Neural Network Wavefunction for Topological Insulator Detection

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The challenge of unraveling the complex quantum phases of topological insulators, particularly at larger scales and fractional fillings, is hindered by impractical computational scaling in existing methods like exact diagonalization and density matrix renormalization group, limiting the understanding of strong correlation systems.

Innovation Solution

A deep learning simulation method using neural networks to determine the wavefunction of a moiré system based on electron positions and spins, combined with quantum Monte Carlo techniques, to accurately identify topological insulators.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional methods like exact diagonalization and density matrix renormalization group are used to study topological insulators, then measurement precision can be achieved, but computational complexity becomes impractical at larger scales

Engineering Contradiction:

Improveaccuracy in identifying topological phasesVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces traditional mechanical/computational diagonalization methods with a neural network-based deep learning system. The neural network learns to identify topological phases directly from wavefunction data, substituting the computationally intensive exact diagonalization process with a trained model that provides similar or superior accuracy at reduced computational cost.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the fundamental parameters of the computational approach by using neural network architectures with adjustable weights and biases to represent wavefunctions. This allows the system to adapt to different system sizes and complexities without the computational scaling limitations of traditional methods.

Inventive Principle:

Principle #35Parameter changes

2Loss of information

If traditional methods are applied to larger scales and fractional fillings, then understanding of strong correlation systems can be improved, but computational resources required become impractical

Engineering Contradiction:

Improveunderstanding of quantum phasesVSAvoidcomputational resources

Core Design Contradiction:

Loss of informationVSUse of energy by moving object

Solution Approach 1:

The patent applies preliminary action by pre-training the neural network on representative quantum systems before applying it to new problems. The network learns general features of topological phases during training, enabling it to quickly analyze new systems without requiring full diagonalization computations for each case.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses the neural network as a computational copy or surrogate model that replicates the functionality of expensive diagonalization calculations. Once trained, the network can rapidly predict topological properties of new systems without requiring the same computational resources as the original training data generation process.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS20250209240A1Deep learning simulation of topological insulator

Publication Date: 2025.06.26 BEIJING YOUZHUJU NETWORK TECH CO LTD

AI summary

A method is proposed for a deep learning simulation of a topological insulator. The method includes determining, by using a neural network, a wavefunction of a moiré system based on positions and spins of electrons of the moiré system; and determining, based on the wavefunction, whether the moiré system is a topological insulator.