Detecting Unexpected Residual Spectral Structure in X-Ray Astronomy

• Activity Number: 308
• Type: Topic Contributed
• Date/Time: Tuesday, August 5, 2008 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Jason Kramer*+ and David A. van Dyk+
• Companies: University of California, Irvine and University of California, Irvine
• Keywords: astronomy ; multi-scale ; Markov random field ; MCMC ; EM ; Poisson

Abstract

As X-ray telescopes have become more powerful, datasets collected for high-energy spectral analysis have become larger and increasingly complex. Spectral analysis is the study of the energy distribution of photons arriving from an astronomical source and gives clues as to the temperature, composition, and physical processes of the source. Photon counts are modeled as a Poisson process while blurring of energies, stochastic censoring, and background contamination are built into a highly structured hierarchical model. We use a finite mixture model with one component for a physics-based parametric spectral model and a second component that uses a smoothing multi-scale prior distribution to account for unexpected residual spectral structure in the data. Bayesian methodology and computing are used to fit the model and to compute estimates, error bars, and credible intervals.