Finding the most distant objects in the Universe by finding anomalous entries in big databases
Seminar Room 1, Newton Institute
Modern astronomical surveys can detect billions of distinct sources, which are made available to the international community in the form of on-line databases. Each object is characterised by tens or hundreds of attributes, generally heuristic statistics calculated from the raw data to encode the source's more important characteristics. In my research I attempt to find unusual objects in such databases, and here I describe a case study:
searching for the most distant quasars (the glowing material falling onto the super-massive black holes that exist at the centres of most galaxies). This search was ultimately successful, in large part due to a Bayesian candidate selection algorithm, although with hindsight several aspects of the search could have been improved.