PPT Slide
In addition to providing a framework for analysis of galaxy surveys, Heavens and Taylor, with Ballinger and Tadros (Oxford) have developed Data Compression techniques for improving the accuracy with which cosmological parameters can be determined.
To get as much information as possible out of the data requires simultaneous analysis of many spectral modes. Accurate analysis of these modes can be very slow, and more seriously, many of these data are noisy and correlated, so much so that the data covariance matrix can be numerically singular. This has an unfortunate consequence: ordinary maximum likelihood analysis methods will not work. However, the noisy and correlated nature of the modes allows allows data to be thrown away without significant loss of information. We have developed techniques for doing this in a way which preserves as much information as possible about the quantities we wish to measure. This allows more modes to be analysed, and better accuracy results.