Metropolis-Hastings algorithm
(1.4 hours to learn)
Summary
Markov Chain Monte Carlo (MCMC) is a method for approximately sampling from a distribution p by defining a Markov chain which has p as a stationary distribution. Metropolis-Hastings is a very general recipe for finding such a Markov chain: choose a proposal distribution and correct for the bias by stochastically accepting or rejecting the proposal. While the mathematical formalism is very general, there is an art to choosing good proposal distributions.
Context
This concept has the prerequisites:
- Markov chain Monte Carlo (M-H is an example of an MCMC algorithm.)
- multivariate Gaussian distribution (Gaussian proposals are a canonical example of an M-H proposal distribution.)
See also
-No Additional Notes-