Events

Events at Physics

<< Summer 2009 Fall 2009 Spring 2010 >>
Subscribe your calendar or receive email announcements of events
NPAC (Nuclear/Particle/Astro/Cosmo) Forum
Quantifying the Unknown in Astronomy: A Bayesian Approach
Date: Friday, October 23rd
Time: 2:30 pm
Place: 5280 Chamberlin
Speaker: Brian Connolly, University of Pennsylvania
Abstract: Over the last few years the Bayesian statistics has played an increasingly important role in astronomical data analyses, from classifying quasars to fitting cosmological models to WMAP data. One aspect of Bayesian statistical methods used to classify astronomical objects is that traditionally they assume that the object being classified falls into a finite set of modeled (or known) astronomical objects. However, astronomical research continually reveals new, unexplained phenomena; new large-scale surveys coming up in the next decade (such as the Dark Energy Survey, the Large Synoptic Survey Telescope, the Joint Dark Energy Mission, etc.) are expected to greatly enrich our catalog of understood (or at least modeled) astronomical objects. In my
talk, I will first review the two approaches to statistics, Bayesian and Frequentist. I will describe how and why the Bayesian approach has been so successful in astronomy in general, and in particular how the Bayesian approach can be extended to account for as yet unmodeled objects. I will then show how this method can be used to quantify the differences in the spectra of
Ultra-Luminous Infra-Red Galaxies (ULIRGs) and aid in the identification of Type Ia supernovae, a staple of modern cosmological research. I will also discuss how network diagrams and graph theory can be used in conjunction with these Bayesian methods to enhance our understanding of the evolution of ULRGs
and Type Ia supernovae.
Host: Stefan Westerhoff
Add this event to your calendar