Abstract:Recently, we proposed a weak Galerkin finite element method for the Laplace eigenvalue problem. In this paper, we present two-grid and two-space skills to accelerate the weak Galerkin method. By choosing parameters properly, the two-grid and two-space weak Galerkin method not only doubles the convergence rate, but also maintains the asymptotic lower bounds property of the weak Galerkin method. Some numerical examples are provided to validate our theoretical analysis.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: how to accelerate the computational efficiency of the Weak Galerkin (WG) finite - element method in solving the Laplacian eigenvalue problem while maintaining its asymptotic lower - bound property.
Specifically, the author proposes two techniques to accelerate the Weak Galerkin method:
1. **Two - grid method**: By solving the nonlinear system on the coarse grid and the linear system on the fine grid, the computational cost is reduced.
2. **Two - space method**: Computations are carried out using polynomial spaces of different orders on the same grid to improve computational efficiency.
By choosing appropriate parameters, these two methods can not only double the convergence rate but also maintain the asymptotic lower - bound property of the Weak Galerkin method. Numerical experiments are also provided in the paper to verify the correctness of the theoretical analysis.
### Summary of Mathematical Formulas
- **Weak Galerkin finite - element space**:
\[
V_h=\{v = (v_0, v_b):v_0|_T\in P_k(T), v_b|_e\in P_{k - 1}(e),\forall T\in T_h, e\in E_h,\text{ and }v_b = 0\text{ on }\partial\Omega\}
\]
- **Definition of weak gradient**:
\[
(\nabla_w v, q)_T=-(v_0,\nabla\cdot q)_T+\langle v_b, q\cdot n\rangle_{\partial T},\quad\forall q\in[P_{k - 1}(T)]^d
\]
- **Bilinear form**:
\[
s(v, w)=\sum_{T\in T_h}h_T^{- 1+\varepsilon}\langle Q_b v_0 - v_b, Q_b w_0 - w_b\rangle_{\partial T}
\]
\[
a_s(v, w)=(\nabla_w v,\nabla_w w)+s(v, w)
\]
\[
b_w(v, w)=(v_0, w_0)
\]
- **Weak Galerkin algorithm**:
\[
\text{Find }(\lambda_h, u_h)\in\mathbb{R}\times V_h\text{ such that }\|u_h\|_b = 1\text{ and }
\]
\[
a_s(u_h, v_h)=\lambda_h b_w(u_h, v_h),\quad\forall v_h\in V_h
\]
Through these methods and formulas, the author has successfully improved the computational efficiency of the Weak Galerkin method in solving the Laplacian eigenvalue problem and maintained its important mathematical properties.