Neural Network Wave Function Determination for Solid System Polarization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calculating the electric polarization of solid systems are limited in precision and scale, necessitating a more efficient and accurate approach to determine this critical property related to electric effects such as ferroelectric and piezoelectric effects.
Innovation Solution
A method utilizing a neural network to determine the wave function of a solid system by minimizing an objective function based on enthalpy in the presence of an electric field, allowing for the calculation of electric polarizability, which improves processing efficiency and precision by considering the periodicity of the solid system.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional quantum mechanical methods (Hartree-Fock, density functional theory, coupled-cluster theory) are used to calculate electric polarization, then measurement precision can be achieved, but device complexity and computational cost increase significantly
Solution Approach 1:
The patent replaces traditional mechanical quantum mechanical calculations with a neural network-based computational model. The neural network is trained to predict electric polarization properties directly from crystal structure data, substituting complex many-body quantum mechanical equations with a data-driven approach that maintains high accuracy while significantly reducing computational complexity
Solution Approach 2:
The patent changes the fundamental parameters used in calculations by transitioning from wave function-based quantum mechanical parameters to machine learning parameters. The neural network model uses crystal structure parameters (atomic positions, lattice vectors) as inputs and predicts polarization properties directly, avoiding the need for complex many-body wave function calculations
2Measurement precision
If traditional quantum mechanical methods are used to calculate electric polarization, then measurement precision can be achieved, but processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by pre-training the neural network model on a comprehensive dataset of crystal structures and their corresponding electric polarization properties. This pre-training phase allows the model to learn complex relationships between crystal structure and polarization properties, enabling fast predictions during actual calculations without requiring time-consuming quantum mechanical computations for each new structure
Solution Approach 2:
The patent uses copying by creating a simplified computational model (neural network) that replicates the essential physics of electric polarization in solid systems. Instead of solving the full many-body quantum mechanical problem, the neural network copies the input-output relationships from training data to provide accurate predictions much faster than traditional methods
3Measurement precision
If conventional calculation methods are used for solid systems, then existing results can be obtained, but the scale and precision are limited
Solution Approach 1:
The patent applies segmentation by focusing calculations on the unit cell structure of crystalline materials rather than attempting to calculate the entire bulk system. The neural network model processes crystal structure information at the unit cell level, leveraging the periodicity of crystalline materials to achieve high precision for bulk properties without requiring proportional increases in computational resources with system size
Data Source
AI summary
Embodiments of the present disclosure relate to a method and an apparatus for determining an electric polarization of a solid system, an electronic device, a computer-readable storage medium, and a computer program product. The method includes: determining a wave function of the solid system by inputting electron coordinates of a periodic unit of the solid system into a neural network and by minimizing an objective function, where the objective function is determined based on an enthalpy in the presence of an electric field; and determining an electric polarizability of the solid system based on the wave function of the solid system.