Hamiltonian Monte Carlo with Constrained Molecular Dynamics as Gibbs Sampling
| Autor: | David D. L. Minh, Laurentiu N. Spiridon |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2017 |
| Předmět: |
Physics
010304 chemical physics Monte Carlo method Sampling (statistics) 010402 general chemistry 01 natural sciences Article 0104 chemical sciences Computer Science Applications Hybrid Monte Carlo symbols.namesake Classical mechanics 0103 physical sciences Dynamic Monte Carlo method symbols Monte Carlo method in statistical physics Monte Carlo integration Statistical physics Physical and Theoretical Chemistry Monte Carlo molecular modeling Gibbs sampling |
| Popis: | Compared to fully flexible molecular dynamics, simulations of constrained systems can use larger time steps and focus kinetic energy on soft degrees of freedom. Achieving ergodic sampling from the Boltzmann distribution, however, has proven challenging. Using recent generalizations of the equipartition principle and Fixman potential, here we implement Hamiltonian Monte Carlo based on constrained molecular dynamics as a Gibbs sampling move. By mixing Hamiltonian Monte Carlo based on fully flexible and torsional dynamics, we are able to reproduce free energy landscapes of simple model systems and enhance sampling of macrocycles. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|