|
|
|
Developing Methods to Incorporate Calibration Uncertainties in Data Analysis
|
|
|
| The aim of this proposal is to develop and make publicly available a set of tools designed to incorporate knowledge of calibration errors into data analysis. We have developed a method to handle in a practical way the effect of uncertainties in instrument response on astrophysical modeling, with specific application to Chandra/ACIS instrument effective area. This groundbreaking work holds great practical promise for a generalized treatment of instrumental uncertainties in not just X-ray spectra but also imaging, or any kind of higher-dimensional analyses. We apply a combination of Principal Component Analysis and Markov chain Monte Carlo methods to calibration uncertainties and include these uncertainties directly into analyses. We propose to develop modular tools that can be easily extended to different instruments and calibration products, and will provide a systematic framework to describe the true errors in the estimated model parameters, even in non-Gaussian regimes.
Calibration is the foundation on which all data are analyzed and interpreted. Therefore, any efforts to improve handling of calibration data and to use it in a better fashion is immediately relevant to Space Science programs. Considerable effort will be saved by reducing the propensity of systematic calibration-based biases from skewing the results. Our proposed project will also have specific impacts. Firstly, because algorithm development will mostly take place in the context of Chandra data, it will lead directly to an improvement in the quality of the analysis of Chandra data. Secondly, it will result in the useful characterization of a number of calibration products from Chandra, Suzaku, and other current missions. Finally, we will set up groundbreaking standards in the manner and specification of calibration data for future missions. |
|
|
