Multiple-Time Step Ab Initio Molecular Dynamics Based on Two-Electron Integral Screening.

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations

Background:

• Ab initio molecular dynamics (AIMD) simulations are crucial for studying molecular systems.
• Standard AIMD methods face computational cost limitations, especially for large systems and long timescales.
• Existing integral screening methods can introduce discontinuities, hindering their use in dynamics.

Purpose of the Study:

• To develop an efficient multiple-timestep ab initio molecular dynamics scheme.
• To address the issue of discontinuities in energy and gradient arising from standard screening methods.
• To accelerate AIMD simulations while maintaining accuracy.

Main Methods:

• A novel multiple-timestep scheme based on adaptive two-electron integral screening in Hartree-Fock (HF) and density functional theory (DFT) calculations.
• Development of a consistent window-screening method to smooth discontinuities in energy and gradient.
• Algorithmic improvements including reuse of electronic-structure information within dynamics steps.

Main Results:

• The proposed scheme significantly reduces computational cost for HF and B3LYP simulations.
• Demonstrated efficiency gains compared to naive multiple-timestepping protocols.
• Successful application to a protonated sarcosine/glycine dipeptide in a water cluster.

Conclusions:

• The developed multiple-timestep AIMD scheme offers a practical acceleration of simulations.
• The consistent window-screening method effectively resolves numerical inconsistencies.
• This approach enhances the feasibility of large-scale electronic structure calculations in molecular dynamics.