Martin Hardcastle

Yes. In fact maybe (i) is sufficient for the DDFacet test and (ii) is our internal ddf-pipeline unit test.

OK, so we see how Cyril gets on with subtraction.

Sure, sounds like a good idea to use Bootes. We will want this when the merge of Cyril's code with mainline DDF is done since there are features in that...

Thing is we want to test several things -- diffuse emission, bright sources, bad ionosphere... not sure one test dataset is going to do what we want -- which is...

So we kind of always expect the bootstrap to produce amplitude errors at some level -- the idea is that they should go away at the next amplitude self-cal, which...

Well, in a sense, that's the problem, not bootstrap (-: But I don't know why it's such a problem for this field...

Yes, we discovered this problem a little while ago. The old software still works. I assume it's incompatibility in the solutions files....

I'm leaving this open with the 'new version' flag because when we start using new DDF versions we're going to have to put a fix in. (This is I think...

Thanks, the thing I'd be worried about with this solution is the assumption that the beam is well characterized by a Gaussian, but I expect you're right about the nature...

Well that's a killms fail at the first dd stage... but presumably works OK in the old pipeline? or not?