Title:
Cosmological Model Selection
Andrew Liddle
University of Sussex, UK
Abstract:
Most analyses of cosmological data focus on estimation of cosmological
parameters. However many of the most interesting questions concern not
the values of parameters, but rather which \emph{set} of parameters
gives the preferred fit to data. This is the statistical problem of
\emph{model selection}. I describe the main tools of model selection,
and describe an implementation of Bayesian model selection for
cosmological data. As an example, these techniques are applied to a
data compilation including WMAP3, focussing on inflationary cosmology.