NEW: Implement Flow operator for CuPy

carterbox commented 4 years ago

Purpose

Allow CuPy backend with align module by implementing the Flow operator.

Approach

Switch to Lanczos interpolation for Flow operator because map_coordinates for CuPy only has 1st order spline interpolation.

Pre-Merge Checklists

Submitter

[x] Write a helpfully descriptive pull request title.
[x] Organize changes into logically grouped commits with descriptive commit messages.
[x] Document all new functions.
[x] Write tests for new functions or explain why they are not needed.
[ ] Build the documentation successfully
[x] Use yapf to format python code.

Reviewer

[ ] Actually read all of the code.
[ ] Run the new code yourself.
[ ] Write a summary of the changes as you understand them.
[ ] Thank the submitter.

dgursoy commented 4 years ago

Do you know the performance (speed) difference between Lanczos and the 1st order interpolation? Because we solve those problems in ADMM approximately anyways, and an approximate (but fast) interpolator may work too.

carterbox commented 4 years ago

I don't know. It's probably more expensive because of needing to call sinc() in the filter.

AdvancedPhotonSource / tike