Adaptive resolution simulations : combining multi-particle collision dynamics and molecular dynamics simulations for fluids

摘要

In soft matter physics there is a variety of systems where phenomena occur on different time- and length scales which are inherently coupled. Examples of such systems are colloidal suspensions, polymer solutions or biological macromolecules. To simulate such systems, it is necessary to consistently take into account atomistic and hydrodynamic interactions within one computational scheme, which is feasible from the requirements of memory consumption and CPU time usage. The hybrid simulation approach presented in this work solves this problem by coupling of Molecular Dynamics and Multi-Particle Collision dynamics simulations. It allows to change the representation of the molecules composing the fluid “on the fly”, taking into account the atomistic details where it is needed, while keeping the description of the rest of the fluid on the mesoscale level. Due to the application of such hybrid coupling between fine- and coarse-grained description, it is possible to simulate larger systems for longer times efficiently, while taking into account solvent properties and hydrodynamics. The main goal of this work is to construct a hybrid description of the solvent in such a way that hydrodynamic interactions are properly accounted for. To reveal the hydrodynamic properties of the hybrid fluid, a number of correlations functions for various systems with different hybrid states were calculated. It was found that transverse current correlation functions related to the viscosity coefficient are equal for all states of the fluid, i.e. “pure” MD, “pure” MPC and all “mixed state” systems. The same applies to the properties of long tile tails in the velocity autocorrelation function, which is influencing the diffusion coefficient of the fluid. Therefore, these results show that the transport properties of the fluid are not altered throughout the hybrid description. In order to verify that hydrodynamics is maintained in the hybrid system, several test flow simulations such as Poiseuille flow, shear flow, and Couette flow were performed. They have shown that the behavior of the hybrid system under flow resembles that of a fluid modeled by a mono-scale method, and it could be shown that the deviation of the velocity profile from the theoretically predicted one is less than 2%. Although the full thermodynamic equilibrium is impossible due the fundamental differences between MD and MPC methods, the hybrid MD/MPC scheme presented in this work is proved to be a very promising approach for simulation of complex fluids. By applying the restraining force in the buffer zone it is possible to maintain dynamical equilibrium throughout the hybrid system. It was shown that the transport properties of the hybrid fluid are conserved across the transition zone between the two fluid representations in one simulation, allowing the consistent description of hydrodynamics in the whole coupled system. By changing the representation of the molecules “on the fly”, the hybrid MD/MPC approach allows to couple within a single simulation atomistic and mesoscale representation of fluids, providing a valuable tool for many problems in soft matter science.