Data with wavelength decreasing along dispersion axis

[StuartLittlefair]

Hi,

I'm trying to reduce some old (2001) WHT/ISIS Red arm data using the TEK detector using ASPIRED.

For this dataset, wavelength decreases with increasing pixel number. From my arc, I have manually identified the following pixel, wavelength pairs:

pix = [67, 165, 365, 531, 598, 860, 908]
waves = [
    6717.08 , 6678.2  , 6598.95 , 
    6532.88 , 6506.53 , 6402.247, 6382.99]

I can get an excellent 4th order polynomial fit to these pairs (RMS ~ 0.07AA).

Glancing through the code, it seems that the assumption of wavelength increasing with pixel number is an essential part of rascal.

Do you have any suggestions for how to deal with this data?

[jveitchmichaelis]

Hi Stuart

Thanks for giving rascal a go! Yes, currently we assume this. Theoretically there's no reason we couldn't support it, it'd just need swapping some signs in a few places and checking for monotonic decreasing.

The way we currently suggest users deal with different dispersion directions is to use a higher level library like ASPIRED (@cylammarco do we have a example?) I believe we just flip signs of the inputs and then flip back at the end, but Marco can confirm.

I'll have a look at how much work it'd be to add this an option though, but it might have to wait for a future release.

Cheers
Josh

[StuartLittlefair]

Thanks - in the meantime I can just flip the dispersion axis around before reducing, but it feels a bit unsatisfying...

[cylammarco]

Hi Stuart, I have an example of WHT/ISIS reduction at https://github.com/cylammarco/ASPIRED-example/tree/master/example_WHT_ISIS. The notebook is too large to open on github directly so here is the viewer: https://nbviewer.org/github/cylammarco/ASPIRED-example/blob/master/example_WHT_ISIS/ISIS_PSO1801p6254_G93m48.ipynb.

To account for the rotation, flipping or both, the TwoDSpec can be set with:

https://aspired.readthedocs.io/en/latest/modules/twodspec_api.html?highlight=flip#aspired.spectral_reduction.TwoDSpec.set_properties

If you provide the arc from an ImageReduction object, the apply_mask_to_arc() method would be applied automatically. However, if the arc is added independently from the spectral frame after the initialisation step, the apply_mask_to_arc() has to be applied manually.

P.S. I am still trying to work out a more user-friendly decision tree to flip and rotate an arc frame. I am hoping to have it improved in v0.5.

[StuartLittlefair]

Great - thanks!