Martin W. Licht
Abstract:Mixed boundary conditions are introduced to finite element exterior calculus. We construct smoothed projections from Sobolev de Rham complexes onto finite element de Rham complexes which commute with the exterior derivative, preserve homogeneous boundary conditions along a fixed boundary part, and satisfy uniform bounds for shape-regular families of triangulations and bounded polynomial degree. The existence of such projections implies stability and quasi-optimal convergence of mixed finite element methods for the Hodge Laplace equation with mixed boundary conditions. In addition, we prove the density of smooth differential forms in Sobolev spaces of differential forms over weakly Lipschitz domains with partial boundary conditions.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to provide a numerical analysis framework for the Hodge - Laplace equation with mixed boundary conditions, especially to construct smoothed projections that satisfy specific properties. Specifically, the author aims to:
1. **Introduce mixed boundary conditions in finite - element exterior calculus**: By introducing mixed boundary conditions into finite - element exterior calculus, it becomes possible to handle the situation where essential boundary conditions are imposed on part of the boundary and natural boundary conditions are imposed on the remaining boundary.
2. **Construct smooth projection operators**: Construct smooth projection operators from the Sobolev de Rham complex to the finite - element de Rham complex. These projection operators need to satisfy the following conditions:
- Commute with the exterior derivative,
- Preserve homogeneous boundary conditions along the fixed boundary part,
- Satisfy consistent boundary conditions for shape - regular triangulations and bounded polynomial degrees.
3. **Prove the density of smooth differential forms**: On weakly Lipschitz domains, smooth differential forms in Sobolev spaces with partial boundary conditions are dense. This helps to ensure the stability and convergence of numerical methods.
4. **Achieve the stability and quasi - optimal convergence of the mixed finite - element method**: By constructing the above - mentioned smooth projection operators, prove that the mixed finite - element method has stability and quasi - optimal convergence for the Hodge - Laplace equation with mixed boundary conditions.
### Summary of Mathematical Formulas
- **Hodge - Laplace Equation**:
\[
\text{Solve } u \text{ such that }-\Delta u = f, \quad u|_{\Gamma_D} = 0, \quad \nabla u|_{\Gamma_N}\cdot\vec{n} = 0
\]
where $\Gamma_D$ and $\Gamma_N$ represent the boundary parts where Dirichlet and Neumann boundary conditions are imposed respectively.
- **Energy Functional**:
\[
J(v):=\frac{1}{2}\int_\Omega|\nabla v|^2\,dx - \int_\Omega fv\,dx
\]
- **Vector - Laplace Equation**:
\[
\text{Solve }\vec{u}\text{ such that }\nabla\times(\nabla\times\vec{u})-\nabla(\nabla\cdot\vec{u})=\vec{f}
\]
The boundary conditions are:
\[
\vec{u}|_{\Gamma_T}\times\vec{n} = 0, \quad (\nabla\cdot\vec{u})|_{\Gamma_T} = 0, \quad \vec{u}|_{\Gamma_N}\cdot\vec{n} = 0, \quad (\nabla\times\vec{u})|_{\Gamma_N}\times\vec{n} = 0
\]
- **Smooth Projection Operator**:
\[
\pi_k : L^2\Lambda^k(\Omega)\to P\Lambda^k(T, U)
\]
such that the following diagram commutes:
\[
\begin{CD}
\cdots @>d>> H\Lambda^k(\Omega, \Gamma_T) @>d>> H\Lambda^{k + 1}(\Omega, \Gamma_T) @>d>> \cdots \\
@V\pi_k VV @V\pi_{k + 1} VV \\
\cdots @>d>> P\Lambda^k(T, U) @>d>> P\Lambda^{k + 1}(T, U)
\end{CD}
\]