Real-parameter evolutionary Monte Carlo with applications to Bayesian mixture models

We propose an evolutionary Monte Carlo algorithm to sample from a target di stribution with real-valued parameters. The attractive features of the algo rithm include the ability to learn from the samples obtained in previous st eps and the ability to improve the mixing of a system by sampling along a t emperature ladder. The effectiveness of the algorithm is examined through t hree multimodal examples and Bayesian neural networks. The numerical result s confirm that the real-coded evolutionary algorithm is a promising general approach for simulation and optimization.