Correct CTFFilter formula/units

[garrettwrong]

Looks like we had some errors in our CTFFilter formula. Because this is mainly tested tautologically, multiple errors slipped by.

To confirm error wrt MATLAB and test I performed the following:

• Created a simulation data set with CTF in Python
• Saved the dataset (as a STARfile)
• Read the STARfile in MATLAB
• Ran MATLABs phaseflip
• Loaded the MATLAB result mrcs back into Python
• Compared Python phase_flip with the MATLAB result
• Compared loading the STAR file in Python then phase_flipping and compare with MATLAB result

I am considering adding the hard coded reference test of CTFFilter.evaluate_grid (say at 5px) so this is less likely to happen again. While I prefer less hard coded tests, clearly testing our formula against our own formula is not enough 😇 . Can discuss other options at our next dev meeting.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 90.66%. Comparing base (a29670b) to head (a890725).
Report is 2 commits behind head on develop.

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1217   +/-   ##
========================================
  Coverage    90.66%   90.66%           
========================================
  Files          132      132           
  Lines        13706    13707    +1     
========================================
+ Hits         12427    12428    +1     
  Misses        1279     1279           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[j-c-c]

Looks good!

[garrettwrong]

Merging patch.

[janden]

I am considering adding the hard coded reference test of CTFFilter.evaluate_grid (say at 5px) so this is less likely to happen again. While I prefer less hard coded tests, clearly testing our formula against our own formula is not enough 😇 . Can discuss other options at our next dev meeting.

I think this is fine. Another reference we might want to consider is RELION. We can use relion_project to generate a projection (with and without CTF). This won't be as tight of course (considering projection is done slightly differently), but I'd consider it closer to a gold standard than Matlab ASPIRE.