MSPA-AST: Multilevel Modeling of Active Galaxy Populations

点击翻译按钮获取中文摘要

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).Dr Loredo and his team will develop Bayesian multi-level modeling for astrophysical problems. Bayesian multi-level models decompose the random variation in measured quantities into distinct layers, each representing a different contribution: layers 'close' to the data might describe the uncertainty in the measurements, while layers 'farther away' typically describe cosmic scatter arising from unobserved variables. Their first project aims to account for two common sources of bias in estimating the distribution of luminosities within a sample of objects: bias from the way that the sample is selected, and bias stemming from the fact that the 'measured' brightnesses of objects are not in fact independent of each other. Astronomers often neglect the second of these, although it can be significant. Dr Loredo and his team will apply their methods to data from the Two-Degree Field quasar survey and the Sloan Digital Sky Survey. In their second project, the team will use a Bayesian multi-level model to test whether the active galactic nuclei seen near the directions from which ultra-high energy cosmic rays are observed are truly associated with the cosmic-ray sources, or whether the alignment is likely to be coincidental. The team will apply their methods to cosmic-ray measurements from HiRes and the Auger observatory. The project team includes experts in both astronomy and statistics. Dr Loredo is an associate of the Center for Astrostatistics at Penn State University, where he lectures at the annual Summer School. At that summer school, he will offer topical sessions on statistical methods for astronomical surveys and multilevel models. The research should produce significant innovations in both disciplines, and will improve the science return from large astronomical surveys.