ratt-ru/CubiCal

need to honour no_beam tag for LSM sources

o-smirnov opened this issue · 4 comments

The tigger LSM format allows for a no_beam tag on sources, which is an indicator that the source is given in apparent flux units, and no beam should be applied. This is not currently honoured by Montblanc. A quick kludge may be required. @JSKenyon @ulricharmel @sjperkins

If my memory serves, Montblanc doesn't really have a mechanism for doing different things per source. I think the only way around this will be be to cluster each direction into a beam required group and no beam group. Each group will need to be predicted separately (much like the directions), one with the FitsBeamSourceProvider and one without. The results would need to be accumulated into the relevant direction.

Just so that this is documented somewhere:

nobeamcomp

Images made with identical parameters. G at (1,1), dE at (64, 32). Left image is current master (nobeam not honoured), right image is with the issue-431 PR (nobeam honoured). Performance was virtually identical (PR was actually slightly faster).

@ulricharmel you can try the issue-431 branch when you have a moment. May interest you too @o-smirnov!

@JSKenyon sure I will

@JSKenyon I tested it on the C-MS. I think your patch is doing the right. I get similar and expected results.