Least‐squares virtual element method for the convection‐diffusion‐reaction problem

Gang Wang,Ying Wang,Yinnian He

DOI: https://doi.org/10.1002/nme.6636

IF: 3.021

2021-02-26

International Journal for Numerical Methods in Engineering

Abstract:<p>In this paper, we introduce a least‐squares virtual element method for the convection‐diffusionreaction problem in mixed form. We use the <b><i>H</i></b>(div) virtual element and continuous virtual element to approximate the flux and the primal variables, respectively. The method allows for the use of very general polygonal meshes. Optimal order a priori error estimates are established for the flux and the primal variables in suitable norms. The least‐squares method offers an efficient a posteriori error estimator without extra effort. Moreover, the hanging nodes are naturally treated in the virtual element method, which provides the high flexibility in mesh refinement because the local mesh post‐processing is never required. Both attractive features motivate us to develop the a posteriori error estimate of the method. Numerical experiments are shown to illustrate the accuracy of the theoretical analysis and demonstrate that the adaptive mesh refinement driven by the proposed estimator can efficiently capture the boundary and the interior layers.</p>

engineering, multidisciplinary,mathematics, interdisciplinary applications

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to develop a least - squares virtual element method (LS - VEM) for solving the convection - diffusion - reaction equation in mixed form. Specifically, the paper focuses on how to effectively use non - conventional polygonal meshes to approximate fluxes and primitive variables, and can accurately capture the characteristics of boundary layers and internal layers while ensuring computational efficiency. In addition, the paper is also committed to providing a reliable a posteriori error estimator to guide the application of adaptive mesh refinement strategies.

### Main research contents

1. **Method introduction**:
- Use \(H(\text{div})\) virtual elements and continuous virtual elements to approximate fluxes and primitive variables respectively.
- This method allows the use of non - conventional polygonal meshes, providing a high degree of flexibility.
- The least - squares method naturally provides an effective a posteriori error estimator without additional effort.

2. **Theoretical analysis**:
- Establish the optimal - order a priori error estimates of fluxes and primitive variables in appropriate norms.
- Prove the reliability and effectiveness of the a posteriori error estimator.

3. **Numerical experiments**:
- Verify the accuracy of the theoretical analysis through numerical experiments.
- Demonstrate the effectiveness of the adaptive mesh refinement strategy in capturing boundary layers and internal layers.

### Key technical points

- **Virtual element method (VEM)**:
- Allows the use of non - conventional polygonal meshes, handles hanging nodes naturally, and provides highly flexible mesh refinement.
- Does not need to know the basis functions explicitly, and only calculates through degrees of freedom.

- **Least - squares method**:
- Provides a symmetric positive - definite algebraic system, and can efficiently use existing solvers (such as the pre - conditioned conjugate gradient method).
- Naturally provides an a posteriori error estimator without additional effort.

### Mathematical model

Consider the convection - diffusion - reaction equation:
\[
-\kappa \Delta u+\beta \cdot \nabla u+\gamma u = f \quad \text{in } \Omega,
\]
\[
u = 0 \quad \text{on } \Gamma,
\]
where:
- \(\Omega \subset \mathbb{R}^2\) is a bounded convex polygonal region with Lipschitz boundary \(\Gamma\).
- \(u\) is the physical quantity of interest.
- \(\kappa>0\) is the constant diffusion coefficient.
- \(\beta \in L^\infty(\Omega)\) is the convection field.
- \(\gamma \in L^\infty(\Omega)\) is the reaction coefficient.
- \(f\) is the source function.

By introducing the variable \(\sigma := -\tilde{\kappa} \nabla u\), the above equation can be rewritten in the form of a first - order system:
\[
\sigma+\tilde{\kappa} \nabla u = 0 \quad \text{in } \Omega,
\]
\[
\tilde{\kappa} \nabla \cdot \sigma+\beta \cdot \nabla u+\gamma u = f \quad \text{in } \Omega,
\]
\[
u = 0 \quad \text{on } \Gamma.
\]

### Conclusion

The method proposed in the paper performs well in dealing with convection - diffusion - reaction problems, especially when using non - conventional polygonal meshes and adaptive mesh refinement strategies. Through theoretical analysis and numerical experiments, the effectiveness and accuracy of this method are verified.