Spatial Clustering of Galaxies in Large Datasets
- Alexander S. Szalay ,
- Tamás Budavari ,
- Andrew Connolly ,
- Jim Gray ,
- Takahiko Matsubara ,
- Adrian Pope ,
- István Szapudi
MSR-TR-2002-86 |
The 22nd Annual ACM Symposium on Applied Computing (SAC 2007)
Datasets with tens of millions of galaxies present new challenges for the analysis of spatial clustering. We have built a framework that integrates a database of object catalogs, tools for creating masks of bad regions, and a fast (NlogN) correlation code. This system has enabled unprecedented efficiency in carrying out the analysis of galaxy clustering in the SDSS catalog. A similar approach is used to compute the three-dimensional spatial clustering of galaxies on very large scales. We describe our strategy to estimate the effect of photometric errors using a database. We discuss our efforts as an early example of data-intensive science. While it would have been possible to get these results without the framework we describe, it will be infeasible to perform these computations on the future huge datasets without using this framework.