About geometrical convergence of general iterative methods applied to nonunique solvable convex problems. I (Q1347165): Difference between revisions

Let \(H\) be a Hilbert space, \(M\) its nonempty, convex and closed subset representing the solution set of any convex problem and \((x_k)\) a sequence of elements in \(H\) generated by any iterative method. At the outset the author establishes several general statements on the geometrical convergence of the sequence \((x_k)\). Further he investigates more concrete rules of iteration and starts with the iterative scheme \(x_{k+1} = T_k (x_k - s_k)\), where \(s_k\) denotes the shift element \((s_k \in H)\) and the operators \(T_k : H \to H\) are supposed to be Féjer monotone with respect to the solution set \(M\) of the related problem, \(x_0\) arbitrary. He represents a thorough discussion how to improve the corresponding error estimates. Finally, he considers some applications to the approximate solution of convex problems with a solution set \(M\) (e.g. the convex feasibility problem, an optimization problem leading to the projection subgradient method).