Machine Learning-Assisted Hybrid ReaxFF Simulations

Abstract

We have developed a machine learning (ML)-assisted Hybrid ReaxFF simulation method (“Hybrid/Reax”), which alternates reactive and non-reactive molecular dynamics simulations with the assistance of ML models to simulate phenomena that require longer time scales and/or larger systems than are typically accessible to ReaxFF. Hybrid/Reax uses a specialized tracking tool during the reactive simulations to further accelerate chemical reactions. Non-reactive simulations are used to equilibrate the system after the reactive simulation stage. ML models are used between reactive and non-reactive stages to predict non-reactive force field parameters of the system based on the updated bond topology. Hybrid/Reax simulation cycles can be continued until the desired chemical reactions are observed. As a case study, this method was used to study the cross-linking of a polyethylene (PE) matrix analogue (decane) with the cross-linking agent dicumyl peroxide (DCP). We were able to run relatively long simulations [>20 million molecular dynamics (MD) steps] on a small test system (4660 atoms) to simulate cross-linking reactions of PE in the presence of DCP. Starting with 80 PE molecules, more than half of them cross-linked by the end of the Hybrid/Reax cycles on a single Xeon processor in under 48 h. This simulation would take approximately 1 month if run with pure ReaxFF MD on the same machine.

C/H/O/F/Al ReaxFF Force Field Development and Application to Study the Condensed-Phase Poly(vinylidene fluoride) and Reaction Mechanisms with Aluminum

“…We considered three systems to include the training set: hexafluoroethane (C 2 F 6 ), dodecafluoropentane, neo-dodecafluoropentane (C 5 F 12 ), as shown in Figure S2, and perfluorooctane (C 8 F 18 ) . We generated OPLS parameters and charges for these molecules using a machine learning-based technique. , We build crystal structures, starting with hexafluoroethane and using a body-centered cubic (BCC) system with two molecules per unit cell and lattice constants 4.8, 4.15, and 9.9 Å, found in the literature . For the rest of the molecules, we build BCC crystal structures with matching density of the hexafluoroethane, 2.3 g/cm 3 .…”

Modeling and simulations for 2D materials: a ReaxFF perspective

“…MoS 2 after fewer search iterations than that of genetic algorithm based ML models (figure 28(g)). Another work [332] addressed the acceleration of chemical reactions that are observed during the growth and characterization of 2D materials and require larger sizes and/or longer time scales than that are accessible to MD simulations, by developing ML assisted hybrid ReaxFF simulation method. Combining nonreactive (i.e.…”

“…(h) ML-assisted hybrid ReaxFF simulations: (i) molecular composition of the system at ML-assisted Hybrid Reax level, (j) snapshot of the system and (k) focus on the cross-linked decane after 20 million MD steps. Reprinted with permission from [332]. Copyright (2021) American Chemical Society.…”

A Gear Decision of Hybrid Electric Truck Based on Neural Network

“…(1) Improvement history Gear decision technology has its own improvement track since its initial research and improvement. With the continuous improvement and innovation of technology, related gear decision technology has also developed rapidly [ 8,9], which can be divided into four stages, as shown in Table 1:…”