Centroid shifts from lensing

[rmjarvis]

In addition to shear and convergence/magnification, lensing also moves the positions of objects around. We should have our lensing modules also be able to return a shift that is appropriate for the location of the galaxy. Then the user could apply that as well as the shear and magnification to get the complete lensing effect.

[reikonakajima]

Shifts like in this paper? http://arxiv.org/pdf/1405.1432.pdf :)

[pmelchior]

I can implement the deflection angles for the NFW halo, so that one gets the new positions by iteratively searching for the observed positions given the pre-lensing positions.

[rmandelb]

What would that look like in terms of how the user interacts with the code? Right now they have to enter post-lensing positions. Is the idea that they would now be told to enter pre-lensing positions, which get shifted and the appropriate shear/magnifications applied?

[pmelchior]

I think there are two ways about that:

1. We stick to post-lensing positions, then a new method, say getPosition() would return their pre-lensing positions. That's the easiest way, but not really useful because it doesn't solve the problem that the post-lensing positions do not reflect that presence of a halo (or any sort of matter distribution).
2. We change to pre-lensing positions, then a new wrapper method would compute the post-lensing ones and call the existing shear/convergence methods with it. A second function call to getPosition() would return these coordinates. That's the right way of doing it, but it is costly computationally since there's an iterative solver involved for each (x/y) transformation.

[rmjarvis]

I favor the latter option. The typical use case in which one would care about these shifts is to have a uniform density of objects in the pre-lensing plane, and then want them plotted in the correctly lensed positions. So even though it is more work on our part, I think it's what the user will expect.

As for the UI, I propose the following:

getLensedPositions(prelensing_positions) # New function
getPreLensedPositions(lensed_positions) # New function
getShear(..., prelensed_positions=False) # New kwarg with default = current behavior
... likewise for getConvergence, etc.

[rmandelb]

So even though it is more work on our part, I think it's what the user will expect.

+1.