Markov Chain Monte Carlo and a data assimilation example from cloud microphysics
Markov Chain Monte Carlo (MCMC) algorithms are designed to sample probability distributions, a rather difficult task in many applications. I plan to introduce the underpinnings of MCMC algorithms and present the affine invariant ensemble MCMC sampler (aka "The MCMC Hammer" ). After discussing the advantages of the Hammer, I will present my current research which implements this ensemble sampler. My current research involves tuning model parameters in a Bayesian framework. Specifically, I am working on estimating the model parameters of a scalar delay differential equation based on data from a large-eddy simulation, a computational model for turbulence in the atmosphere.
(Refreshments will be served.)