Closed schlafly closed 2 years ago
This PR now additionally removes any NaN entries in the locations of the stationary positioners. NaNs in the locations of the positions at which a correction is desired generate NaNs, which seems fair.
@julienguy , I'll plan to merge this tomorrow absent objections.
This PR adds an option to turbulence.correct_using_stationary so that outliers can be clipped from that analysis. Running on all images since 10/20 shows that typically 0 or 1 positioners are clipped. A small number of tiles with large turbulence have ~15 positioners clipped, and ~10 test images that have garbage positions have hundreds of clipped positioners.
This will make us more robust if one of our "stationary" positioners gets bumped in the night. Two positioners have been bumped at the 20-30 micron level since original "stationary" measurements on 10/10. This feature will prevent those from causing problems during the night on which they are bumped.