Representing potential energy surfaces by high-dimensional neural network potentials

By J. Behler. 2014

385 citations recorded by the source corpus.

Subfields: Atomic and Molecular Physics, and Optics, Materials Chemistry, Physical and Theoretical Chemistry

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Atomic and Molecular Physics, and Optics

In Materials Chemistry

In Physical and Theoretical Chemistry

Citations in the corpus

Fast and Accurate Uncertainty Estimation in Chemical Machine Learning. (2018, result): “This is in stark contrast to similar approaches based on neural networks 16,38 in which the evaluation of multiple models entails a substantial overhead.”
Fracture of graphene: a review (2015, background): “It can be a very promising direction to applymodern techniques ofmachine learning in the development of accurate interatomic potentials for complicated atomic systems (Behler 2014).”
Many-Body Coarse-Grained Interactions Using Gaussian Approximation Potentials. (2016, background): “This is already an approximation in all-atom simulations [24], where ongoing research is studying extensions such as polarisable force fields [111, 112] and more general many-body approaches [30, 113].”
Machine Learning for High-entropy Alloys: Progress, Challenges and Opportunities (2022, background): “Since the neural networks are applied to every atom, NNP is also referred to as high-dimensional NNP (HDNNP) [146].”
How van der Waals interactions determine the unique properties of water (2016, background): “The form of the potentials is not constructed using simplified physically motivated models, but instead consists of a set of highly flexible functions in the form of artificial neural networks (25, 26) trained to a broad range of condensed phase configurations.”
Active learning of linearly parametrized interatomic potentials (2016, methodology): “It is hardly expected that a MLIP can extrapolate beyond the training domain, but even developing a reliable problem-speciﬁc MLIP that would accurately interpolate within the training domain is nontrivial, as pointed out, for instance, by Behler [9, Section 4].”
Machine Learning and Materials Informatics: Recent Applications and Prospects (2017, background): “Several candidates, including those based on symmetry functions [76–78], bispectra of neighborhood atomic densities [79], Coulomb matrices (and its variants) [80, 81], smooth overlap of atomic positions (SOAP) [82–85], and others [86,87], have been proposed.”
Moment Tensor Potentials: A Class of Systematically Improvable Interatomic Potentials (2015, methodology): “In this context, one common approach to constructing nonparametric potentials, called the neural networks potentials (NNP) [7, 8], is using artificial NNs as the regression model and a family of descriptors first proposed by Behler and Parrinello [8].”