Hybrid Quantum-Classical Neural Network for DFT Functional Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current density functional theory (DFT) methods for quantum chemistry modeling face challenges in achieving accurate results, especially for strongly correlated systems, due to reliance on approximations and high computational costs, and are limited by the quality and availability of experimental data for training machine learning models.
Innovation Solution
A hybrid quantum-classical computing method that uses quantum processing units to generate high-quality training data for neural networks, enabling more accurate and efficient determination of DFT functionals, which can model complex physical systems that are impractical for classical computing systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional computational methods are used for strongly correlated systems, then computational resources and time are significantly consumed, but accurate results can be obtained
Solution Approach 1:
The patent introduces a neural network as an intermediary between quantum processing and classical DFT implementation. The neural network is trained on quantum-processed data to learn accurate electron correlation effects, then applies these learned patterns efficiently to new systems, mediating between the accuracy of quantum methods and the speed of classical computations.
Solution Approach 2:
The patent performs preliminary quantum processing to generate training data for the neural network. By pre-computing accurate reference data using quantum processing for various electronic densities and configurations, the system prepares a trained neural network model that can then rapidly predict DFT functional parameters without requiring real-time quantum computations for each new system.
2Measurement precision
If machine learning techniques trained on experimental data are used, then DFT accuracy can be improved, but the data quality and quantity are insufficient
Solution Approach 1:
The patent uses quantum processing as an intermediary to generate high-quality training data. Instead of relying on noisy experimental data or inaccurate classical simulations, the quantum processor acts as a mediator that produces accurate reference data for electronic densities and energies, which then serves as high-quality training data for the neural network.
Solution Approach 2:
The patent substitutes experimental measurement systems and classical simulation systems with quantum processing systems for data generation. Quantum processing replaces the mechanical and chemical processes of physical experimentation with quantum mechanical computations that can be precisely controlled and repeated, generating abundant high-quality training data without experimental limitations.
3Productivity
If approximations to the universal functional are used in DFT, then computational cost is reduced, but accuracy deteriorates in many situations
Solution Approach 1:
The patent changes the parameters of the DFT functional by using a neural network to dynamically determine functional parameters based on the electronic density. Instead of using fixed approximate functionals, the neural network learns optimal functional parameters from quantum-processed training data and adapts these parameters to each specific system, improving accuracy while maintaining computational efficiency.
Solution Approach 2:
The patent creates a copied model of the complex quantum many-body problem through the neural network. The neural network learns the complex electron correlation patterns from quantum-processed training data and creates a simplified computational model that copies these patterns, allowing accurate DFT functional determination without directly solving the full quantum many-body problem for each new system.
Data Source
AI summary
There is described a hybrid quantum-classical computing method and system that leverages quantum processing to generate data that enables the training of neural networks for the purpose of density functional theory (DFT) functional determination. Physical systems are modeled on a quantum processing module and simulated to generate energy values and electronic density functions as training data with sufficient degrees of quality and accuracy. The training data may be in classical form and are used to train a neural network. The trained neural network may then be employed to parameterize DFT functionals.