VOStat: A Distributed Statistical Toolkit for the Virtual Observatory

Abstract

The nature of astronomical data is changing: data volumes are following Moore's law with a doubling every 18 months and data sets consisting of a billion data vectors in a 100-dimensional parameter space are becoming commonplace. Sophisticated statistical techniques are crucial to fully and efficiently exploit these and maximize the scientific return. A long-standing limitation, however, on the range and capability of such analyses has been the paucity of non-proprietary software. VOStat is the result of a cross-disciplinary collaboration between astronomers and statisticians to meet these challenges; it is a prototype knowledge-based statistical toolkit implemented within the VO paradigm for the entire astronomical community. VOStat consists of an easily extensible distributed web-based framework transparently accessed via a single science endpoint. An exploratory science application is presented to demonstrate some of the functionality currently offered by VOStat.