Abstract:In this paper, we propose a new hybridized discontinuous Galerkin method for the Poisson equation with homogeneous Dirichlet boundary condition. Our method has the advantage that the stability is better than the previous hybridized method. We derive $L^2$ and $H^1$ error estimates of optimal order. Some numerical results are presented to verify our analysis.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: propose a new Hybridized Discontinuous Galerkin Method (HDG) to improve the stability and computational efficiency of existing methods. Specifically, the author proposes an improved HDG method for the Poisson equation with homogeneous Dirichlet boundary conditions and ensures the stability of this method under any penalty parameter by introducing a lifting operator.
### Main problems and solutions
1. **Limitations of existing methods**:
- Although traditional Discontinuous Galerkin Finite Element Methods (DGFEMs) are robust in dealing with geometric variations, the matrices they generate are large in size and bandwidth, resulting in high computational costs.
- Although existing hybrid DGFEMs can reduce the system size, in some cases, their stability depends on the choice of penalty parameters within a specific range, which limits their application scope.
2. **Objectives of the new method**:
- Propose a new hybrid DGFEM that remains stable under any penalty parameter.
- Improve the stability and accuracy of the method by introducing a lifting operator to define the penalty term.
3. **Specific implementation**:
- Introduce the lifting operator \( L_e \) and the penalty term \( j_h(u, v) \) so that the new method can remain stable under any penalty parameter.
- Perform a weak - form transformation of the Poisson equation and derive a new discrete system.
- Conduct error - estimate analysis to prove the convergence and optimal order of the new method.
### Summary of mathematical formulas
- Strong form of the Poisson equation:
\[
-\Delta u = f \quad \text{in } \Omega, \quad u = 0 \quad \text{on } \partial \Omega
\]
- Discrete system after weak - form transformation:
\[
B_L^h(u_h, v_h) := a_h(u_h, v_h) + b_h(u_h, v_h) + b_h(v_h, u_h) + j_h(u_h, v_h) = (f, v_h) \quad \forall v_h=(v_h, \hat{v}_h) \in V_h
\]
- Definition of the penalty term:
\[
j_h(u, v)=\sum_{K \in T_h}(L_{\partial K}(u - \hat{u}), L_{\partial K}(v - \hat{v}))_K+\sum_{K \in T_h}\sum_{e \subset \partial K}\int_e \eta_e h_e^{-1}(u - \hat{u})(v - \hat{v}) \, ds
\]
- Error estimate:
\[
|||u - u_h||| \leq C h^k |u|_{k + 1, \Omega}, \quad ||u - u_h||_{0, \Omega} \leq C h^{k+1} |u|_{k + 1, \Omega}
\]
### Conclusion
By introducing the lifting operator and redefining the penalty term, this paper successfully proposes a new hybrid discontinuous Galerkin method. This method not only remains stable under any penalty parameter but also has its optimal convergence verified in numerical experiments. Future research can be extended to other types of boundary - value problems and time - dependent problems.