Abstract:We present a meshless finite difference method for multivariate scalar conservation laws that generates positive schemes satisfying a local maximum principle on irregular nodes and relies on artificial viscosity for shock capturing. Coupling two different numerical differentiation formulas and adaptive selection of the sets of influence allows to meet a local CFL condition without any a priori time step restriction. Artificial viscosity term is chosen in an adaptive way by applying it only in the vicinity of the sharp features of the solution identified by an algorithm for fault detection on scattered data. Numerical tests demonstrate a robust performance of the method on irregular nodes and advantages of adaptive artificial viscosity. The accuracy of the obtained solutions is comparable to that for standard monotone methods available only on Cartesian grids.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to develop a mesh - free finite - difference method for scalar conservation laws. This method can generate positive schemes satisfying the local maximum principle on irregular nodes and capture shocks through adaptive artificial viscosity. Specifically, the authors aim to:
1. **Achieve numerical stability on irregular nodes**: Traditional finite - difference methods usually rely on regular grids, but when dealing with problems of complex geometries or dynamic changes, regular grids may be inconvenient. Mesh - free methods allow nodes to be freely distributed in space without creating grids or re - meshing, thus increasing flexibility and efficiency.
2. **Capture shock phenomena**: When solving nonlinear hyperbolic conservation laws, such as compressible fluid flow models, shocks (i.e., discontinuities in the solution) are often encountered. In order to accurately capture these shocks, an appropriate artificial viscosity term needs to be introduced. However, traditional methods may require global time - step limitations or other additional structures, which limit their application scope.
3. **Reduce numerical dissipation**: Although artificial viscosity is helpful for stability and shock capture, it also introduces additional numerical dissipation, affecting the accuracy of the solution. Therefore, the authors propose an adaptive artificial viscosity strategy based on a fault - detection algorithm, applying artificial viscosity only near the sharp features of the solution to reduce the overall numerical dissipation.
### Main contributions
- **Mesh - free finite - difference method**: A mesh - free finite - difference method applicable to multi - variable scalar conservation laws is proposed, which can generate positive schemes on irregular nodes and ensure the local maximum principle.
- **Adaptive artificial viscosity**: By adaptively selecting the artificial viscosity term and applying artificial viscosity only near the sharp features of the solution (such as shocks), unnecessary numerical dissipation is reduced and the accuracy of the solution is improved.
- **Local CFL condition**: By continuously expanding the influence set, each node is ensured to satisfy the local CFL condition, thus avoiding global time - step limitations and improving computational efficiency.
### Representation of mathematical formulas
The key formulas involved in the paper include:
- The quasi - linear form of the conservation law:
\[
u_t+F'(u)\nabla_x u = 0
\]
where \(F'(u)\) is the derivative of the flux function \(F(u)\), and \(\nabla_x\) is the spatial gradient operator.
- The recurrence formula after discretization:
\[
U(t+\Delta t,x_i)=U(t,x_i)-\Delta t\sum_{x_j\in X_i}w_{ij}U(t,x_j)
\]
where \(X_i\) is the influence set of node \(x_i\), and \(w_{ij}\) is the weight of the numerical differentiation formula.
- The modified scheme after adding the artificial viscosity term:
\[
U(t + \Delta t,x_i)=U(t,x_i)-\Delta t\sum_{x_j\in X_i}w_{ij}U(t,x_j)+\mu_i\Delta t\sum_{x_j\in X_i}v_{ij}U(t,x_j)
\]
where \(\mu_i\) is the artificial viscosity coefficient, and \(v_{ij}\) is the numerical differentiation weight of the Laplace operator.
Through these improvements, the authors show that this method is robust and efficient in dealing with scalar conservation law problems on irregular nodes.