improve tweak script - Githubissues

tjlane / pypad

Intuitive, high-level interface to the CSPAD

Other

2 stars 1 forks source link

improve tweak script #33

Closed tjlane closed 11 years ago

tjlane commented 11 years ago

This turns out to be a good thing to have during beam times. Right now the interface is terrible. @rkirian if you have ideas let me know.

sellberg commented 11 years ago

@tjlane I tried optimizing the hexagonal Bragg peaks, and the rotation seems to screw things up. This system with 9 rings should clearly be overdetermined, so the PyPad should be able to get this right. It could be because the peaks are too speckle-y and not ring-like enough. I'm gonna try and apply the filters and/or disable the quad_rotation to see if the optimization is improved. Otherwise we may have to brainstorm tomorrow about a better algorithm... Partly this need may be covered by the tweak script, just to have something robust that always works.

screen shot 2013-07-27 at 16 42 53

screen shot 2013-07-27 at 17 21 30

screen shot 2013-07-27 at 17 23 44

tjlane commented 11 years ago

@sellberg gross. Definitely not so good.

Some things you might want to try immediately:

turn off quad rotations
there is a different optimization algo in pypad you can try by setting objective_type : peak_height in the prepgeom yaml
make a filter -- it could be that a few strong Bragg peaks are dominating the optimization

I agree this should work and the fact it doesn't is a flaw.

sellberg commented 11 years ago

@tjlane I'm gonna play around with it and see if I can improve it, definitely looks better without the quad rotations, but so far I can't say it's better than the manual tweaking Rick did during the beamtime. My worry is that if we can't ensure it improves stuff, one may just skip the optimization and go for manual tweaking instead....

PYPAD - NO ROTATIONS screen shot 2013-07-30 at 13 44 29

MANUAL TWEAKING ice-t229k-cxi64813-r0001-r0009