On the convergence of Newton-type methods under mild differentiability conditions (Q1047178): Difference between revisions

The authors introduce the idea of recurrent functions to provide a new semilocal convergence analysis for a Newton-type method (NTM) \[ x_{n + 1} = x_n - A(x_n )^{ - 1}P(x_n ),\quad(n \geq 0),\;(x_0 \in D), \] \[ P(x) = F(x) + G(x), \quad (x \in D) \] for the following equation \[ F(x) + G(x) = 0. \] Here \(F\), \(G\) are Fréchet-differentiable and continuous operators respectively, \(F:D \to Y,\,\,G:D \to Y,\) where \(D \subset R\) is the convex subset and \(X\) and \(Y\)are Banach spaces. The NTM has been used by several authors to generate a sequence \(\{x_n \}\) approximating a locally unique solution \(x^\ast \) of the considered equation. \(A(x)\) belongs to the space of bounded linear operators from \(X\) to \(Y\) and is an approximation to the Fréchet derivative \({F}'(x)\) of the operator \(F(x)\). At each step one operator evaluation \(P(x_n)\) is required, and one inverse \(A(x_n )^{ - 1}\). Using the idea of recurrent functions a combination of Lipschits and center-Lipschitz conditions, instead of only Lipschitz conditions the authors get weaker sufficient convergence conditions and larger convergence domain than in earlier studies of many scientists. Applications and numerical examples are given too.