Abstract:In this paper, we introduce a type of tensor neural network. For the first time, we propose its numerical integration scheme and prove the computational complexity to be the polynomial scale of the dimension. Based on the tensor product structure, we develop an efficient numerical integration method by using fixed quadrature points for the functions of the tensor neural network. The corresponding machine learning method is also introduced for solving high-dimensional problems. Some numerical examples are also provided to validate the theoretical results and the numerical algorithm.
What problem does this paper attempt to address?
The problem this paper attempts to address is the numerical solution of high-dimensional partial differential equations (PDEs). Specifically, traditional numerical methods encounter the "curse of dimensionality" when dealing with high-dimensional PDEs, meaning that as the dimensionality of the problem increases, the computational complexity grows exponentially, making it difficult for traditional numerical methods to effectively solve high-dimensional PDEs. To solve this problem, the paper introduces a new Tensor Neural Network (TNN) structure and proposes an efficient numerical integration method.
### Main Contributions:
1. **Tensor Neural Network (TNN)**: The paper introduces the concept of TNN for the first time, which is a neural network based on tensor product structures that can be used to construct trial functions for high-dimensional PDEs.
2. **Numerical Integration Method**: The paper proposes an efficient numerical integration method that calculates the integral of the TNN function by fixing integration points. This method avoids the low convergence rate and computational instability issues of traditional Monte Carlo methods, making the numerical solution of high-dimensional PDEs more efficient and stable.
3. **Computational Complexity**: The paper proves that the computational complexity of the numerical integration of TNN is only polynomial in the dimensionality, rather than exponential, thereby overcoming the "curse of dimensionality" to a certain extent.
### Specific Problems Addressed:
- **Numerical Solution of High-Dimensional PDEs**: Traditional methods have high computational complexity when dealing with high-dimensional PDEs, making it difficult to obtain high-precision and stable solutions. TNN, through low-rank approximation and efficient numerical integration methods, can effectively solve PDEs in high-dimensional cases.
- **Efficient Computation of High-Dimensional Integrals**: Traditional methods such as the Monte Carlo method have slow convergence rates and low computational efficiency in high-dimensional integrals. TNN, through tensor product structures and fixed integration points, can efficiently perform high-dimensional integral computations.
### Application Areas:
- **Physics**: Solving high-dimensional PDEs such as the many-body Schrödinger equation and the Boltzmann equation.
- **Chemistry**: High-dimensional problems in fields such as reaction dynamics and molecular dynamics.
- **Biology**: Modeling and simulation of biological systems.
- **Engineering**: Solving high-dimensional PDEs in fields such as fluid mechanics and materials science.
In summary, by introducing TNN and an efficient numerical integration method, this paper provides a new and effective solution for the numerical solution of high-dimensional PDEs.