In statistical physics, by far the most often used Monte Carlo (MC)
method is Markov chain Metropolis sampling. But in some niches also other
methods have been developed. The most important one, having applications
in Bayesian statistics, neutron transport theory, quantum MC simulations,
polymers and other fields, is sequential MC. We have developed a new version
of it, called pruned-enriched Rosenbluth method
(PERM), and applied it to various problems. First applications
were to polymers (including protein folding, DNA melting, critical unmixing,
etc.), and we are still working in this field. Just now we are developing
an improved version for systems at very low temperatures, which is much
more efficient than previous versions, and apply it to model proteins.
But we also applied PERM to other problems like the spreading of epidemics,
reaction-diffusion models, and multiple percolating clusters. Applications
to noise reduction and filtering are considered.
|
