On the existence of generally convergent algorithms

From MaRDI portal
(Redirected from Publication:1077876)





Let \(F_ d\) be the space of polynomials of degree \(\leq d\) and let S be the Riemann sphere \({\mathbb{C}}\cup \{\infty \}\). Then Newton's method \(N: F_ d\times S\to S\) with \(N(f,z):=z-f(z)/f'(x)\) is rational over \({\mathbb{C}}\) in f and z. If f is sufficiently close to a zero \(\zeta\) of f, then the iterates \(z_ k=N^ k(f,z_{k-1})\) converge to \(\zeta\) as \(k\to \infty\). However it has been shown, e.g., in earlier papers of the second author that there is an open set U in \(F_ d\times S\) (if \(d>2)\) such that this convergence will not happen for (f,z)\(\in U\). In the present paper it is shown that if one adds the operation of complex conjugation to the rational operations, then there do exist convergent purely iterative algorithms for finding zeros of polynomials.











This page was built for publication: On the existence of generally convergent algorithms

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1077876)