. it isn't clean to me what the path forward is.

Before I do that, can we come to a consensus on the first question? Should I submit a MR in bilby_pipe so that the spline nodes are in the same place as LALInference? Or does that not matter?

I vote for not changing the spline nodes at this stage. They're only different by at most 1/duration, right?

I agree with Colm here. Until we have a sensible reason to change them (they are "fundamentally wrong" or they cause a severe level of difference with L.I) I'd leave it as is.

I'm kind of tied up with other things at the moment so I'm going to leave it to others to push getting this seen to by Matt/Simon at somepoint.

closed

unassigned @sylvia.biscoveanu