On Point Spread Function modelling: towards optimal interpolation - 1 views
-
Joe Zuntz on 14 Oct 11Interpolation schemes for modelling the PSF, including some fun stuff like Gaussian Processes.
-
Phil Marshall on 14 Oct 11Now we need to get our paper out ASAP and start doing comparisons! :-) Re: "optimal" interpolation: I think they do a nice job of showing (and agreeing with us ;-) that you do best by having a very flexible, multi-scale model for the underlying maps. We also incorporate information about the statistics of the atmospheric PSF pattern in our prior; Kriging is more local, in that the covariance function is estimated only from the data in hand. Which makes me wonder what happens when the data sparseness goes up...