pygamma
pygamma copied to clipboard
signed gamma evaluation
Is the current implementation correct? The determination of the sign should probably be calculated between the sample and the value closest to it in the reference under the distance footprint, rather than a straight diff of the sample and the reference?
Your gamma_evaluation is nice and compact, but somewhat difficult to understand. Can you put some more comments in?
In the JoMP paper, it looks like they are just determining the sign from the un-shifted points.
I have added a few comments to the code.