Function to compute WCS for a given FITS HIPS tile

[cdeil]

@tboch - I see you gave an example of a FITS WCS for a given FITS HIPS tile: https://hips.readthedocs.io/en/latest/drawing_algo.html#wcs-for-fits-tiles

Can you describe how it could be computed for any tile?

Maybe @adl1995 (or you if it's too complex) could add a function (e.g. in hips/utils.py or a new module) that returns such an astropy.wcs.WCS for given HiPS tile parameters?

I think that would be valuable to have, both for debugging and precise tile drawing (e.g. reproject just calls into wcslib via astropy.wcs.WCS, so probably reprojecting a tile with WCS into any other sky image will just work with one call into reproject (and they have the slow flux conserving method also implemented, which is nice).

[tboch]

My colleague François-Xavier provided me with a Java executable to compute this WCS. I know he planned to release his HEALPix-related code under a LGPL licence, I will check if he can already provide me the source code.

[cdeil]

Moving to v0.2 milestone, this isn't urgent. Still, @tboch - any news?

[tboch]

I have the Java source code available. I'll provide a Python version by end of this week. Where should I put this?

[cdeil]

If you have time, you could add this directly to the hips package. Maybe a property or method on https://hips.readthedocs.io/en/latest/api/hips.HipsTileMeta.html would be a good location?

If you want @adl1995 to add it (and add tests / docstring), then you could just put the code you have in a https://gist.github.com/ and then @adl1995 will make a PR adding everything.

[cdeil]

@tboch - Coming back to this: is it really possible to have a WCS for a HiPS tile?

If yes, I think this would be very nice. E.g. one could just use http://reproject.readthedocs.io/ to get a precise tile painting or tile generation if going the other way. And also for debugging, it's very nice to be able to add a WCS and e.g. open a tile in DS9 or check world and pixel coords from Python.

[tboch]

@cdeil : yes, it's possible, I have some working code I still have to properly validate. Should I create a different branch for this?

[cdeil]

Yes, always work in a git branch.

Probably it's a function that makes a WCS or a Header object? Could go here: https://github.com/hipspy/hips/tree/master/hips/utils

If you have an implementation, validating shouldn't be too hard. Pick one or two example tiles and a few pixels in the tile and check their sky position from the WCS against the known sky position from HEALPix.

I don't have much time, but could also try to go back to hipspy a bit in the next weeks and do some things.

[cdeil]

@tboch - Do you have code for a WCS for the HiPS tile? Or should we move this from the v0.3 milestone (in the next weeks) to v1.0 (aka some time in the future)?

[tboch]

thanks to the work of @fxpineau and @bmatthieu3 , I have now a working and validated code (tested with Montage and reproject) to create a WCS for a given HiPS tile: https://gist.github.com/tboch/636913974763f462fb4b51e61e76ee67 It requires cdshealpix package.

Still have to write proper tests, and it will be ready for inclusion in the hips package

[cdeil]

🎉