Improve documentation for linop.optics
#233
Conversation
Codecov Report
@@ Coverage Diff @@
## main #233 +/- ##
==========================================
+ Coverage 93.83% 93.94% +0.10%
==========================================
Files 50 50
Lines 3681 3681
==========================================
+ Hits 3454 3458 +4
+ Misses 227 223 -4
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
| k0: Illumination wavenumber, :math:`k_0`, corresponding to | ||
| :math:`2 \pi` / wavelength. | ||
| z: Propagation distance, :math:`z`. | ||
| pad_factor: The padded input shape is the input shape |
There was a problem hiding this comment.
@lukepfister: Can you suggest some guidance to the user on selection of the padding factor?
There was a problem hiding this comment.
That's a complicated subject; it's the same as the choice of padding for DFT-based linear convolution but has extra physical interpretation. Would probably just leave a pointer to Chapter 5 of voelz.
There was a problem hiding this comment.
Better than nothing, I guess, but I was hoping to provide some guidance that didn't require the user to first do enough reading to be able to write their own implementation!
There was a problem hiding this comment.
🤷♂️ I think it is too complicated to get into the docs. Unlike a standard analysis of convolution in a signal processing class, the side lengths have physical meaning. Sometimes the size of your domain is just the size of your domain and any spatial aliasing is really indicative that you haven’t modeled the transmission channel properly (not accounting for reflections, etc)
there is a case to be made that those options should be removed entirely and the user should set up the domain side lengths to be larger.
Improve documentation for
linop.optics.