Asymptotic-preserving semi-Lagrangian discontinuous Galerkin schemes for the Boltzmann equation

Basic Information

• Paper ID: 2510.14375
• Title: Asymptotic-preserving semi-Lagrangian discontinuous Galerkin schemes for the Boltzmann equation
• Authors: Xiaofeng Cai, Zhen Hao, Liu Liu, Jiayu Wan
• Classification: math.NA cs.NA
• Submission Date: October 16, 2025
• Paper Link: https://arxiv.org/abs/2510.14375

Abstract

This paper proposes an asymptotic-preserving semi-Lagrangian discontinuous Galerkin (AP-SLDG) scheme for the Boltzmann equation that effectively handles multiscale transport phenomena. The main challenge lies in designing appropriate moment update mechanisms for penalty methods within the semi-Lagrangian framework. Inspired by the work of Ding et al., the key technique is to utilize the Shu-Osher formulation in the implicit-explicit Runge-Kutta (IMEX-RK) setting, capturing the correct limiting system through carefully constructed moment update procedures. Theoretical analysis establishes accuracy order conditions for IMEX-RK time integration and the new moment update steps, and employs coercivity techniques to establish stability of the linearized model. Numerical experiments verify the accuracy, asymptotic-preserving properties, and robustness of the proposed scheme on various test problems.

Research Background and Motivation

Problem Background

1. Importance of the Boltzmann equation: The Boltzmann equation is the fundamental kinetic equation describing rarefied gas dynamics, correctly describing gas flow physics in regimes where continuum models such as the Navier-Stokes equations fail.
2. Computational Challenges:
   • Computational complexity arising from high-dimensional phase space
   • Nonlinear and nonlocal nature of the collision operator
   • Stiffness problems caused by multiscale characteristics
3. Limitations of Existing Methods:
   • Traditional Eulerian methods are restricted by CFL conditions, limiting time step size
   • For small Knudsen numbers ε, explicit methods require strict time step restrictions of ∆t = O(ε)
   • Semi-Lagrangian methods, while avoiding CFL restrictions, face new challenges in designing asymptotic-preserving schemes

Research Motivation

To develop numerical schemes that maintain accuracy and efficiency across all physical scales, particularly schemes that automatically reduce to the correct Euler equations in the fluid limit (ε→0).

Core Contributions

1. Innovative moment update mechanism: Designs a new moment update procedure within the semi-Lagrangian framework, resolving the failure of traditional Eulerian moment update strategies in the SL framework
2. Application of Shu-Osher formulation: Adapts the Shu-Osher formulation to the IMEX-RK setting, enabling correct identification of the limiting system and construction of corresponding moment updates
3. Theoretical Analysis:
   • Establishes accuracy order conditions for IMEX-RK schemes
   • Proves asymptotic-preserving properties
   • Analyzes stability of the linearized model using coercivity techniques
   • Provides sufficient conditions for positivity preservation
4. Numerical Verification: Validates the scheme's accuracy, asymptotic-preserving properties, and robustness through multiple test problems

Methodology Details

Problem Formulation

Solve the Boltzmann equation: $\partial_t f + v \cdot \nabla_x f = \frac{1}{\varepsilon} Q(f)$

where f(x,v,t) is the particle distribution function, ε is the Knudsen number, and Q(f) is the collision operator. The goal is to design numerically efficient and accurate schemes for all values of ε.

Model Architecture

1. BGK Penalty Method

Employs BGK penalty decomposition: $Q(f) = Q(f) - \beta Q_{BGK}(f) + \beta Q_{BGK}(f)$ where $Q_{BGK}(f) = M_f - f$ and $M_f$ is the Maxwellian distribution associated with f.

2. Semi-Lagrangian Discretization

Based on characteristic Galerkin weak formulation: $\int_{I} f(t_2,x,v)\Psi(x)dx = \int_{I-v\Delta t} f_0(x,v)\Psi(x+v\Delta t)dx + \text{collision term integral}$

3. Nodal DG Spatial Discretization

Uses Lagrange polynomial basis functions: $f(x,v) = \sum_{j=1}^{N_x} \sum_{p_j=0}^k f(x_{j,p_j},v)\ell_{j,p_j}(x)$

4. IMEX-RK Time Discretization

s-stage IMEX-RK scheme: $f^{(i)} = \tilde{S}_{i,0}[f^n] + \Delta t \sum_{j=1}^{i-1} a_{ij}S_{i,j}[\frac{1}{\varepsilon}G_P(f^{(j)})] + \Delta t \sum_{j=1}^i \tilde{a}_{ij}\tilde{S}_{i,j}[\frac{1}{\varepsilon}Q_P(f^{(j)})]$

Technical Innovations

1. New Moment Update Strategy

In traditional Eulerian methods, moment updates eliminate stiff terms through integration: $U^{n+1} = U^n - \langle v \cdot \nabla_x f^n, \phi(v) \rangle$

However, in the SL framework, due to the presence of operator $S_{v,\Delta t}$, stiff terms cannot be eliminated. This paper proposes moment updates based on the limiting scheme: $U^{(i)} = (1-\bar{\tilde{A}}_{i-1}\hat{\tilde{A}}^{-1}_{(i-2)}e_{(i)})\langle S_{i,0}[f^n], \phi(v)\rangle + \bar{\tilde{A}}_{i-1}\hat{\tilde{A}}^{-1}_{(i-2)}\langle \tilde{S}^i[\hat{F}_{i-1}], \phi(v)\rangle$

2. Adaptation of Shu-Osher Formulation

Rewrites the original IMEX-RK scheme in Shu-Osher form, enabling clear analysis of limiting behavior and design of corresponding moment updates.

3. Asymptotic Accuracy Conditions

Establishes conditions for moment updates to achieve high-order accuracy:

• First-order: $c_s = 1$
• Second-order: $D_s = 1/2, B_s = 0$
• Third-order: $G_s = H_s = 1/6, B_s^* = B_s^{**} = B_s^{***} = 0$

Experimental Setup

Numerical Scheme

• Spatial discretization: Nodal DG method with polynomial orders k=2,3
• Time discretization: Three IMEX-RK schemes
  • FBEuler: First-order forward-backward Euler
  • DP2A242: Second-order 4-stage method
  • ARS443: Third-order 5-stage GSA method
• Collision operator: Maxwell molecule model, computed via fast spectral method

Test Problems

1. Accuracy test: Smooth Maxwellian initial values, verifying convergence rates
2. AP property test: Non-equilibrium initial values, verifying relaxation to equilibrium
3. Sod shock problem: Riemann problem, testing shock capturing capability
4. Mixed-scale test: Spatially varying ε(x), testing multiscale performance

Evaluation Metrics

• Relative L1 and L2 errors
• Convergence rates
• AP error $\|f - M_f\|_{\ell^1}$
• Comparison with reference solutions

Experimental Results

Main Results

1. Accuracy Verification

• For k=2, all schemes achieve second-order accuracy across different ε values
• For k=3, FBEuler and DP2A242 achieve expected accuracy, while ARS443 reduces to second-order at ε=10^{-6} (consistent with theoretical predictions)

2. AP Property Verification

• Type A methods (DP2A242) exhibit strong AP properties: at ε=10^{-4}, 10^{-6}, AP error rapidly decreases to ε level after one time step
• Type CK methods (ARS443) exhibit weak AP properties: require well-prepared initial values

3. Shock Problem

• At CFL=0.5, all methods accurately capture shock structure
• At CFL=2, FBEuler and DP2A242 remain stable, while ARS443 becomes unstable at small ε due to positivity constraints

4. Mixed-Scale Test

Higher-order methods (ARS443) maintain good accuracy under large CFL numbers, demonstrating the robustness of the method.

Theoretical Verification

Numerical results well validate the theoretical analysis:

• Convergence rates match theoretical predictions
• AP properties align with Type A/CK classification
• Positivity preservation conditions are verified

Related Work

Main Research Directions

1. AP schemes: Asymptotic-preserving methods pioneered by Jin et al., ensuring accuracy across all scales
2. Semi-Lagrangian methods: Transport solution methods avoiding CFL restrictions
3. DG methods: Spatial discretization methods with conservation and hp-adaptivity
4. IMEX methods: Time integration methods for handling stiff problems

Innovation of This Work

Compared to existing work, this paper is the first to achieve high-order AP schemes for the Boltzmann equation within the semi-Lagrangian DG framework, solving the key technical challenge of moment updates.

Conclusions and Discussion

Main Conclusions

1. Successfully develops high-order AP-SLDG schemes for the Boltzmann equation
2. Theoretical analysis reveals additional accuracy constraints in the SL framework
3. Numerical experiments verify the effectiveness and robustness of the method

Limitations

1. Positivity constraints: Some high-order schemes (e.g., ARS443) require ε-dependent time step restrictions
2. Stability analysis: Currently, stability is established only for linearized models; nonlinear cases require further investigation
3. Spatial dimension: Current analysis primarily addresses one-dimensional spatial problems

Future Directions

1. Extend stability analysis to nonlinear Boltzmann equations
2. Design IMEX-RK tableaus simultaneously satisfying high-order, asymptotic accuracy, and positivity preservation
3. Develop adaptive strategies for multiscale characteristics
4. Extend to high-dimensional spatial problems

In-Depth Evaluation

Strengths

1. Theoretical contribution: Establishes complete AP scheme theory within the SL framework, including accuracy conditions, stability, and positivity analysis
2. Technical innovation: Clever application of Shu-Osher formulation resolves the key challenge of moment updates
3. Comprehensive experiments: Multiple test problems thoroughly verify all properties of the method
4. Clear presentation: Well-structured paper with rigorous mathematical derivations

Weaknesses

1. Positivity limitations: Positivity preservation conditions for high-order schemes may restrict practical applications
2. Nonlinear stability: Stability analysis is limited to linearized cases
3. Computational complexity: The paper lacks detailed analysis of computational complexity and efficiency comparisons

Impact

1. Academic value: Provides new theoretical framework for numerical methods of kinetic equations
2. Practical value: Potential applications in rarefied gas dynamics, plasma physics, and related fields
3. Reproducibility: Detailed algorithm description and complete theoretical analysis ensure good reproducibility

Applicable Scenarios

1. Rarefied gas flow simulation
2. Plasma transport problems
3. Kinetic simulations requiring multiscale phenomena treatment
4. Scientific computing applications with high accuracy and efficiency requirements

References

The paper cites 42 relevant references covering important works in numerical methods for Boltzmann equations, AP schemes, semi-Lagrangian methods, DG methods, and related fields, demonstrating thorough and authoritative literature review.