CONVERGENCE THEOREMS OF ITERATIVE ALGORITHMS FOR A FAMILY OF FINITE NONEXPANSIVE MAPPINGS

Abstract

Let $E$ be a Banach space, $C$ a nonempty closed convex subset of $E$, $f : C \to C$ a contraction, and $T_i : C \to C$ a nonexpansive mapping with nonempty $F := \bigcap_{i = 1}^N Fix(T_i)$, where $N \ge 1$ is an integer and $Fix(T_i)$ is the set of fixed points of $T_i$. Let $\{x_t^n\}$ be the sequence defined by $x_t^n = tf(x_t^n) + (1-t) T_{n+N} T_{n+N-1} \cdots T_{n+1} x_t^n$ ($0 \lt t \lt 1$). First, it is shown that as $t \to 0$, the sequence $\{x_t^n\}$ converges strongly to a solution in $F$ of certain variational inequality provided $E$ is reflexive and has a weakly sequentially continuous duality mapping. Then it is proved that the iterative algorithm $x_{n+1} = \lambda_{n+1} f(x_n) + (1-\lambda_{n+1}) T_{n+1} x_n$ ($n \ge 0$) converges strongly to a solution in $F$ of certain variational inequality in the same Banach space provided the sequence $\{\lambda_n\}$ satisfies certain conditions and the sequence $\{x_n\}$ is weakly asymptotically regular. Applications to the convex feasibility problem are included.