I will start with a short introduction to Bayesian statistics, including the magic of Markov chain Monte Carlo. With these concepts, I will describe how to build and use an emulator, a fast statistical approximation to a computationally intensive simulation. I will present this in the context of the Cosmic Emu, an emulator for the matter power spectrum based on large n-body simulations. The work involves various types of Gaussian process models, a flexible approach to regression that's useful in lots of contexts, cosmological and otherwise.
Speaker: Earl Lawrence, Los Alamos National Labs
Contact:Website: Click to Visit
Save this Event:iCalendar
Windows Live Calendar