Stochastic Effective Core Potentials, toward Efficient Quantum Monte Carlo Simulations of Molecules with Large Atomic...

Area of Science:

• Computational Chemistry
• Quantum Chemistry
• Molecular Dynamics

Background:

• Accurate molecular simulations are computationally intensive, particularly for large systems.
• Core electrons often contribute significantly to computational cost without substantial impact on valence electron behavior.

Purpose of the Study:

• To develop a Monte Carlo method that minimizes the computational cost associated with core electron regions.
• To accelerate molecular dynamics simulations by efficiently handling core electron contributions.

Main Methods:

• A novel Monte Carlo approach treating core regions as physically independent.
• On-the-fly computation of effective core potentials via efficient subsampling.
• A two-time-step process to accelerate valence region dynamics.

Main Results:

• Numerical overhead O(N) is negligible compared to the O(N^3) scaling.
• Achieved numerical efficiency gains of one to two orders of magnitude for large N.
• Demonstrated parameter transferability from atoms to molecules, allowing calibration with single atoms.

Conclusions:

• The proposed method significantly reduces computational cost in molecular simulations.
• Effective handling of core regions enables accelerated dynamics and broad applicability.
• The method offers a computationally efficient alternative for large-scale molecular modeling.