Cluster energy prediction method and system

Abstract

The invention provides a cluster energy prediction method, and the method comprises the steps: obtaining all atomic coordinates of a nonmetal cluster, unfolding the obtained atomic coordinates, and carrying out the preprocessing of the unfolded atomic coordinate data; obtaining an energy prediction result of the nonmetal cluster according to the preprocessed atomic coordinate data and a preset multilayer neural network model, wherein a mean square error loss function is adopted in the preset multilayer neural network model; obtaining all atomic coordinates of the metal cluster, expanding the obtained atomic coordinates, and preprocessing the expanded atomic coordinate data; and according to the preprocessed atomic coordinate data and a preset polynomial regression prediction model, obtaining an energy prediction result of the metal cluster. The method achieves the more accurate prediction of the energy of the current clusters of different structures, enables the atomic coordinates to be used for polynomial regression and neural network prediction through the transverse expansion of the atomic coordinates, and further improves the prediction precision.

Description

technical field [0001] The present disclosure relates to the technical field of cluster energy prediction, in particular to a cluster energy prediction method and system. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. [0003] The basic problem of cluster science research is to determine how atoms and molecules evolve into clusters, and how the structure and properties of clusters change accordingly. Structure is the key factor determining the performance of clusters. Novel and unique structures often have novel properties. Therefore, the prediction of cluster energy and structure has become an important issue in the study of clusters. [0004] Clusters can be classified into metallic and nonmetallic clusters, and metallic nanoclusters exhibit fundamentally different properties from their plasmonic counterparts due to quantum size effects as we...

AI Technical Summary

Problems solved by technology

[0005] The inventors found that metal nanoclusters have attracted much attention due to their good optical, catalytic, chiral, magnetic, electrochemical and other properties. However, because the potential energy surface of the cluster is too complex, and sometimes relativistic effects need to be considered, the search group The globally optimal structure (i.e., the lowest energy) of clusters is particularly difficult

Among them, the traditional theoretical calculation method needs numerical iteration to solve the Schrödinger equation, and as the number of atoms increases, the high-precision theoretical calculation time increases exponentially, which is very time-consuming

Images