Test function at the bottom of indexing.py

[jadball]

Following on from #239

Do you have an example where a double loop (N^2) gives a better answer than zip( npks, tols )? I had the double loop before and decided to get rid of it for some reason. The order of the loops was not clear for me. With box beam I let the error increase first (grain position effect) and with weak data (small crystals) the fewer spots is better than larger errors.

We could also look at a smooth function like exp(-drlv/hkl_tol) and only count once per h,k,l,sign(y) peak. The cutoff stops being so sharp.

Write a notebook that tests tolerance loops and a smooth function.

We should also think about writing tests/profiling from the simulated data:

• How many grains does it find that it should find?
• How many grains does it find that it shouldn't find?
• How long does it take to do that?

[jonwright]

Great! Maybe just use the notebook for plotting results?

A few years ago, I tried to set up a comparison of peak fitting here: https://gitlab.esrf.fr/wright/fitgolf

It has a folder of functions and a folder of testcases and it runs each function on each testcase. We could start a game of indexing code-golf instead...