Gaussian Moments as Physically Inspired Molecular Descriptors for Accurate and Scalable Machine Learning Potentials.

@article{Zaverkin2020GaussianMA,
  title={Gaussian Moments as Physically Inspired Molecular Descriptors for Accurate and Scalable Machine Learning Potentials.},
  author={Viktor Zaverkin and Johannes K{\"a}stner},
  journal={Journal of chemical theory and computation},
  year={2020}
}
  • V. Zaverkin, J. Kästner
  • Published 16 July 2020
  • Physics, Mathematics, Medicine, Computer Science
  • Journal of chemical theory and computation
Machine learning techniques allow a direct mapping of atomic positions and nuclear charges to the potential energy surface with almost ab-initio accuracy and the computational efficiency of empirical potentials. In this work we propose a machine learning method for constructing high-dimensional potential energy surfaces based on feed-forward neural networks. As input to the neural network we propose an extendable invariant local molecular descriptor constructed from geometric moments. Their… Expand
16 Citations
Fast and Sample-Efficient Interatomic Neural Network Potentials for Molecules and Materials Based on Gaussian Moments.
TLDR
An improved NN architecture based on the previous GM-NN model is presented, which shows an improved prediction accuracy and considerably reduced training times and extends the applicability of Gaussian moment-based interatomic potentials to periodic systems and demonstrates the overall excellent transferability and robustness of the respective models. Expand
Linear Atomic Cluster Expansion Force Fields for Organic Molecules: Beyond RMSE.
TLDR
It is demonstrated that fast and accurate linear force fields can be built for molecules using the atomic cluster expansion (ACE) framework, and the four- or five-body ACE force fields improve on the accuracy of the empirical force fields by up to a factor of 10, reaching the accuracy typical of recently proposed machine-learning-based approaches. Expand
Exploration of transferable and uniformly accurate neural network interatomic potentials using optimal experimental design
TLDR
It is shown that the application of the proposed active learning scheme leads to transferable and uniformly accurate potential energy surfaces constructed using only a small fraction of data points, and it is possible to define a natural threshold value for the proposed uncertainty metric which offers the possibility to generate highly informative training data on-the-fly. Expand
Neural Network Potential Energy Surfaces for Small Molecules and Reactions.
TLDR
This work considers NN-based approaches to build PESs in the sums-of-product form important for quantum dynamics, ways to treat symmetry, and issues related to sampling data distributions and the relation between PES errors and errors in observables. Expand
Strategies for the construction of machine-learning potentials for accurate and efficient atomic-scale simulations
Recent advances in machine-learning interatomic potentials have enabled the efficient modeling of complex atomistic systems with an accuracy that is comparable to that of conventionalExpand
Linear Atomic Cluster Expansion Force Fields for Organic Molecules: beyond RMSE
We demonstrate that accurate linear force fields can be built using the Atomic Cluster Ex- pansion (ACE) framework for molecules. Our model is built from body ordered symmetric polynomials whichExpand
Linear Atomic Cluster Expansion Force Fields for Organic Molecules: beyond RMSE
We demonstrate that accurate linear force fields can be built using the Atomic Cluster Expansion (ACE) framework for molecules. Our model is built from body ordered symmetric polynomials which makesExpand
Modeling electronic response properties with an explicit-electron machine learning potential
Explicit-electron force fields introduce electrons or electron pairs as semi-classical particles in force fields or empirical potentials, which are suitable for molecular dynamics simulations. EvenExpand
Choosing the right molecular machine learning potential
TLDR
This work evaluates the performance of popular machine learning potentials in terms of accuracy and computational cost, and delivers structured information for non-specialists in machine learning to guide them through the maze of acronyms, recognize each potential's main features, and judge what they could expect from each one. Expand
Gaussian Process Regression for Materials and Molecules
TLDR
The focus of the present review is on the regression of atomistic properties: in particular, on the construction of interatomic potentials in the Gaussian Approximation Potential (GAP) framework; beyond this, the fitting of arbitrary scalar, vectorial, and tensorial quantities is discussed. Expand
...
1
2
...

References

SHOWING 1-10 OF 55 REFERENCES
Embedded Atom Neural Network Potentials: Efficient and Accurate Machine Learning with a Physically Inspired Representation.
TLDR
This work demonstrates that the so-called embedded atom neural network approach is equally accurate as several established ML models in representing both big molecular and extended periodic systems, yet with much fewer parameters and configurations. Expand
A novel approach to describe chemical environments in high-dimensional neural network potentials.
TLDR
A set of invariant, orthogonal, and differentiable descriptors for an atomic environment is proposed, implemented in a neural network potential for solid-state silicon, and tested in molecular dynamics simulations. Expand
PhysNet: A Neural Network for Predicting Energies, Forces, Dipole Moments, and Partial Charges.
TLDR
PhysNet is introduced, a DNN architecture designed for predicting energies, forces, and dipole moments of chemical systems, and it is shown that explicitly including electrostatics in energy predictions is crucial for a qualitatively correct description of the asymptotic regions of a potential energy surface (PES). Expand
Machine learning of accurate energy-conserving molecular force fields
TLDR
The GDML approach enables quantitative molecular dynamics simulations for molecules at a fraction of cost of explicit AIMD calculations, thereby allowing the construction of efficient force fields with the accuracy and transferability of high-level ab initio methods. Expand
Fourier series of atomic radial distribution functions: A molecular fingerprint for machine learning models of quantum chemical properties
We introduce a fingerprint representation of molecules based on a Fourier series of atomic radial distribution functions. This fingerprint is unique (except for chirality), continuous, andExpand
A reactive, scalable, and transferable model for molecular energies from a neural network approach based on local information.
TLDR
A data-driven method to construct a potential energy surface based on neural networks is presented, which is accurate across chemical and configurational space and demonstrated by applying it to datasets from nonreactive and reactive molecular dynamics simulations and a diverse database of equilibrium structures. Expand
Machine learning of molecular properties: Locality and active learning.
TLDR
A new machine learning algorithm for predicting molecular properties that is based on a local model of interatomic interactions providing high accuracy when trained on relatively small training sets and an active learning algorithm of optimally choosing the training set that significantly reduces the errors for the outliers is proposed. Expand
Quantum-chemical insights from deep tensor neural networks
TLDR
An efficient deep learning approach is developed that enables spatially and chemically resolved insights into quantum-mechanical observables of molecular systems, and unifies concepts from many-body Hamiltonians with purpose-designed deep tensor neural networks, which leads to size-extensive and uniformly accurate chemical space predictions. Expand
SchNet - A deep learning architecture for molecules and materials.
TLDR
The deep learning architecture SchNet is presented that is specifically designed to model atomistic systems by making use of continuous-filter convolutional layers and employs SchNet to predict potential-energy surfaces and energy-conserving force fields for molecular dynamics simulations of small molecules. Expand
Towards exact molecular dynamics simulations with machine-learned force fields
TLDR
A flexible machine-learning force-field with high-level accuracy for molecular dynamics simulations is developed, for flexible molecules with up to a few dozen atoms and insights into the dynamical behavior of these molecules are provided. Expand
...
1
2
3
4
5
...