Analysis and approximation to parabolic optimal control problems with measure-valued controls in time

¹ The State Key Laboratory of Scientific and Engineering Computing, Institute of Computational Mathematics & National Center for Mathematics and Interdisciplinary Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, 100190 Beijing, PR China
² The Hong Kong Polytechnic University Shenzhen Research Institute, Shenzhen 518057, PR China

^* Corresponding author: dongdong.liang@polyu.edu.hk

Received: 2 April 2024
Accepted: 28 November 2024

Abstract

In this paper, we investigate an optimal control problem governed by parabolic equations with measure-valued controls over time. We establish the well-posedness of the optimal control problem and derive the first-order optimality condition using Clarke’s subgradients, revealing a sparsity structure in time for the optimal control. Consequently, these optimal control problems represent a generalization of impulse control for evolution equations. To discretize the optimal control problem, we employ the space-time finite element method. Here, the state equation is approximated using piecewise linear and continuous finite elements in space, alongside a Petrov–Galerkin method utilizing piecewise constant trial functions and piecewise linear and continuous test functions in time. The control variable is discretized using the variational discretization concept. For error estimation, we initially derive a priori error estimates and stabilities for the finite element discretizations of the state and adjoint equations. Subsequently, we establish weak-* convergence for the control under the norm ℳ(Ī_c;L²(ω)), with a convergence order of O(h^1/2 + τ^1/4) for the state.