Optimal error estimates of the local discontinuous Galerkin method for surface diffusion of graphs on Cartesian meshes (Q427218): Difference between revisions

The authors consider the local discontinuous Galerkin (LDG) method for surface diffusion of graphs \[ \begin{cases} u_t+\nabla \left( Q \left( I-\frac{\nabla u \otimes \nabla u }{Q^2}\right) \nabla H \right)=0 &\text{for }(x.t) \in \Omega \times (0,T] ,\\ u(x,0)=u_0(x) &\text{in } \Omega,\end{cases} \] subject to periodic boundary conditions, where \(\Omega\) is a bounded rectangular domain in \(\mathcal{R}^d, d \leq 3\), \(Q:=\sqrt{1+|\nabla u|^2}\) is the area element and \(H:= \nabla \cdot (\nabla u/Q)\) is the mean curvature of the domain boundary. The LDG method for surface diffusion of graphs was developed in an earlier paper by \textit{Y. Xu} and \textit{C.-W. Shu} [J.\ Sci.\ Comput.\ 40, No. 1--3, 375--390 (2009; Zbl 1203.65190)], where also its energy stability was proved. In the present paper, a priori error estimates are analyzed and the optimal convergence order \(k+1\) in the \(L^2\) norm is obtained when using completely discontinuous polynomials of degree \(k\) as trial functions.