Convolutional Neural Network of Atomic Surface Structures to Predict Binding Energies for High-Throughput Screening of Catalysts

We present an application of deep-learning convolutional neural network of atomic surface structures using atomic and Voronoi polyhedra-based neighbor information to predict adsorbate binding energies for the application in catalysis.

Convolutional Neural Network of Atomic Surface Structures to Predict Binding Energies for High-Throughput Screening of Catalysts

We present an application of deep-learning convolutional neural network of atomic surface structures using atomic and Voronoi polyhedra-based neighbor information to predict adsorbate binding energies for the application in catalysis.

Компьютерное моделирование структуры и механических свойств слоев силицена на графите при движении иона лития

The molecular dynamics method is applied to study structural and mechanical effects appearing during the lithium ion motion in a dc electric field along a planar channel formed by perfect silicene sheets and sheets containing vacancy-type defects. Mono-, di-, tri-, and hexavacancies of rather densely and uniformly filled silicene sheets are arranged one above the other on a graphite substrate. The times of Li^+ ion passage through silicene channels with various gaps are determined. The construction of Voronoi polyhedra and truncated polyhedrons, whose centers coincide with the moving ion position allowed revealing the structural features inherent to the two-dimensional layered structure. The nature of stresses appearing in silicene sheets most critical to ion motion over the channel is determined.

The complete classification of five-dimensional Dirichlet–Voronoi polyhedra of translational lattices

This paper reports on the full classification of Dirichlet–Voronoi polyhedra and Delaunay subdivisions of five-dimensional translational lattices. A complete list is obtained of 110 244 affine types (L-types) of Delaunay subdivisions and it turns out that they are all combinatorially inequivalent, giving the same number of combinatorial types of Dirichlet–Voronoi polyhedra. Using a refinement of corresponding secondary cones, 181 394 contraction types are obtained. The paper gives details of the computer-assisted enumeration, which was verified by three independent implementations and a topological mass formula check.

The local environment of the molecules in water–DMSO mixtures, as seen from computer simulations and Voronoi polyhedra analysis

The local structure of DMSO–water mixtures is studied by computer simulation and Voronoi analysis.

Voronoi polyhedra probing of hydrated OH radical

Voronoi polyhedron method is employed to extract the smallest volume shared by ˙OH radical in liquid water at the biologically important temperature (37 °C). The 3D-visualization and the probability distributions of the metric and topological properties of ˙OH solvation cage are provided.