Lang and I pair-coded (although really Lang coded and I stared at the terminal) two extensions for The Tractor today, both related to our hierarchical weak lensing project (now called MBI) with Bard and Marshall and Schneider and Dawson and others. In the first, we extended our mixture of Gaussians approximations to standard galaxy profile to general Sérsic indices; this requires some optimizations, running now in the background. In the second, we built something (that Lang wants to call "Trogdor the Burninator") that very reliably initializes emcee for industrial-scale sampling in galaxy morphological parameters for millions of galaxies. We start with an optimization. A third issue, which we did not address, is that we have four qualitatively different galaxy models (exp, dev, composite exp+dev, and Sérsic) and we would like, ideally, to sample in all four, but this is not totally trivial, given that either we have to make reversible jumps or else compute fully marginalized likelihoods. Right now we just do an early model selection (by BIC) and go with the selected model only. That's a hack, to be sure.
"In the second, we built something (that Lang wants to call "Trogdor the Burninator") that very reliably initializes emcee for industrial-scale sampling in galaxy morphological parameters for millions of galaxies."
ReplyDeletePresumably each galaxy is treated separately? Otherwise you'll have ye olde label switching degeneracy.
yes, we are doing one galaxy at a time for now
DeleteCool. If you find that some are close together it shouldn't be too hard to then do those together.
DeleteI'm also curious to know what you are doing to take into account the fact that Sersic profiles and the like are wrong. If you assume they're correct you can get very misleading error bars on important quantities like the total flux.
totally agreed on both points and don't yet have a plan for the latter, which, as you know, is something I am working on with Vakili
Delete