Improved generalized differentiability conditions for Newton-like methods (Q983186): Difference between revisions

The authors are concerned with the problem of approximating a locally unique solution \(x^{*}\) of a nonlinear equation \(F(x)=0\) where \(F\) is a Fréchet differentiable operator defined on a subset \(\mathcal{D}\) of the Banach space \(\mathcal{X}\) with values in the Banach space \(\mathcal{Y}\). The Newton-like approximating sequence is given by \[ x_{n+1}=x_{n}-A(x_{n})^{-1}F(x_{n}) \quad (n>0). \] Here \( A(x)\in\mathcal{L(X,Y)}, (x\in\mathcal{D}),\quad \mathcal{L(X,Y)}\) being the space of linear bounded operators from \(\mathcal{X}\) to \(\mathcal{Y}, \) and \( A(x)\) an approximation to the Fréchet-derivative. A theorem proves the semilocal convergence of the sequence. The method and the results are then extended to the equation \(F(x)+G(x)=0\) where \(G:\mathcal{D}\rightarrow\mathcal{Y}\) and the Newton-like method generating the corresponding sequence approximating the exact solution reads as follows \[ x_{n+1}=x_{n}-A(x_{n})^{-1}(F(x_{n})+G(x_{n})) \quad (n\geq 0). \] The conditions under which the convergence and the error estimates take place are better than the known ones. Examples are presented.