Loss of Molecular Roughness upon Coarse-Graining Predicts the Artificially Accelerated Mobility of Coarse-Grained Molecular Simulation Models

Coarse-grained models include only the most important degrees of freedom to match certain target properties and thus reduce the computational costs. The dynamics of these models is usually accelerated compared to those of the parent atomistic models. We propose a new approach to predict this acceleration on the basis of the loss of geometric information upon coarse-graining. To this end, the molecular roughness difference is calculated by a numerical comparison of the molecular surfaces of both the atomistic and the coarse-grained systems. Seven homogeneous hydrocarbon liquids are coarse-grained using the structure-based iterative Boltzmann inversion. An acceleration factor is calculated as the ratio of diffusion coefficients of the coarse-grained and atomistic simulation. The molecular roughness difference and the acceleration factor of the seven test systems reach a very good linear correlation.