Diffusion Constants

For the computation of diffusion coefficients from MD simulations in equilibrium cases we use classical methods such as Mean Sqare Displacement and Velocity Autocorrelation Functions. However in cases where the particles are subject to convection or even acceleration, these models are insufficient. We have improved the existing methods to determine the presence of steady states and compute correct diffusion coefficients by accounting for the non-diffusive motion. The figure illustrates the conventional way versus the corrected method.

1. C. Neuen: A Multiscale Approach to the Poisson-Nernst-Planck Equation. Diploma thesis, Institute for Numerical Simulation, University of Bonn, July 2010.