Open Conference Systems, StatPhys 27 Main Conference

Font Size: 
Rejection-free cluster Monte Carlo in arbitrary external fields
Jaron Kent-Dobias, James P Sethna

##manager.scheduler.building##: Edificio San Jose Auditorio 1
Date: 2019-07-10 04:45 PM – 05:00 PM
Last modified: 2019-06-08


The discovery of cluster algorithms for Markov chain Monte Carlo was a boon to the study of critical behavior on lattice models, as these methods often overcome critical slowing down. However, they do not operate in any kind of external potential without modification, usually in the form of adding a rejection step. We present recently published work (Phys Rev E 98, 063306) that details a general method for such an extension without modifying the cluster-flip dynamics. This relies on the addition of a "ghost transformation" to the lattice that records the accumulated flips of the external field itself. We demonstrate that this extension is efficient and natural in the scaling sense for several canonical models, and provide nontrivial examples of its use.