search

benjaminpope / drpangloss

the best of all possible interferometry models
https://benjaminpope.github.io/drpangloss/
MIT License
2 stars 0 forks source link

Dori vs New Absil Implementation #9

Closed benjaminpope closed 4 months ago

benjaminpope commented 4 months ago

I am a little concerned that the Absil formulation does not match the results in the original notebook by Dori, but it looks otherwise alright to me.

Dori's version: image

New version: (note the iffy bit, where the real binary is clearly visible) image

Dori's was suggesting that it did not detect a pretty clear injected signal, so I am not confident that that was right either.

IMO the issue fundamentally is that the Absil method is about establishing the contrast limits on a non-detection and we have an injected binary here, so we should rethink how we are doing & visualizing this.

benjaminpope commented 4 months ago

I think this is fixed in #14. Will leave open a few days in case people find more bugs.