On representing chemical environments
By A. P. Bart'ok, Risi Kondor, Gábor Csányi. 2012
1,908 citations recorded by the source corpus.
We review some recently published methods to represent atomic neighborhood environments, and analyze their relative merits in terms of their faithfulness and suitability for fitting potential energy surfaces. The crucial properties that such representations (sometimes called descriptors) must have are differentiability with respect to moving the atoms and invariance to the basic symmetries of physics: rotation, reflection, translation, and permutation of atoms of the same species. We demonstrate that certain widely used descriptors that initially look quite different are specific cases of a general approach, in which a finite set of basis functions with increasing angular wave numbers are used to expand the atomic neighborhood density function. Using the example system of small clusters, we quantitatively show that this expansion needs to be carried to higher and higher wave numbers as the number of neighbors increases in order to obtain a faithful representation, and that variants of the descriptors converge at very different rates. We also propose an altogether different approach, called Smooth Overlap of Atomic Positions, that sidesteps these difficulties by directly defining the similarity between any two neighborhood environments, and show that it is still closely connected to the invariant descriptors. We test the performance of the various representations by fitting models to the potential energy surface of small silicon clusters and the bulk crystal.
Subfields: Atomic and Molecular Physics, and Optics, Materials Chemistry, Computational Chemistry & Drug Discovery
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
- Alexander V. Shapeev
- Yuri Mishin
- Evgeny V. Podryabinkin
- Gábor Csányi
- Risi Kondor
- K. Müller
- Rampi Ramprasad
- Rohit Batra
In Materials Chemistry
- Yuri Mishin
- Tess E. Smidt
- Alexander V. Shapeev
- Risi Kondor
- K. Müller
- Gábor Csányi
- Michael Gastegger
- Kristof T. Schütt
In Computational Chemistry & Drug Discovery
- Tess E. Smidt
- K. Müller
- Michael Gastegger
- Kristof T. Schütt
- Risi Kondor
- Stephan Günnemann
- Johannes Klicpera
- Oliver T. Unke
Citations in the corpus
- Directional Message Passing for Molecular Graphs (2020, methodology): “The classical way of using machine learning for predicting molecular properties is combining an expressive, hand-crafted representation of the atomic neighborhood (Bartók et al., 2013) with Gaussian processes (Bartók et al.”
- SchNet: A continuous-filter convolutional neural network for modeling quantum interactions (2017, methodology): “Previous work has used neural networks and Gaussian processes applied to hand-crafted features to fit potential energy surfaces [11–16].”
- Cormorant: Covariant Molecular Neural Networks (2019, methodology): “Note the similarity of this expression to the bispectrum [Kakarala, 1992, Bendory et al., 2018], which is an already established tool in the force field learning literature [Bartók et al., 2013].”
- Machine learning unifies the modeling of materials and molecules (2017, positive): “This kernel is equivalent to the scalar product of the spherical power spectra of the neighbor density ( 39 ), which therefore constitutes a chemical descriptor of the neighbor environment.”
- Wavelet Scattering Regression of Quantum Chemical Energies (2016, methodology): “Deformation stability: f(x) is Lipschitz continuous to variations of the distances |rk − rl| between atoms and hence Lipschitz continuous to diffeomorphism actions [5].”
- Equivariant message passing for the prediction of tensorial properties and molecular spectra (2021, background): “A common inductive bias of the neural network potential is a decomposition of the energy into atomwise contributions E = =1 ( s i ), where an output network (cid:15) predicts energy contributions from atoms embedded within their chemical environment, represented by s i ∈ R F 21,37 .”
- Spectral neighbor analysis method for automated generation of quantum-accurate interatomic potentials (2014, methodology): “ they are not invariant under rotation of the polar coordinate frame. However, the following scalar triple products of expansion coecients can be shown to be real-valued and invariant under rotation [7]. B j 1;j 2;j = Xj 1 m 1;m0 1 = j 1 Xj 2 m 2;m0 2 = j 2 Xj m;m0= j (uj m;m 0 ) H jmm0 j 1mm0 1 j 2”