Markov Chain Monte Carlo (MCMC) is one the most powerful and versatile methods developed in the 20th century. It uses a sequences of random numbers to solve important problems in physics, computational biology, econometrics, political science, Bayesian inference, machine learning, data science, optimization, etc. For many of these problems, simple Monte Carlo ("integration by darts") is inefficient. Often, MCMC is the answer.
This course provides a gentle "hands-on" introduction to MCMC and some of its applications.
Topics include:
-
Basics of probability theory
Random number Generation
Importance Sampling
Simple Monte Carlo versus MCMC
Basic sampling algorithms
Analysis of convergence and error