Abstract:Based on the Hermite expansion of the distribution function, we introduce a Galerkin spectral method for the spatially homogeneous Boltzmann equation with the realistic inverse-power-law models. A practical algorithm is proposed to evaluate the coefficients in the spectral method with high accuracy, and these coefficients are also used to construct new computationally affordable collision models. Numerical experiments show that our method captures the low-order moments very efficiently.
What problem does this paper attempt to address?
The core problem that this paper attempts to solve is to improve the efficiency and accuracy of the numerical simulation of the Boltzmann collision operator, especially for the spatially homogeneous Boltzmann equation with the actual inverse - power - law model. Specifically, the author proposes the Galerkin spectral method based on the Hermite spectral method and develops an efficient algorithm to calculate the coefficients in the spectral method. These coefficients are not only used for discretizing the collision operator but also for constructing new collision models with lower computational costs.
### Main problems and solutions in the paper
1. **Complexity of high - dimensional integrals**:
- The Boltzmann equation is a six - dimensional problem, and its collision operator involves complex integral operations, which makes numerical simulation very difficult.
- The author uses the Hermite spectral method to avoid the need to periodize the distribution function, thus providing a more natural way to deal with this problem.
2. **Challenges in calculating coefficients**:
- In the Hermite spectral method, calculating the expansion coefficients of the collision operator is a major problem because these coefficients involve the expansion of a large number of polynomials and the computational cost is extremely high.
- By using the properties of polynomials, the author derives explicit formulas for all coefficients. These formulas can be directly used for programming implementation, and for a moderate number of degrees of freedom, the computational cost is acceptable.
3. **Optimization of computational cost**:
- The time complexity of directly calculating the collision operator is the cube of the number of degrees of freedom, while the time complexity of the Fourier spectral method is the square of the number of modes.
- To reduce the computational cost, the author proposes a new discretization method, in which only a part of the truncated series expansion is "quadratically" processed, and the rest decays exponentially in the way of the BGK model. This method greatly reduces the computational cost while still being able to accurately capture the evolution of low - order moments.
### Summary of mathematical expressions
- **Boltzmann equation**:
\[
\frac{\partial f}{\partial t}+\nabla_x\cdot(v f) = Q[f],\quad t\in\mathbb{R}^+,x\in\mathbb{R}^3,v\in\mathbb{R}^3
\]
where \(Q[f]\) is the collision operator and has a quadratic form:
\[
Q[f](t,x,v)=\int_{\mathbb{R}^3}\int_{n\perp g}\int_0^\pi [f(t,x,v')f(t,x,v'_1)-f(t,x,v)f(t,x,v_1)]B(|g|,\chi)d\chi dn dv_1
\]
- **Hermite expansion**:
\[
f(t,v)=\sum_{k_1 k_2 k_3}f_{k_1 k_2 k_3}(t)H_{k_1 k_2 k_3}(v)M(v)
\]
where \(H_{k_1 k_2 k_3}(v)\) is the Hermite polynomial and \(M(v)\) is the Maxwellian distribution.
- **Expansion coefficients of the collision operator**:
\[
A^{k_1 k_2 k_3}_{i_1 i_2 i_3,j_1 j_2 j_3}=\sum_{m_1 = 0}^{\lfloor k_1/2\rfloor}\sum_{m_2 = 0}^{\lfloor k_2/2\rfloor}\sum_{m_3 = 0}^{\lfloor k_3/2\rfloor}\sum_{n_1 = 0}^{\lfloor l_1/2\rfloor}\sum_{n_2 = 0}^{\lfloor l_2/2\rfloor}\sum_{n_3 = 0}^{\lfloor l_3/2\rfloor}(2k - 4m+ 1)C^{k_1 k_2 k_3}_{m_1 m_2 m_3}C^{l_1 l_2 l_3}_{n_1 n_2 n_3}
\]