Explore a one-hour seminar presentation from Dr. Giorgos Vasdekis of Newcastle University discussing the development and application of skew-symmetric schemes for sampling from diffusion processes. Delve into the innovative locally balancing algorithms, with particular focus on the Barker algorithm and its robust performance in handling heteroskedasticity of posterior targets and step size variations. Learn how concepts from skew-symmetric distribution sampling can be applied to diffusion process simulation, leading to a new class of unadjusted MCMC algorithms. Understand the theoretical foundations and practical implications of these methods in the context of stochastic systems for anomalous diffusion, presented as part of the research program at the Isaac Newton Institute. Gain insights from collaborative research conducted with S. Livingstone, N. Nusken, and R. Zhang, examining how these algorithms maintain high dimensional properties while offering improved robustness compared to existing approaches.
Overview
Syllabus
Date: 11th Nov 2024 - 14:00 to
Taught by
INI Seminar Room 2