Abstract:

"the Tractor" is a new project to frame astronomical source detection and measurement as an inference problem in a pixel-level generative (forward) model. This approach naturally handles the multi-band, many-epoch data sets that will be produced by surveys such as Pan-STARRS, the Dark Energy Survey, and the Large Synoptic Survey Telescope (LSST), but it is also well suited to more diverse imaging sets from different instruments and different wavelengths. Unlike traditional approaches, we avoid co-adding, degrading, or discarding any of the images; the goal is to make use of the (often complementary) information in all the images. This strength of the Tractor makes it ideally suited to producing the accurate wide-area photometric catalogs required for targeted spectroscopic surveys such as BigBOSS.