Comments on "Simulation of nonlinear circuits in the frequency domain" [with reply]

In their discussion of optimization in the above-named work (see ibid., vol.CAD-5, p.521-35, Oct. 1986) the authors claim that optimization is inefficient, that information is lost in calculating the cost function to be minimized, and that it is difficult to exploit the structure of the node equations when using optimization methods. The authors conclude that relaxation and Newton´s method are the preferred techniques. The commenters address these issues and comment on the relationship between the optimization approach and Newton´s method. In replying, the original authors state that while the commenters have started from a minimization problem and derived an iteration formula that is identical to Newton-Raphson, in doing so they have added enough constraints so that their method is no longer capable of finding arbitrary minima (and so can no longer be considered an optimization method), but only those minima associated with roots of the underlying problem. Thus, it is claimed that the commenters have simply rederived the classic Newton-Raphson algorithm in a circuitous manner.<>