Cognitive Models logo Cognitive Models

Migration operator is one crucial operator in the distributed genetic algorithm (Tanese, 1989; Hu & Tsai, 2005), originated from the genetic algorithm (Holland, 1975; Goldberg, 1989). The algorithm uses a similar scheme, like chromosomes exchange gene information.

Migration in MCMC computation uses the same method as in random-walk Metropolis to propose a new proposal, which is then subjected to the Metropolis decision step to accept or reject as a valid sample from a target distribution. The method of migration operator to propose a proposal is that it adds random noise vector onto a “current” parameter vector as a new proposal.