Information-based complexity of linear operator equations (Q1194382)

The author investigates the complexity of linear operator equations generated by bounded operators on Hilbert spaces. To formulate the problem precisely the author specifies the following components of the setting: 1. what is the class of instances; 2. how the accuracy of an approximate solution is measured. The main goal is to evaluate the complexity for the ``traditional'' classes of equations and to point out the corresponding optimal and suboptimal methods. The author studies complexity bounds and optimality of Chebyshev methods and suboptimality of the conjugate gradient methods. In concluding remarks some generalizations of the obtained results to the nonlinear case are given.