TL;DR
This paper introduces a novel reconstruction technique to create high-order, conservative semi-Lagrangian schemes for kinetic equations. This method addresses the challenge of maintaining accuracy and conservation, especially near discontinuities, which is crucial for simulating phenomena like shock waves in fluid dynamics and plasma physics.
What Problem Does This Paper Address?
Kinetic equations, like the Boltzmann transport equation, describe the behavior of rarefied gases and are closely related to quasi-linear systems of conservation laws. When these systems develop discontinuities (shocks), accurately simulating them numerically is difficult. Traditional methods often struggle with either conservation (leading to errors) or oscillations near these discontinuities. Semi-Lagrangian methods offer an advantage by avoiding CFL-type time step restrictions inherent in Eulerian approaches. However, they require reconstructing numerical solutions at off-grid points, and standard polynomial reconstructions can introduce spurious oscillations or lose conservation, especially when using non-linear, non-oscillatory techniques like ENO or WENO. This paper aims to bridge this gap by developing a reconstruction method that is both conservative and non-oscillatory, suitable for high-order semi-Lagrangian schemes.
Methods
The core contribution is a conservative reconstruction technique that builds upon existing non-oscillatory piecewise polynomial reconstructions (like CWENO). The key idea is to compute a "sliding average" of a precomputed non-oscillatory reconstruction. This sliding average is designed to inherit the non-oscillatory properties of the base reconstruction while guaranteeing conservation of discrete moments.
The method is detailed for both 1D and 2D cases. In 1D, given cell averages $\bar{u}_i$, a piecewise polynomial reconstruction $R(x)$ is first constructed. Then, the conservative approximation $Q_{i+\theta}$ is obtained by taking the integral of $R(x)$ over a shifted interval, weighted by coefficients $\alpha_l(\theta)$ and $\beta_l(\theta)$ derived from the polynomial reconstruction and the shift parameter $\theta$. The paper shows that with an even degree $k$ for the base polynomial reconstruction, the proposed method achieves $k+2$-th order accuracy.
The technique is then extended to 2D, where a similar sliding average approach is applied to a 2D piecewise polynomial reconstruction $R_{i,j}(x,y)$. The resulting 2D conservative reconstruction $Q_{i+\theta, j+\eta}$ is shown to achieve $k+2$-th order accuracy for an even degree $k$.
Datasets
Not discussed in the paper. The paper focuses on numerical methods and their theoretical properties, validated through numerical experiments on well-posed problems.
Key Insights and Ideas
- Bridging the gap: The paper successfully combines the benefits of non-oscillatory reconstructions (like CWENO) with the conservation properties required for accurate shock capturing in semi-Lagrangian schemes.
- Sliding average as a conservator: The novel "sliding average" of a polynomial reconstruction is the key innovation. It cleverly leverages the properties of the underlying reconstruction while ensuring conservation.
- High-order accuracy: The method achieves $k+2$-th order accuracy when using a $k$-th degree polynomial reconstruction with an even $k$. This is a significant improvement over linear reconstructions.
- Generalizability: The technique is presented in a general framework and extended to two dimensions, suggesting its applicability to a wider range of problems.
- Flexibility: The method can readily incorporate various existing non-oscillatory reconstructions (CWENO, CWENOZ, etc.) as its base, offering flexibility to users.
Key Results
The paper demonstrates the effectiveness of the proposed conservative reconstruction through numerical experiments on:
- 1D Xin-Jin model: The method shows good accuracy and conservation properties, particularly in capturing shock positions, outperforming some existing methods like GWENO34 and GWENO46 in shock tests.
- 1D Broadwell model: Similar positive results are observed for accuracy and conservation.
- 2D Xin-Jin model: The 2D extension also exhibits accurate shock capturing and conservation.
The numerical results consistently show that the proposed reconstruction maintains conservation errors within machine precision, even in the presence of discontinuities, validating its effectiveness.
Limitations and Future Work
- Smoothness requirement: The theoretical analysis of high-order accuracy relies on the assumption that the underlying function $u$ is sufficiently smooth. The paper notes that order reductions can occur in intermediate regimes, especially for discontinuous solutions.
- Positivity: While the paper discusses how to achieve positivity (e.g., by using positive limiters in the base reconstruction), it's not an inherent property of the reconstruction itself but rather inherited from the base reconstruction. The paper shows that with appropriate choices (e.g., for CWENO), negative solutions can be avoided.
- Future Work: The paper explicitly states that applications to BGK and Vlasov-Poisson equations will be presented in a companion paper, indicating these as immediate future directions. The general framework also suggests potential applications to other kinetic and hyperbolic systems.
Summary updated