Criteria for the metric generalized inverse and its selections in Banach spaces (Q2479881)

Suppose that \(X\) and \(Y\) are Banach spaces and \(T:X\to Y\) is a linear mapping with domain \(D(T)\). Using the concept of best approximation solution [cf.\ Gen.\ Inverses Appl.\ Proc.\ adv.\ Semin., Madison 1973 (1976; Zbl 0324.00009)], the authors call the set-valued mapping \(T^\partial : Y\to D(T)\) defined by \[ T^{\partial}(y)=\{x_0 \in D(T)\mid x_0{\text{ is a best approximation solution to }}T(x)=y\} \] for any \[ y\in D(T^{\partial})=\{y\in Y\mid T(x)=y\text{ has a best aproximation solution in }D(T)\} \] the set-valued metric generalized inverse of \(T\). An operator (in general nonlinear) \(T^{\sigma}: D(T^{\partial})\to D(T)\) such that \(T^{\sigma}(y)\in T^{\partial}(y)\) for all \(y\in D(T^{\partial})\) is said to be a selection for the set-valued metric generalized inverse. The main aim of this paper is to construct a kind of bounded homogeneous selection for the set-valued metric generalized inverse.