Active Learning Accelerates Ab Initio Molecular Dynamics on Pericyclic Reactive Energy Surfaces

Abstract

Modeling dynamical effects in chemical reactions, such as post-transition state bifurcation, requires ab initio molecular dynamics simulations due to the breakdown of simpler static models like transition state theory. However, these simulations tend to be restricted to lower-accuracy electronic structure methods and scarce sampling because of their high computational cost. Here, we report the use of statistical learning to accelerate reactive molecular dynamics simulations by combining high-throughput ab initio calculations, graph-convolution interatomic potentials and active learning. This pipeline was demonstrated on an ambimodal trispericyclic reaction involving 8,8-dicyanoheptafulvene and 6,6-dimethylfulvene. With a training dataset size of approximately 31,000 M062X/def2-SVP quantum mechanical calculations, the computational cost of exploring the reactive potential energy surface was reduced by an order of magnitude. These tools are then used to automatically predict a reaction mechanism that is in agreement with the experimentally-reported product distribution. In addition, thousands of virtually costless picosecond-long reactive trajectories suggest that post-transition state bifurcation plays a very minor role in the reaction. Furthermore, a transfer-learning strategy effectively allowed to upgrade the potential energy surface to higher levels of theory (def2-TZVPD basis set and double hybrid functional) using less than 10% additional calculations. Since these methods capture intramolecular non-covalent interactions more accurately, they uncover longer lifetimes for entropic intermediates. This overall approach is broadly applicable and opens the door to the study of dynamical effects in larger, previously-intractable reactive systems.

Effect of Framework Composition and NH3 on the Diffusion of Cu+ in Cu-CHA Catalysts Predicted by Machine-Learning Accelerated Molecular Dynamics

“…Then, five more generations of active learning were used, with biased MD simulations at temperatures ranging from 600 to 1000 K to force larger deviations from the equilibrium structures, thus ensuring a better configurational sampling. The last generation of the NNP was trained on a complete data set containing 42K revPBE+D3 force calculations on structural models containing from 290 to 323 atoms per supercell, with a diverse set of atomic local environments in which the negative charges arising from Al substitution were compensated with [Cu­(NH 3 ) 2 ] + , NH 4 + , or H + − as summarized in Table and plotted in Figure b.…”

A Transferable Active-Learning Strategy for Reactive Molecular Force Fields

“…3 Machine learning (ML) approaches have the potential to revolutionize force-field based simulations, aiming to provide the best of both worlds, [4][5][6] and have indeed begun to provide new insights into a range of challenging research problems. [7][8][9][10][11][12][13][14][15] The development of a truly general ML potential mapping nuclear coordinates to total energies and forces is, however, precluded by the curse of dimensionality. Within small chemical subspaces, models can be achieved using neural networks (NNs), 6,[16][17][18][19][20] kernel-based methods such as Gaussian processes (GP) 21,22 and gradient-domain machine learning (GDML), 23 and linear fitting with properly chosen basis functions, 24,25 each with different data requirements and transferability.…”

“…Very recently, AL approaches have started to be adopted for fitting reactive potentials for organic molecules based on single point evaluations at quantum-chemical levels of theory. Notable examples include the modelling of gas-phase pericyclic reactions, 13 the exploration of reactivity during methane combustion, 49 and the decomposition of urea in water. 40 Here, we show how ab initio quality GAPs can be trained with hundreds of total ground truth evaluations for liquid water, requiring no a priori knowledge of the system.…”

Active learning and neural network potentials accelerate molecular screening of ether-based solvate ionic liquids

“…An active learning (AL) strategy was used to acquire training data efficiently by carrying out DFT only on areas with high model uncertainty. 16,17 AL augments the data space and automatically improves the model by iteratively acquiring training data and deploying molecular dynamics (MD) simulations with the learned potential. Additional DFT simulations are carried out on points in the trajectory for which the model is uncertain.…”