On the optimal control and relaxation of nonlinear infinite dimensional systems (Q914140)

The author considers distributed parameter control systems governed by a semilinear evolution equation. Existence of an optimal solution is shown for the problem with a convexity assumption and also for the relaxed problem. The linear problem is then investigated and two bang-bang type results are obtained. The first of these shows that an optimal relaxed trajectory can be arbitrarily approximated by using an extremal control of the original system and the second result shows that for any admissible trajectory there exists a bang-bang control that generates the same end points. Some examples from systems governed by partial differential equations are investigated.