On the Convergence of Hillam's Iteration Scheme
Bernd-Jürgen Falkowski · Mathematics Magazine · 1996
In their excellent paper [1] on digital halftoning, Geist, et al., develop an interesting approach to neural network simulation. In this context they reaffirm a conjecture on the convergence of a certain numerical iteration scheme originally due to Hillam, cf. [2), since they found substantial numerical evidence supporting it. In this note we prove the conjecture under a more restrictive condition than the one given in [1]. Moreover, we present a numerical example providing evidence that the original conjecture does not hold.