Developing image simulations with non-uniform shear

[CMacM]

The aim of this sprint is to develop a code which allows non-affine shear transformations to be applied to simulated galaxy images.

Ideally, it will be easily integrated with galsim, so it can be available for general use. Non-affine shear transforms can be used to represent flexion, non-affine WCS transforms, and intrinsic alignment shear as a function of galaxy radius, making this a potentially useful tool for various groups within the DESC. Towards this end, the code should also allow for the user to pass their own non-affine shear functions.

If these goals are addressed before the end of the week, we will look to begin work on creating simulations and validating they are fit for purpose, which is initially to quantify the potential for shear estimators with different radial weightings to extract IA information.

Deliverables:

• [ ] Get GitHub repo more structured and set up
• [ ] Implement IA transform function
• [ ] Implement flexion transform function
• [ ] Implement optical field distortion transforms (post PSF)
• [ ] Test implementations with small-scale simulations
• [ ] Optimise and structure code (stretch)
• [ ] Push and merge final code (stretch)
• [ ] Begin developing larger simulations for testing FPFS radial weighting (stretch)
• [ ] Implement passing of custom transform functions (stretch goal)

Contacts: @CMacM Day/Time: All week Main communication channel: Slack #desc-batsim-sprint GitHub repo: https://github.com/CMacM/BATSim Zoom room (if applicable): Can set up if remote participants are interested

[sidneymau]

Just a quick comment—I briefly thought about this during last year's sprint week (slide here: https://docs.google.com/presentation/d/1eeC8j4H_u5S0c8y-tr3y_fM3hd3-zplrmDaPnXjTh4c/edit#slide=id.g16f5d0ddd52_44_0). I didn't figure out how to do the lensing correctly (only the inverse transformation which is of course much easier). I'm not sure I have too much insight, but I can at least point to a few places where this has been discussed in the past:

• https://github.com/GalSim-developers/GalSim/issues/361
• https://github.com/LSSTDESC/SprintWeek2022/issues/4

[CMacM]

Just a quick comment—I briefly thought about this during last year's sprint week (slide here: https://docs.google.com/presentation/d/1eeC8j4H_u5S0c8y-tr3y_fM3hd3-zplrmDaPnXjTh4c/edit#slide=id.g16f5d0ddd52_44_0). I didn't figure out how to do the lensing correctly (only the inverse transformation which is of course much easier). I'm not sure I have too much insight, but I can at least point to a few places where this has been discussed in the past:

• Support flexion GalSim-developers/GalSim#361
• (Un)recognized blend metrics in BTK for BLxPZ and BLxCL SprintWeek2022#4

Hi Sidney, thanks for sharing these past discussions and joining, any help you can offer would be welcome!

[c-d-leonard]

Update: I think Charlie and I have understood how the measurements galaxy-scale-dependent IA of Georigou et al and some notes are here.