magnitudes aren't being converted properly

[jbudynk]

This issue was first discovered in PR ChandraCXC/iris#275. Changing the units from any flux/flux density to magnitudes (ABMAG or STMAG) converts the flux and flux errors incorrectly.

I've been checking setFluxUnits() in iris-common's SegmentStarTable class to see what's happening. My guess is we should be always going back to the original flux values when computing any unit changes, but what does Iris 2.1 do?

The SPV units code in Iris is the same as it is in Specview, so I don't think there's anything wrong with spv's YUnits conversion methods.

[jbudynk]

OK, a few things are at play with this bug:

• SPV's YUnits conversion method for AB magnitudes isn't perfect. At line 862, it should be

return -2.5 * Math.log10(arg) + ABZERO);

this is what was causing the flux values to change when the user converts back and forth between ABMAG and another unit multiple times.

• setFluxUnits() in iris-common's SegmentStarTable should be using units.convertErrors for the errors and not units.convertY, as there are rules for converting errors to/from magnitude units in the specview code.
• The order in which the setFlux... methods in SegmentStarTable.setFluxUnits() matters. If the flux values are converted first, then the wrong flux values are passed into convertErrors for converting the flux errors. The flux values should be converted after the errors are converted.
  • Or, we can change the code in SPV or SegmentStarTable so this issue doesn't arise.
• There is something funky going on with SPV's conversion from AB magnitude errors that I haven't found yet. I'm still looking for a solution here.