DPG Phi
Verhandlungen
Verhandlungen
DPG

SKM 2021 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 14: Condensed-Matter Simulations augmented by Advanced Statistical Methodologies (joint session DY/CPP)

CPP 14.2: Talk

Friday, October 1, 2021, 10:15–10:30, H2

Population Annealing Monte Carlo Using the Rejection-Free n-Fold Way Update Applied to a Frustrated Ising Model on the Honeycomb Lattice — •Denis Gessert1,2 and Martin Weigel1,31Centre for Fluid and Complex Systems, Coventry University, Coventry, CV1 5FB, United Kingdom — 2Institut für Theoretische Physik, Leipzig University, Postfach 100920, D-04009 Leipzig, Germany — 3Institut für Physik, Technische Universität Chemnitz, D-09107 Chemnitz, Germany

Population annealing (PA) is a MC method well suited for the study of systems with a rough free energy landscape, e.g. glassy systems. PA is similar to an equilibrium version of parallel simulated annealing runs with the addition of a resampling step at each temperature. While a large population may improve imperfect equilibration, it is evident PA will fail when almost no spins are flipped in the equilibration routine.

This is the case in systems with a low temperature phase transition where high Metropolis rejection rates make sampling phase space near infeasible. To overcome this slowdown we propose a combination of the PA framework with the rejection-free “n-fold way” update and achieve an exponential speed-up at low temperatures compared to Metropolis.

To test our method we study the Ising model with competing ferromagnetic nearest and antiferromagnetic next-to-nearest neighbor interactions of strengths J1 > 0 and J2 < 0, resp., on the honeycomb lattice. As Tc becomes arbitrarily small, when approaching the special point J2=−J1/4 with Tc = 0, we consider this a good choice to test the efficacy of our method.

100% | Mobile Layout | Deutsche Version | Contact/Imprint/Privacy
DPG-Physik > DPG-Verhandlungen > 2021 > SKM