Existence for minimization with nonconvex constraints (Q1080169)

We prove the existence of a solution for the constrained optimization problem \(\inf \{f(x);\quad x\in K\}\) with the admissible set \(K=\{x\in X\); \(g_ i(x)\leq 0\), \(i=1,...,m\}\), where f and each \(g_ i\) are differentiable real-valued functions on a real Banach space X with strictly convex dual \(X^*\). The basic hypothesis consists in a certain decomposition of the derivative f'(x) taking into consideration K. The obtained result extends an existence theorem of \textit{V. Barbu} and \textit{T. J. Seidman} [Existence for minimization in Banach spaces with some applications, J. Math. Anal. Appl. (to appear)] treating the case where K is a finite intersection of half spaces. We also give a geometric version which can be regarded as a generalization of the Palais-Smale existence theorem [see \textit{R. S. Palais} and \textit{S. Smale}, Bull. Am. Math. Soc. 70, 165-172 (1964; Zbl 0119.092) and \textit{R. S. Palais}, Lectures on the differential topology on infinite-dimensional manifolds, Mimeographed notes, Brandeis Univ., Waltham/Mass. (1964-1965)] to constrained minimization problems. Finally, the abstract result is applied to a distributed parameter optimal control problem.