A New Finite Element Approximation of a State-constrained Optimal Control Problem.pdfVIP

Journal of Computational Mathematics, Vol.27, No.1, 2009, 97–114. A NEW FINITE ELEMENT APPROXIMATION OF A STATE-CONSTRAINED OPTIMAL CONTROL PROBLEM* Wenbin Liu KBS Institute of Mathematics and Statistics, The University of Kent, Canterbury, CT2 7NF, England Email: W.B.Liu@ukc.ac.uk Wei Gong and Ningning Yan LSEC, Institute of Systems Science, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China Email: gongwei@amss.ac.cn, ynn@amss.ac.cn Abstract In this paper, we study numerical methods for an optimal control problem with point- wise state constraints. The traditional approaches often need to deal with the delta- singularity in the dual equation, which causes many di?culties in its theoretical analysis and numerical approximation. In our new approach we reformulate the state-constrained optimal control as a constrained minimization problems only involving the state, whose optimality condition is characterized by a fourth order elliptic variational inequality. Then direct numerical algorithms nonconforming ?nite element approximation are proposed for the inequality, and error estimates of the ?nite element approximation are derived. Numerical experiments illustrate the e?ectiveness of the new approach. Mathematics subject classi?cation: 49J20, 65N30. Key words: Optimal control problem, State-constraints, Fourth order variational inequal- ities, Nonconforming ?nite element method. 1. Introduction In this paper, we consider the following state-constrained optimal control problem: 1 2 α 2 min y ? y dx + u dx 1.1 0 y ≤? 2 ? 2 ? subject to ??y u in ? y 0 on ??. where α 0 is a parameter, ? is a bounded domain in R2 with the Lipschitz continuous 2 boundary ??, y0 ∈ L ? is the desired state, and ? is a given function. We further assume that ?|?? 0, and more details will be speci?ed later. Such a state-con