|
|
|
Automated Classification of X-ray Sources for Very Large Datasets
|
|
|
| The enormous amount of multidimensional data generated by increasingly higher quality observatories is resulting in very large astronomical databases. The growing data archives of the Chandra X-ray Observatory (CXO) and the X-ray Multi-Mirror Mission (XMM-Newton) provide excellent examples of this conundrum. Unbiased source classification methods would augment and enhance standard spectral and temporal analysis techniques that are routinely applied to CXO and XMM data. We are proposing to develop a novel statistical source clustering and classification algorithm to maximize the science return from these and other large, multidimensional astronomical datasets. Our initial emphasis is on spectral classification of hundreds of X-ray sources detected in CXO observations of the Orion Nebula Cluster. We will generalize our classification techniques to take into account temporal attributes of X-ray sources and thereby create a robust clustering algorithm. The algorithm also will be extended for use in blindly classifying X-ray sources in other regions of the sky. Inputs and procedures specific to X-ray wavelengths will be modularized so the algorithm may be used with data from other regions of the electromagnetic spectrum. The ultimate goal is to develop the capability to group sources in large fields, such as a sky survey, independent of the requirement of a model or a priori knowledge of the nature of the sources (i.e., young stars, binaries, active galactic nuclei). Methods to visualize the results will be explored. The resulting algorithm will be developed into a tool and made available to the established centers for astronomical data analysis. The proposed research is relevant to Goal II, Astronomical Search for Origins, RFA 2(a). The final algorithm will increase science return from observational data through advanced knowledge discovery. The algorithm will be of immediate use to X-ray astronomers studying the mechanisms underlying X-ray emission to improve our understanding of the nature and timescale of accretion onto young, solar-mass stars from protoplanetary disks. |
|
|
