detecting sources in imaging

I spent a large part of the morning going through the figures, exposition, and connection to the literature in the draft by Dustin Lang (Toronto) and me about how to detect a star in a collection of probably multi-band images. There is interesting and interestingly different prior work, which wasn't prior when we started this project! Yes, it's an old paper. One of the points Lang makes is that what you detect—if, say, you use Bayes or posterior ratios—depends both on your priors and your utility. And when there are complex combinations of things creating detections in your images (including asteroids, cosmic rays, red and blue galaxies, quasars, stars, and high-redshift g dropouts, say), you want those priors to be reasonable if you want your catalogs to be reasonable.

I also spoke with Kate Storey-Fisher (NYU) about adversarial systematics in cosmology and other radical projects, and [unnamed collaborators] about our HST proposal, which is nearly ready (and anonymous, which is cool!).