Direct construction of mesoscopic models from microscopic simulations.

Area of Science:

• Computational physics
• Materials science
• Statistical mechanics

Background:

• Microscopic simulations like molecular-dynamics (MD) are computationally intensive for large systems.
• Coarse-graining reduces complexity by representing groups of atoms as single particles.
• Mesoscopic models offer a balance between atomistic detail and computational efficiency.

Purpose of the Study:

• To develop and validate mesoscopic models (Langevin dynamics and DPD) from microscopic MD simulations.
• To investigate the static and dynamic properties of these coarse-grained models.
• To assess the accuracy of mesoscopic models in reproducing atomistic system behavior.

Main Methods:

• Constrained molecular-dynamics (MD) simulations of Lennard-Jones (LJ) clusters.
• Coarse-graining LJ clusters into single particles for Langevin dynamics and DPD models.
• Calculation of effective mean force fields and friction force coefficients.
• Comparison of mesoscopic model properties with MD results across various densities and cluster sizes.

Main Results:

• The mean force field derived from mesoscopic models scales linearly with particle count and temperature.
• Mesoscopic models accurately reproduce the equation of state across a wide density range.
• Radial distribution functions are accurately reproduced only in dilute and semidilute regimes.
• Friction and diffusivity are overestimated/underestimated at high densities due to omitted many-body effects.

Conclusions:

• Mesoscopic models, particularly DPD, can effectively reproduce atomistic system properties under specific conditions.
• The accuracy of coarse-grained models depends on the system's density and the inclusion of many-body effects.
• DPD shows promise for simulating systems where many-body interactions are less dominant.