Possible MIRI/MRS world 2 detector bug.

[baileyji]

Hi Folks,
I'm seeing very unexpected behavior in the world to detector transform as captured in this gist (note you will need to contact me for the supporting fits files).

The short of it is that computing a set of x,y pixel coordinates from a mesh of ra, dec, and lambda yields regions on the detector that are outside (If I'm using the code properly) of the slices on on the detector.

https://gist.github.com/baileyji/1a1c4762bbc303f9f0a3a10503a5da64

[nden]

After offline discussions, I think there's no bug. The reason for seeing values between slices is that the regions mask is not applied to output values of the inverse transform. I'm closing this but two new issues cam out of the discussions - #1427 and #1428

[nden]

After much back and forth I think I understand the issue now.
When going from alpha, beta to detector, the beta value alone is used to determine which slice the (alpha, beta) value belongs to. Instead, it is necessary to also look at the range of alpha values for each slice.
Here's an example for a 12SHORT exposure:

alpha, beta = ( 3.19529725, 2.0776295)

Based on beta only this point belongs to slice 216. However, the alpha range for slice 216 is
(-2.17462231, 2.13811708). Based on the alpha value this point should have a value of NaN on the detector, while now it goes through the "alpha_beta to detector" transform and has a real value on the detector. Looking at the pixel location on the detector and using the forward transform, it actually belongs to slice 207 (another indication it's wrong).

[hbushouse]

Can this be retested to see if it's still a problem in B7.8?