Hamiltonian Monte Carlo by Jonathan Patrick ’24

Hamiltonian Monte Carlo: Leveraging Hamiltonian Dynamics to Sample from Complex Probability Distributions by Jonathan Patrick ’24, Wednesday February 28, 1:10 – 1:50pm, North Science Building 015, Wachenheim, Statistics Colloquium

Abstract:
Obtaining a sequence of random samples from well-studied, low-dimensional distributions is simple enough, but efficiently sampling from arbitrary, high-dimensional distributions is a challenging problem in computational statistics. This talk will explore Hamiltonian Monte Carlo sampling, which leverages Hamiltonian dynamics to generate random samples from complex distributions with notable efficiency. We will address the mechanics underlying this innovative sampling method before evaluating its effectiveness and surveying a subset of its applications.