Automatically growing global reactive neural network potential energy surfaces: A trajectory-free active learning strategy.

  title={Automatically growing global reactive neural network potential energy surfaces: A trajectory-free active learning strategy.},
  author={Qidong Lin and Yaolong Zhang and Bin Zhao and Bin Jiang},
  journal={The Journal of chemical physics},
  volume={152 15},
An efficient and trajectory-free active learning method is proposed to automatically sample data points for constructing globally accurate reactive potential energy surfaces (PESs) using neural networks (NNs). Although NNs do not provide the predictive variance as the Gaussian process regression does, we can alternatively minimize the negative of the squared difference surface (NSDS) given by two different NN models to actively locate the point where the PES is least confident. A batch of… 

High-Fidelity Potential Energy Surfaces for Gas Phase and Gas-Surface Scattering Processes from Machine Learning.

This work reviews recent advances in constructing high-fidelity potential energy surfaces (PESs) from discrete ab initio points, using machine learning tools and focuses on inelastic and reactive scattering processes, which are more challenging than bound systems because of the involvement of continua.

Committee neural network potentials control generalization errors and enable active learning

This work adapts committee models to interatomic potentials based on artificial neural networks so that multiple models that share the same atomic environment descriptors yield an average that outperforms its individual members as well as a measure of the generalization error in the form of the committee disagreement.

Adversarial Attacks on Uncertainty Enable Active Learning for Neural Network Potentials

Adversarial attacks are new ways to simultaneously sample the phase space and bootstrap NN potentials, increasing their robustness and enabling a faster, accurate prediction of potential energy landscapes.

Machine Learning for Electronically Excited States of Molecules

This review focuses on not only how machine learning is employed to speed up excited-state simulations but also how this branch of artificial intelligence can be used to advance this exciting research field in all its aspects.

Machine learning and excited-state molecular dynamics

Recent advances for excited-state dynamics based on machine learning in quantum chemistry are surveyed to highlight successes, pitfalls, challenges and future avenues for machine learning approaches for light-induced molecular processes.

Active learning of potential-energy surfaces of weakly bound complexes with regression-tree ensembles.

This work proposes to use a regression version of stochastic query by forest, a hybrid method that samples points corresponding to large uncertainties while avoiding collecting too many points from sparse regions of space, which is fully automatic and does not require any prior knowledge of the PES.

Machine learning, artificial intelligence, and chemistry: How smart algorithms are reshaping simulation and the laboratory

This review provides an overview of machine learning and artificial intelligence from a chemist’s perspective and focuses on a number of examples of the use of these approaches in computational chemistry and in the laboratory.

Active Learning for Saddle Point Calculation

An active learning framework consisting of a statistical surrogate model, Gaussian process regression (GPR) for the energy function, and a single-walker dynamics method, gentle accent dynamics (GAD), for the saddle-type transition states to reduce the number of expensive computations of the true gradients.

2020 JCP Emerging Investigator Special Collection.

Gaussian process models of potential energy surfaces with boundary optimization.

A strategy is outlined to reduce the number of training points required to model intermolecular potentials using Gaussian processes, without reducing accuracy, by up to ∼49%, compared to the previous work based on a sequential learning technique.



Bridging the Gap between Direct Dynamics and Globally Accurate Reactive Potential Energy Surfaces Using Neural Networks.

This paper proposes this unified way to investigate gaseous and gas-surface reactions of medium size, initiating with hundreds of preliminary direct dynamics trajectories, followed by low-cost and high-quality simulations on full-dimensional analytical PESs.

Construction of reactive potential energy surfaces with Gaussian process regression: active data selection

ABSTRACT Gaussian process regression (GPR) is an efficient non-parametric method for constructing multi-dimensional potential energy surfaces (PESs) for polyatomic molecules. Since not only the

Constructing High-Dimensional Neural Network Potential Energy Surfaces for Gas–Surface Scattering and Reactions

While the ab initio molecular dynamics (AIMD) approach to gas–surface interaction has been instrumental in exploring important issues such as energy transfer and reactivity, it is only amenable to

Ab initio potential-energy surfaces for complex, multichannel systems using modified novelty sampling and feedforward neural networks.

A neural network/trajectory approach is presented for the development of accurate potential-energy hypersurfaces that can be utilized to conduct ab initio molecular dynamics (AIMD) and Monte Carlo

Active learning in Gaussian process interpolation of potential energy surfaces.

Three active learning schemes are used to generate training data for Gaussian process interpolation of intermolecular potential energy surfaces, and two notably outperform LHC designs of comparable size and produce an error value an order of magnitude lower than the one produced by the LHC method.

Representing Global Reactive Potential Energy Surfaces Using Gaussian Processes.

Tests for the 3A″ state of SH2 suggest that the Gaussian process is capable of providing a reasonable potential energy surface with a small number of ab initio points, but it needs substantially more points to converge reaction probabilities.

Bayesian optimization for the inverse scattering problem in quantum reaction dynamics

We propose a machine-learning approach based on Bayesian optimization to build global potential energy surfaces (PES) for reactive molecular systems using feedback from quantum scattering

Automating the Development of High-Dimensional Reactive Potential Energy Surfaces with the ROBOSURFER Program System.

The ROBOSURFER program system is presented that completely automates the generation of new geometries, performs ab-initio computations and iteratively improves the PES under development and evidence suggesting that the breakdown of single reference electronic structure theory may contribute significantly to the fitting errors of global reactive PESs is presented.

Communication: Fitting potential energy surfaces with fundamental invariant neural network.

The fundamental invariant neural network (FI-NN) method, which minimizes the size of input permutation invariant polynomials, can efficiently reduce the evaluation time of potential energy, in particular for polyatomic systems.

Permutation invariant polynomial neural network approach to fitting potential energy surfaces. II. Four-atom systems.

A rigorous, general, and simple method to fit global and permutation invariant potential energy surfaces (PESs) using neural networks (NNs) is discussed, resulting in full-dimensional global PESs with average errors on the order of meV.