A new error analysis for parabolic Dirichlet boundary control problems (2306.15911v1)
Abstract: In this paper, we consider the finite element approximation to a parabolic Dirichlet boundary control problem and establish new a priori error estimates. In the temporal semi-discretization we apply the DG(0) method for the state and the variational discretization for the control, and obtain the convergence rates $O(k{\frac{1}{4}})$ and $O(k{\frac{3}{4}-\varepsilon})$ $(\varepsilon>0)$ for the control for problems posed on polytopes with $y_0\in L2(\Omega)$, $y_d\in L2(I;L2(\Omega))$ and smooth domains with $y_0\in H{\frac{1}{2}}(\Omega)$, $y_d\in L2(I;H1(\Omega))\cap H{\frac{1}{2}}(I;L2(\Omega))$, respectively. In the fully discretization of the optimal control problem posed on polytopal domains, we apply the DG(0)-CG(1) method for the state and the variational discretization approach for the control, and derive the convergence order $O(k{\frac{1}{4}} +h{\frac{1}{2}})$, which improves the known results by removing the mesh size condition $k=O(h2)$ between the space mesh size $h$ and the time step $k$. As a byproduct, we obtain a priori error estimate $O(h+k{1\over 2})$ for the fully discretization of parabolic equations with inhomogeneous Dirichlet data posed on polytopes, which also improves the known error estimate by removing the above mesh size condition.