The acceptance rate of Metropolis Hastings algorithm is crucial for the convergence of the MCMC. Sometimes it is advised to be about .28 (e.g. Gelman at al). Often it can be influenced by some sort of scaling parameter.

I am running a simulation with MCMC where different parameters are varied and as a result the acceptance rate varies across simulation conditions. I cannot find the best scale parameter for each conditions in advance. Rather I need to automatize finding the best scaling parameter. Are there any procedures known to achieve this?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.