How can we use upcoming cosmological data sets covering the largest scales accessible to observation to learn about the cosmic beginning, dark energy and dark matter in a way that uses all the ways the data depend on cosmology, is close to optimal, and avoids double-counting?
Traditionally, various derived data products are analyzed separately to derive cosmological constraints which are then combined after the fact. For example, galaxy surveys are used to obtain cluster number counts, measurements of redshift space distortions, velocity information, two-point and higher order spatial correlations, shape statistics etc., each of which can be analyzed further, exploiting various physical effects to yield constraints on cosmological parameters. These constraints are then combined after the fact. This is a sensible approach when data is scarce and the data products obtained from different surveys are very nearly independent.
As we are beginning to sample significant portions of the observable universe, we enter a new regime where cosmological inferences obtained from different summaries of observational data are no longer approximately independent. This is true even if the data sets are derived from different telescopes, and even if the observed volumes are non-overlapping, since the data is taken from only one Universe.
Combining constraints on cosmological physics from different sources without taking into account these somewhat counterintuitive dependencies would therefore either overuse the data or neglect ways in which different data sets can enhance one another. I will discuss the promise and the challenges of a radically different, global approach to cosmological data analysis that avoids these difficulties, unifies the cosmological inference from several physical effects and yet remains modular in its implementation.
Keywords: Cosmology; Astrostatistics; Bayesian statistics
Biography: Ben Wandelt is a theoretical cosmologist with interests in data analysis and computation as applied to cosmological problems. He completed his undergraduate and graduate studies at Imperial College, London from 1991 until 1997 and then held postdoctoral positions at the Niels Bohr Institute and at Princeton University. He joined the Physics and Astronomy faculties at the University at Illinois in 2001. He received a Friedrich Wilhem Bessel prize in 2006, an Arnold E Beckman Research award in 2008, and the Xerox award for faculty research in 2009. In 2010 Ben moved to the Institute for Astrophysics in Paris (IAP), where he is Professor and International Chair as well as joint director of the Initiative in Cosmology and Astroparticle Physics at the IAP.