Existence theorems for set-valued operators in Banach spaces (Q2460924): Difference between revisions

For a nonempty closed convex subset \(C\) of a real Banach space \(E\), let \(N_C(x)\) be the normal cone of \(C\) at \(x\in C\), that is, \(N_C(x)=\{ x^*\in E^*: \langle x-y, x^*\rangle \geq 0\), \(\forall y\in C\}\). The authors study the existence of points \(x\in X\) satisfying \(0\in Tx\), where \(T:X\rightarrow 2^{E^*}\) is a set-valued operator defined on a closed convex subset \(X\) of \(E\). A such \(x\) is called a zero point of \(T\). Extending a result obtained in [\textit{S.-Y. Matsushita} and \textit{W. Takahashi}, J.~Math.\ Anal.\ Appl.\ 323, No. 2, 1354--1364 (2006; Zbl 1110.47053)] from a single-valued operator to a set-valued operator, the authors prove that a monotone operator \(T\) upper semicontinuous on line segments in \(X\), with nonempty weak* compact convex values, has a zero point whenever there exists a nonempty weakly compact convex subset \(C\) of \(X\) such that \(-Tx\subset (N_C(x)\setminus \{0\})^c\), for all \(x\in C\). Applying this result, one obtains a fixed point theorem for set-valued upper semicontinuous operators on Hilbert spaces which generalizes the Kakutani fixed point theorem.