On Spherical Convergence, Convexity, and Block Iterative Projection Algorithms in Hilbert Space

Author: Cohen N.   Ames T.W.F.  

Publisher: Academic Press

ISSN: 0022-247X

Source: Journal of Mathematical Analysis and Applications, Vol.226, Iss.2, 1998-10, pp. : 271-291

Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.

Previous Menu Next

Abstract

We call a sequence {xn} in Hilbert space “spherical” if there exists u such that lim‖xn - u‖ exists and is finite. If moreover u is a weak accumulation point of the sequence, we call the sequence “spherically convergent.”We demonstrate that for large classes of nonexpansive (possibly nonstationary) discrete-time processes in Hilbert space the iterates are spherically convergent. Basic identities and orthogonality relations pertinent to this type of convergence are exhibited. Sufficient conditions for weak and spherical convergence, in terms of the “fullness” of the set of fixed points of the process, are established and compared. Spherical convergence of the general block iterative projection scheme in Hilbert space is established.