Energy drift in molecular dynamics simulations
- 178 Downloads
In molecular dynamics, Hamiltonian systems of differential equations are numerically integrated using the Störmer–Verlet method. One feature of these simulations is that there is an unphysical drift in the energy of the system over long integration periods. We study this energy drift, by considering a representative system in which it can be easily observed and studied. We show that if the system is started in a random initial configuration, the error in energy of the numerically computed solution is well modeled as a continuous-time stochastic process: geometric Brownian motion. We discuss what in our model is likely to remain the same or to change if our approach is applied to more realistic molecular dynamics simulations.
Key wordsHamiltonian systems symplectic numerical methods modified Hamiltonian shadow Hamiltonian backward error analysis molecular dynamics
Unable to display preview. Download preview PDF.
- 4.G. Ciccotti, D. Frenkel, and I. R. McDonald, eds., Simulation of Liquids and Solids: Molecular Dynamics and Monte Carlo Methods in Statistical Mechanics, North-Holland, Amsterdam, 1987.Google Scholar
- 5.D. Cottrell, Symplectic integration of simple collisions: A backward error analysis, Master’s thesis, McGill University, 2004. (www.math.mcgill.ca/∼cottrell/thesis.pdf)Google Scholar
- 8.E. Hairer, C. Lubich, and G. Wanner, Geometric numerical integration illustrated by the Störmer–Verlet method, Acta Numer., (2003), pp. 339–450.Google Scholar
- 12.E. Shaw, H. Sigurgeirsson, and A. M. Stuart, A Markov model for billiards, preprint, Warwick Mathematics Institute, 2004.Google Scholar
- 15.L. Verlet, Computer “experiments” on classical fluids. I. Thermodynamical properties of Lennard–Jones molecules, Phys. Rev., 159 (1967), pp. 98–103.Google Scholar