New Statistical Methods for Detecting Periodicity in Sparse Astronomical Data

University Of California-Berkeley, Berkeley CA

Investigators

Abstract

AST-0507254 Rice This project is concerned with testing for periodicity of astronomical data in a low-count or sparse observation regime, with primary application to gamma-ray emissions from rotation-powered pulsars, from the sequence of photons recorded by a system such as EGRET on the Compton Gamma-Ray Observatory, or the forthcoming Gamma Ray Large Area Space Telescope, GLAST. The methodology is rooted in hypothesis testing for semi-parametric statistical models, extending tests for periodicity of Poisson processes to other processes as well, and also to testing for non-periodic deviations from uniformity. Current approaches are limited by their ability to account for unknown frequency drift without substantially sacrificing detection sensitivity, and their computational demands are enormous. The new tests are potentially of greater sensitivity and dramatically increased computational efficiency, and after the underlying theory and an efficient software implementation are developed, unidentified sources in the EGRET database will be analyzed as the first application. Along with the inter-disciplinary training of students and junior researchers, the extremely large number of periodic and non-periodic variable phenomena means that the potential impact on astronomy is considerable.