Closed beckermr closed 3 years ago
@esheldon There is a lot going here I don't follow.
Can you look through the code?
OK, I figured out the bias. The stack is shredding objects when the S/N is too high. I was using mag=14 in our units and seeing it not find 36 objects every time. This caused big non-linearities in the response to shear which breaks mdet.
I still think we need to track down the minus sign in R22. I'll make an issue.
This PR is ready for review!
This PR has a start for integration tests for shear recovery.