Telescope alignment errors / aberrations inferred from field dependence of moments of PSF on science array agree with those inferred from donut data, to within ??? waves of wavefront error. NOTE: the PSF is bandpass-dependent. This test should be done for r, i filters. g and z may be more problematic (g band PSF is not very good towards the edge of the focal plane; z band affected by CCD transparency).
Static LUT for alignment has been installed and is being applied. F/A donuts are being analyzed, but alignment tweaks need not yet be implemented. Results of F/A analysis are stored in database or equivalent. Focus loop is implemented and working. (One MIGHT be able to do this test with open-loop focus adjustments, but not sure.)
This test can be performed using science data alone. Science data in r and/or i bands must be taken in good seeing (FWHM < 1.2" ???). One computes moments of stars on all science CCDs and uses the patterns to derive a misalignment (including defocus). The numbers are compared with those derived from F/A donuts.
As a cross-check, we also want to obtain in-focus and out-of-focus images in the science array twice per SV night during evening and morning twilight to monitor the correspondence between the F/A data, PSF moments, and science CCD donuts.
The numbers agree at the same level that BCAM and donut data should agree (translation within 200 microns; tilts and defocus to some equivalent values).
This test is intended to determine what is possible using the pattern of PSF shapes. It is possible that one cannot derive a non-degenerate solution.