Abstract:In this paper we are concerned with the stabilization of MUSCL-type finite volume schemes in arbitrary space dimensions. We consider a number of limited reconstruction techniques which are defined in terms inequality-constrained linear or quadratic programming problems on individual grid elements. No restrictions to the conformity of the grid or the shape of its elements are made. In the special case of Cartesian meshes a novel QP reconstruction is shown to coincide with the widely used Minmod reconstruction. The accuracy and overall efficiency of the stabilized second-order finite volume schemes is supported by numerical experiments.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to stabilize the application of the MUSCL (Monotonic Upstream - Centered Scheme for Conservation Laws) - type finite - volume method in arbitrary spatial dimensions. Specifically, the author focuses on how to improve the accuracy and stability of these numerical methods through restrictive reconstruction techniques.
### Main problems
1. **Reduce numerical dissipation**: Although the first - order finite - volume method is conservative and robust, due to its inherent large dissipation characteristics, the discontinuous parts of the solution will be over - smoothed.
2. **Avoid non - physical oscillations**: Although the second - order MUSCL method can provide better resolution, it is prone to generate spurious oscillations and non - physical values, which may lead to the immediate collapse of the numerical simulation.
3. **Applicable to arbitrary grids**: Many existing stabilization techniques are only applicable to specific types of structured grids, and are not applicable to general unstructured grids or multi - dimensional situations.
### Solutions
The author proposes a reconstruction technique based on linear or quadratic programming problems with local inequality constraints to ensure stable high - order accuracy on grids of any shape and size. Specifically:
- **QP (Quadratic Programming) reconstruction**: By minimizing a quadratic objective function and imposing constraints of local acceptability, the optimal linear approximation of each grid cell is calculated.
- **LP (Linear Programming) reconstruction**: Directly reconstruct a locally acceptable solution through a linear programming problem.
- **Minmod restriction theorem**: On the Cartesian grid, it is proved that the proposed QP reconstruction is consistent with the classical Minmod restriction theorem, thus verifying the reliability of the method.
### Key contributions
- A new QP reconstruction method is proposed, which can be extended to multi - dimensional Minmod reconstruction, and has no restrictions on the dimension, consistency or cell shape of the grid.
- Numerical experiments show that the proposed minimization problem is low - cost to solve and can provide more accurate and efficient approximate results.
### Mathematical formulas
The key formulas involved in the paper include:
- **QP reconstruction optimization problem**:
\[
\min_{w \in W(E;u)} J_E(w;u)
\]
where,
\[
J_E(v;u) = \sum_{E' \in N(E)} \frac{\omega_{E,E'}}{2} \| u_{E'} - v(x_{E'}) \|^2
\]
and
\[
W(E;u) = \left\{ w \in P_1(\mathbb{R}^d) \mid w(x_E) = u_E, \min\{u_E, u_{E'}\} \leq w(x_{E'}) \leq \max\{u_E, u_{E'}\} \right\}
\]
- **Gradient optimization problem**:
\[
\min \frac{1}{2} \nabla w \cdot (H \nabla w) - g \cdot \nabla w
\]
Subject to the constraint:
\[
0 \leq \text{sign}(m_{E,E'}) d_{E,E'} \cdot \nabla w \leq |m_{E,E'}|
\]
where,
\[
H = \sum_{E' \in N(E)} \omega_{E,E'} d_{E,E'} \otimes d_{E,E'}
\]
and
\[
g = \sum_{E' \in N(E)} \omega_{E,E'} m_{E,E'} d_{E,E'}
\]
Through these methods, the author has successfully solved the stability and accuracy problems of the MUSCL - type finite - volume method on complex grids.