Dresden 2020 – wissenschaftliches Programm

Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 32: Condensed-matter simulations augmented by advanced statistical methodologies (joint session DY/CPP)

CPP 32.1: Vortrag

Montag, 16. März 2020, 15:30–15:45, HÜL 186

Funnel Hopping Monte Carlo: An efficient method to overcome broken ergodicity — •Jonas Alexander Finkler and Stefan Goedecker — University of Basel, Switzerland

Monte Carlo simulations are a powerful tool in the investigation of thermodynamic properties of atomic systems. In practice however sampling of the complete configuration space is often hindered by high energy barriers between different regions of configuration space which can make ergodic sampling completely infeasible within accessible simulation time.

We present a novel method called Funnel Hopping Monte Carlo (FHMC) to overcome broken ergodicity in Monte Carlo simulations. The method requires a priory knowledge of the funnel structure of the potential energy surface. This information does not have to be obtained from expensive Monte Carlo simulations but can be obtained very cheaply from a Minima Hopping [1] run. By approximating the Boltzmann distribution around local minima using Gaussian mixtures high quality Monte Carlo moves can be proposed that allow the simulation to directly jump between different funnels, circumventing the energy barriers in between.

The methods performance is demonstrated on the example of the 38 and 75 atom Lennard-Jones clusters which are well known for their double funnel energy landscapes that completely prevent ergodic sampling with conventional Monte Carlo simulations.

[1] S. Goedecker, J. Chem. Phys. 120, 9911 (2004)