Multilevel simulation of hard-sphere mixtures.

Area of Science:

• Computational physics
• Statistical mechanics
• Multiscale modeling

Background:

• Analyzing multi-scale physical systems requires efficient simulation methods.
• Coarse-grained representations can simplify complex systems but may lose detail.
• Monte Carlo methods are widely used but can be computationally intensive.

Purpose of the Study:

• To introduce and evaluate a multilevel Monte Carlo (MLMC) simulation method for multi-scale physical systems.
• To obtain numerically exact results at the most detailed level.
• To improve the performance of estimators in simulations.

Main Methods:

• Developed a multilevel Monte Carlo simulation approach using a hierarchy of coarse-grained representations.
• Applied the method to a mixture of size-asymmetric hard spheres in the grand canonical ensemble.
• Compared a three-level MLMC method with a previously studied two-level version.

Main Results:

• The three-level MLMC method demonstrated improved estimator performance compared to the two-level version at fixed computational cost.
• The enhanced performance stems from an additional interpolation level restricting small particles near large ones.
• Asymptotic variance of the estimator was analyzed to understand performance improvements.

Conclusions:

• The developed multilevel Monte Carlo method offers a more efficient way to simulate multi-scale physical systems.
• Introducing intermediate coarse-grained levels can significantly enhance simulation accuracy and computational efficiency.
• This approach provides a pathway to numerically exact results for complex physical systems.