Numerical analysis of a bisection-exclusion method to find zeros of univariate analytic functions (Q2576275): Difference between revisions

The paper under review is a detailed and up to date study of the bisection-exclusion method to locate and approximate the zeros of an analytic function \(f\) in a bounded domain (a square \(S_0\), in this case). The principle of the method is to remove from this domain those subsets, that is, squares obtained by successive bisections, which do not contain any zero of \(f\) and to keep small squares containing zeros. To do so, an exclusion test as well as a strategy to remove subsets of the initial domain, are necessary. The exclusion test used here is based on an exclusion function \(m:\mathbb{C}\rightarrow \mathbb{R_+}\), which is an implicit function obtained by the Taylor formula, i.e., \(m\) is defined implicitly by \[ M(x, m(x))=0, \] where \[ M(x,t)=\left| f(x)\right| -\sum_{k \geq 1}\frac{\left| f^{(k)}(x)\right| }{k!}t^k,\,\,x \in \mathbb{C},\,y \in \mathbb{R_+}. \] The theoretical complexity of the bisection-exclusion algorithm, the behavior of the exclusion function as well as the geometry of zeros, are detailedly discussed and then concrete examples and numerical experiments to illustrate the effectiveness of the method are also presented. To make the reading easier, the proofs are put separately, at the end of the paper.