Moment Tensor Potentials: A Class of Systematically Improvable Interatomic Potentials

By Alexander V. Shapeev. 2015

1,056 citations recorded by the source corpus.

Density functional theory offers a very accurate way of computing materials properties from first principles. However, it is too expensive for modelling large-scale molecular systems whose properties are, in contrast, computed using interatomic potentials. The present paper considers, from a mathematical point of view, the problem of constructing interatomic potentials that approximate a given quantum-mechanical interaction model. In particular, a new class of systematically improvable potentials is proposed, analyzed, and tested on an existing quantum-mechanical database.

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

Cormorant: Covariant Molecular Neural Networks (2019, background): “…community has formed around trying to use machine learning to learn more descriptive force ﬁelds directly from DFT computations [Behler and Parrinello, 2007, Bartók et al., 2010, Rupp et al., 2012, Shapeev, 2015, Chmiela et al., 2016, Zhang et al., 2018, Schütt et al., 2017, Hirn et al., 2017].”
N-body Networks: a Covariant Hierarchical Neural Network Architecture for Learning Atomic Potentials (2018, background): “The closest to our work in certai ways are Moment Tensor Potientials (Shapeev, 2015), although that framework does not have learnable parameters.”
Machine learning for heterogeneous catalyst design and discovery (2018, result): “72 Therefore, these machine-learned interatomic potentials (MLPs) can speed-up simulations by several orders of magnitude while keeping comparable accuracy to QM methods.”
Machine Learning a General-Purpose Interatomic Potential for Silicon (2018, methodology): “78 E. V. Podryabinkin and A. V. Shapeev, Comp Mater Sci 140, 171 (2017).”
Machine Learning for High-entropy Alloys: Progress, Challenges and Opportunities (2022, background): “This feature renders the MTP method attractive for the high-throughput screening of multicomponent alloys compared to the traditional DFT methods.”
Deep Potentials for Materials Science (2022, methodology): “Apart from NNP and SchNet, several other types of ML potentials were introduced; Gaussian approximation potentials (GAP) [30– 33], moment tensor potentials (MTP) [34–36], spectral neighbour analysis potentials (SNAP) [37–39], gradient-domain machine learning (GDML) [40, 41], et al [42].”
Nanoporous C3N4, C3N5 and C3N6 nanosheets; novel strong semiconductors with low thermal conductivities and appealing optical/electronic properties (2020, methodology): “Moment tensor potentials (MTPs) [37,38] were employed as an accurate and computationally efficient model of describing interatomic interaction.”
The design space of E(3)-equivariant atom-centred interatomic potentials (2022, neutral): “The model maps a set of atoms with element types positioned in the three-dimensional Euclidean space to the total potential energy.”
Learning local equivariant representations for large-scale atomistic dynamics (2022, methodology): “MLIPs using strictly local descriptors such as Behler-Parrinello neural networks [5], GAP [6], SNAP [7], DeepMD [20], Moment Tensor Potentials [8], or ACE [12] do not suffer from this obstacle due to their strict locality.”