An analog scheme for fixed-point computation - Part II: Applications

In a companion paper [6] we presented theoretical analysis of an analog net work for fixed-point computation. This paper applies these results to sever al applications from numerical analysis and combinatorial optimization, in particular: 1) solving systems of linear equations; 2) nonlinear programmin g; 3) dynamic programing; and 4) network how computations, Schematic circui ts are proposed for representative cases and implementation issues are disc ussed. Exponential convergence is established for a fixed-point computation that determines the stationary probability vector for a Markov chain. A fi xed-point formulation of the single source shortest path problem (SPP) that will always converge to the exact shortest path is described. A proposed i mplementation, on a 2-mu complementary metal-oxide-semiconductor (CMOS) pro cess, for a fully connected eight-node network is described in detail. The accuracy and settling time issues associated with: the proposed design appr oach are presented.