|Statistical Estimation in Modern Astronomy: Repeated Inverse Problems and Fuctional Interpolation
Darren Homrighausen, Carnegie Mellon University
Monday, January 30, 2012
4:00 p.m., room 237, Weber Bldg
Statistics has for centuries played a critical role in the analysis of astronomical data. With the advent of new technologies, ultra-large telescopes, and automated sky-surveys, data of unprecedented size and precision is becoming available and the need for new statistical tools and ideas has never been greater.
In this talk, I will present new methods for tackling two related problems that arise in the data-processing pipelines of modern Astronomy. The first is a method for combining sequences of low-quality observations into an accurate estimate of the underlying signal of interest. This method has many advantages over existing ones, such as automatic tuning parameter selection and a strong theoretical backing. The second method I will present is a new technique for comparing astronomical images when one of the images is of much lower quality than the other. This is an important step in the automated real-time discovery of interesting transient phenomena. Both methods are being developed for use in the Large Synoptic Survey Telescope (LSST).