I am using the Monte-Carlo metropolis algorithm to simulate the Ising model. Since the convergence is slow near $T_c$, I am looking for a method to speed up the problem. What I did was instead of evaluating the algorithm one site at a time, I am doing it for all sites simultaneously in each iteration. Is this a valid method?

EDIT: What I observe is that the the algorithm converges to the half-full lattice (density = 0.5) at temperatures above $T_c$. Some figures are drawn below:

enter image description here enter image description here

As we can see, there is a horizontal line that we do not expect. What is causing it?

  • $\begingroup$ This is a valid perturbation of the state in the MCMC framework, but it is possible that you get very low acceptance rates. What do you observe? $\endgroup$ May 15 at 0:00


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.