APTT: An accuracy-preserved tensor-train method for the Boltzmann-BGK equation

Zhitao Zhu,Chuanfu Xiao,Kejun Tang,Jizu Huang,Chao Yang

2024-05-21

Abstract:Solving the Boltzmann-BGK equation with traditional numerical methods suffers from high computational and memory costs due to the curse of dimensionality. In this paper, we propose a novel accuracy-preserved tensor-train (APTT) method to efficiently solve the Boltzmann-BGK equation. A second-order finite difference scheme is applied to discretize the Boltzmann-BGK equation, resulting in a tensor algebraic system at each time step. Based on the low-rank TT representation, the tensor algebraic system is then approximated as a TT-based low-rank system, which is efficiently solved using the TT-modified alternating least-squares (TT-MALS) solver. Thanks to the low-rank TT representation, the APTT method can significantly reduce the computational and memory costs compared to traditional numerical methods. Theoretical analysis demonstrates that the APTT method maintains the same convergence rate as that of the finite difference scheme. The convergence rate and efficiency of the APTT method are validated by several benchmark test cases.

Numerical Analysis

What problem does this paper attempt to address?

The paper primarily addresses the high computational cost and memory consumption issues encountered in solving the Boltzmann-BGK equation by proposing a novel method—Accuracy-Preserved Tensor-Train (APTT) method.

The Boltzmann-BGK equation is a high-dimensional partial differential equation. Traditional numerical methods (such as finite element or finite difference methods) face the problem of the curse of dimensionality when dealing with such problems, meaning that as the grid is refined, the computational complexity and memory requirements increase dramatically. To solve this problem, the paper presents the following contributions:

1. **Proposal of the new method**: The authors applied the Crank-Nicolson Leap Frog (CNLF) scheme and second-order upwind scheme to discretize the Boltzmann-BGK equation, resulting in a complete tensor algebra system. Then, they used the low-rank Tensor-Train (TT) representation to approximate the full tensor in this algebraic system, thereby significantly reducing computational and memory costs.

2. **Low-rank representation and accuracy preservation**: In the APTT method, the key is to construct a low-rank TT representation of the collision term. The high-order tensor corresponding to the collision term is decomposed into the product of several lower-order and low-rank TT format tensors. Based on this low-rank representation, the original tensor algebra system is approximated as a TT-based low-rank linear system, which is efficiently solved using methods such as TT-Modified Alternating Least Squares (TT-MALS) or Density Matrix Renormalization Group (DMRG).

3. **Theoretical analysis**: The paper also provides an error analysis of the APTT method, proving that the method can maintain the same convergence rate as the discretization scheme while satisfying the conservation laws of mass, momentum, and energy within a preset accuracy tolerance.

4. **Numerical experiment validation**: The convergence rate and efficiency of the APTT method are validated through multiple benchmark test cases.

In summary, this paper aims to develop a new method that can effectively reduce the computational resource consumption when solving the Boltzmann-BGK equation while ensuring the accuracy of the results.