This is a cross-disciplinary, multi-investigator, multi-faceted suite of astro-statistical research projects bridging the fields of astronomy and statistics. Current and future astronomical research involves vast imaging, photometric and spectroscopic surveys that produce terabyte to petabyte databases. While the scientific promise is tremendous, achieving the goals depends critically on the ability to extract useful knowledge from such large datasets. The statistical problems are diverse, and astronomers typically confront them with inadequate knowledge of advanced methods, due both to limited training and to insufficiently focused methodological research.
The original plan for this project included a large range of activities, covering likelihood and resampling methods, multivariate methods for photometric redshifts, the local classification of multivariate datasets; analysis of mega-datasets with efficient computational algorithms; and classification of low-resolution spectra. The project as supported will cover a rather smaller subset of these topics, to be determined as the work progresses.
The education and dissemination work remains a large fraction of the effort, including organizing the continuation of an annual intensive Summer School in statistical inference, not supported by this award. All of this work will, however, have a substantial impact on both the statistical quality of astronomical survey science and on the training of many young researchers.
