Routine Molecular Dynamics Simulations Including Nuclear Quantum Effects: from Force Fields to Machine Learning Potentials

TL;DR

The paper addresses the challenge of incorporating nuclear quantum effects (NQEs) into routine MD for large condensed-phase systems. It introduces Quantum-HP, a massively parallel platform integrated into Tinker-HP that supports two complementary approaches: Ring-Polymer MD (RPMD) for exact NQEs and adaptive Quantum Thermal Bath (adQTB) for affordable approximate NQEs. The authors demonstrate scalable performance on multi-CPU and multi-GPU architectures, establish compatibility with machine-learning potentials via Deep-HP, and validate the approach by computing hydration free energies with Q-AMOEBA, showing improved accuracy when including NQEs. These results position explicit NQEs as a practical component of standard MD workflows and outline future directions for re-parametrization and extension to exascale computing.

Abstract

We report the implementation of a multi-CPU and multi-GPU massively parallel platform dedicated to the explicit inclusion of nuclear quantum effects (NQEs) in the Tinker-HP molecular dynamics (MD) package. The platform, denoted Quantum-HP, exploits two simulation strategies: the Ring-Polymer Molecular Dynamics (RPMD) that provides exact structural properties at the cost of a MD simulation in an extended space of multiple replicas, and the adaptive Quantum Thermal Bath (adQTB) that imposes the quantum distribution of energy on a classical system via a generalized Langevin thermostat and provides computationally affordable and accurate (though approximate) NQEs. We discuss some implementation details, efficient numerical schemes, parallelization strategies and quickly review the GPU acceleration of our code. Our implementation allows an efficient inclusion of NQEs in MD simulations for very large systems, as demonstrated by scaling tests on water boxes with more than 200,000 atoms (simulated using the AMOEBA polarizable force field). We test the compatibility of the approach with Tinker-HP's recently introduced Deep-HP machine learning potentials module by computing water properties using the DeePMD potential with adQTB thermostating. Finally, we show that the platform is also compatible with the alchemical free energy estimation capabilities of Tinker-HP and fast enough to perform simulations. Therefore, we study how the NQEs affect the hydration free energy of small molecules solvated with the recently developed Q-AMOEBA water force field. Overall, the Quantum-HP platform allows users to perform routine quantum MD simulations of large condensed-phase systems and will participate to shed a new light on the quantum nature of important interactions in biological matter.

Figures (8)

• Figure 1: Schematic representation of the ring-polymer path-integral for $\nu=8$. Each bead $x_1,\hdots,x_\nu$ (represented by a blue circle) is subject to the physical potential and connected to its nearest neighbours via a harmonic potential (represented as springs).
• Figure 2: Schematic representation of the parallelization scheme used for the evaluation of the forces in RPMD simulations. The figure distinguishes the two subcases: $N_{proc}\leq \nu$ (top) and $N_{proc}>\nu$ (bottom). In the top figure, we define $\lambda=\nu/N_{proc}$.
• Figure 3: flow chart of a molecular dynamics simulation using the adQTB thermostat.
• Figure 4: Scaling tests on multi-CPU architecture for the different methods. Performance is indicated by the number of nanoseconds of simulation per walltime day as a function of the number of processes.
• Figure 5: Scaling tests on multi-GPU architecture for the different methods. Performance is indicated by the number of nanoseconds of simulation per walltime day as a function of the number of processes. Nodes are composed of four interconnected V100 GPUs so that when using more than four GPUs, out-of-node communications are required, causing a drop in the efficiency.