Numerical approximation of nonconvex optimal control problems defined by parabolic equations (Q791140): Difference between revisions

We consider an optimal control problem for distributed systems governed by parabolic equations. The state equations are nonlinear in the control variable; the constraints and the cost functional are generally nonconvex. Relaxed controls are used to prove existence and derive necessary conditions for optimality. To compute optimal controls, a descent method is applied to the resulting relaxed problem. A numerical method is also given for approximating a special class of relaxed controls, notably those obtaind by the descent method. Convergence proofs are given for both methods, and a numerical example is provided.