Differential Evolution Monte Carlo sampling

easy Bayesian computation for real parameter spaces

Description

This code implements a Markov chain Monte Carlo algorithm which automatically and efficiently tunes the proposal distribution to the covariance structure of the target distribution. This is achieved while maintaining the target distribution as the stationary distribution of the Markov chain. The algorithm is described in:

Cajo F.T. Ter Braak, "A Markov Chain Monte Carlo version of the genetic algorithm Differential Evolution: easy Bayesian computing for real parameter spaces", Stat Comput (2006) 16:239–249

As of the date of submission, this paper is freely available at:

http://www.stat.columbia.edu/~gelman/stuff_for_blog/cajo.pdf

• MATLAB release: MATLAB 7.1.0 (R14SP3)

[Joseph Catanzarite]

Good entry into differential evolution monte carlo. Code is clear, understandable and easy to modify for your own optimization problem.

[Henry Zhu]

Corey Yanofsky writes a very useful code about MCMC for us.

[Xianzhen]

Dear Corey: The program provides a so good entry into DE-MC that I want to use it to generate random numbers from a target distribution. Although you provide a detail demo to demonstrate the use of the program, it is difficult for me to fully understand it due to the lack of knowledge. It will be greatly appreciate if you could spend some of your time to show me the way to generate random numbers (e.g. from a multivariate normal target with mu = [1 -1] and cov = [0.9 0.4; 0.4 0.3]). Thank you.