Convergence theorems for generalized projections and maximal monotone operators in Banach spaces (Q1811852)

Let \(E\) be a smooth strictly convex reflexive Banach space and \(E^{\ast}\) its dual space. Let \(V:E\times E\rightarrow\mathbb{R},\) \(V\left( x,y\right) =\left\| x\right\| ^{2}-2\langle J\left( x\right) ,y\rangle+\left\| y\right\| ^{2},\) where \(J\) is the normalized duality mapping and \(C\subset E\) is a closed convex set. Then for every \(x\in E\) there exists a unique \(y_{x}\in C\) such that \(V\left( x,y_{x}\right) =\min_{y\in C}V\left( x,y\right) \). The mapping \(\Pi_{C}:E\rightarrow C,\;\Pi_{C}\left( x\right) =y_{x}\) is called a generalized projection on \(C\). Within this framework, the authors establish the following results: (1) If \(\left\{ C_{n}\right\} _{n\geq1}\) is a sequence of nonempty closed convex subsets of \(C\) and if there exists the limit in the Mosco sense \(C_{0}=M-\lim C_{n}\neq\emptyset\), then \(C_{0}\) is closed and convex and, for every \(x\in C\), \(\Pi_{C_{n}}\left( x\right) \) is weakly convergent to \(\Pi_{C_{0} }\left( x\right) \). (2) If the norm of \(E^{\ast}\) is Fréchet differentiable then, for every \(x\in C\), \(\Pi_{C_{n}}\left( x\right) \) converges strongly to \(\Pi_{C_{0}}\left( x\right) \). (3) If the norm of \(E\) is Fréchet differentiable and \(\lim\Pi_{C_{n}}\left( x\right) =\Pi_{C_{0}}\left( x\right) \) for every \(x\in C,\) then \(C_{0} =M-\lim_{n}C_{n}\). Using the above results, the authors prove weak convergence and strong convergence theorems for the resolvents of a sequence of maximal monotone operators.