Biased Metropolis Sampling for Rugged Free Energy Landscapes
Abstract
Metropolis simulations of all-atom models of peptides (i.e. small proteins) are considered. Inspired by the funnel picture of Bryngelson and Wolyness, a transformation of the updating probabilities of the dihedral angles is defined, which uses probability densities from a higher temperature to improve the algorithmic performance at a lower temperature. The method is suitable for canonical as well as for generalized ensemble simulations. A simple approximation to the full transformation is tested at room temperature for Met-Enkephalin in vacuum. Integrated autocorrelation times are found to be reduced by factors close to two and a similar improvement due to generalized ensemble methods enters multiplicatively.
- Publication:
-
The Monte Carlo Method in the Physical Sciences
- Pub Date:
- November 2003
- DOI:
- 10.1063/1.1632118
- arXiv:
- arXiv:cond-mat/0306589
- Bibcode:
- 2003AIPC..690...63B
- Keywords:
-
- 87.15.Aa;
- 02.70.Uu;
- Theory and modeling;
- computer simulation;
- Applications of Monte Carlo methods;
- Condensed Matter - Statistical Mechanics;
- Condensed Matter - Soft Condensed Matter;
- Quantitative Biology
- E-Print:
- Plenary talk at the Los Alamos conference, The Monte Carlo Method in Physical Sciences: Celebrating the 50th Anniversary of the Metropolis Algorithm, to appear in the proceedings, 11 pages, 4 figures, one table. Inconsistencies corrected and references added