The Order of Convergence for Landweber Scheme with $\Alpha,\beta$-Rule

Caifang Wang,Tie Zhou
DOI: https://doi.org/10.3934/ipi.2012.6.133
2012-01-01
Inverse Problems and Imaging
Abstract:The Landweber scheme is widely used in various image reconstruction problems. In previous works, alpha,beta-rule is suggested to stop the Landweber iteration so as to get proper iteration results. The order of convergence of discrepancy principal (DP rule), which is a special case of alpha, beta-rule, with constant relaxation coefficient lambda satisfying 0 < lambda sigma(2)(1) < 1, (parallel to A parallel to(V,W) = sigma(1) > 0) has been studied. A sufficient condition for convergence of Landweber scheme is that the value lambda(m)sigma(2)(1) should be lied in a closed interval, i.e. 0 < epsilon <= lambda(m)sigma(2)(1) <= 2 - epsilon, (0 < epsilon < 1). In this paper, we mainly investigate the order of convergence of the alpha, beta-rule with variable relaxation coefficient lambda(m) satisfying 0 < epsilon <= lambda(m)sigma(2)(1) <= 2 - epsilon. According to the order of convergence, we can conclude that alpha, beta-rule is the optimal rule for the Landweber scheme.
What problem does this paper attempt to address?