Utilizing fast multipole expansions for efficient and accurate quantum-classical molecular dynamics simulations.

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations
• Quantum Mechanics/Molecular Mechanics (QM/MM)

Background:

• Hybrid QM/MM methods are crucial for simulating complex molecular systems.
• Previous QM/MM-MD simulations faced challenges in computational efficiency and accuracy, especially with large environments.
• The Fast Multipole Method (FMM) has been used to describe electrostatic interactions in these systems.

Purpose of the Study:

• To enhance the efficiency and accuracy of hybrid QM/MM-MD simulations.
• To adapt and improve the FMM technique for calculating interactions between DFT and PMM fragments.
• To enable efficient use of high-performance computing for QM/MM-MD simulations.

Main Methods:

• Employed a hybrid approach combining grid-based density functional theory (DFT) for solute and polarizable molecular mechanics (PMM) for solvent.
• Utilized a hierarchical Fast Multipole Method (FMM) for efficient electrostatic interaction calculations.
• Adapted a revised FMM approach with strictly linear scaling computational effort with system size.
• Implemented a jointly parallelized computation of DFT and PMM-MD parts.

Main Results:

• Achieved a significant enhancement in the efficiency and accuracy of DFT/PMM-MD simulations.
• Demonstrated a performance gain of approximately one order of magnitude for alanine dipeptide in water.
• Showcased the efficient utilization of high-performance computing systems through parallelized implementation.

Conclusions:

• The revised FMM approach substantially improves QM/MM-MD simulations.
• The enhanced method offers a powerful tool for accurate and efficient molecular simulations.
• The developed software is publicly available, facilitating further research in computational chemistry.