A simplicial mapping method for locating the zeros of a function

From MaRDI portal





The authors present a method for determining approximate locations of all simple zeros of an n-dimensional function f in an n-cube. The n-cube is divided into \(N_ 1\cdot N_ 2\cdot...\cdot N_ n\) smaller n-cubes by intersecting with hyperplanes parallel to the coordinate planes. Each subcube is partitioned into n! simplices. On each of the \(N_ 1\cdot N_ 2\cdot....\cdot N_ n\cdot n!\) simplices the function f is approximated by a linear function in the usual way and at least 2 (up to \(n+1)\) determinants of \(n\times n\) matrices must be evaluated for deciding whether there is a zero of f near the simplex. If the computed zero does not possess the required accuracy, a refining method is used by building up a simplex of smaller size whose center is the computed zero. The whole method is thought to complement a locally convergent method (e.g. Newton's method). The paper does not contain any reference to one of the well-known simplicial methods for locating the zeros of a function, see e.g. \textit{E. Allgower} and \textit{K. Georg} [SIAM Rev. 22, 28-85 (1980; Zbl 0432.65027)]. These methods are usually based on constructive proofs of Brouwer's fixed point theorem. Since there are no requirements on f in this paper, the presented method is of heuristic nature. It is very doubtful that the search procedure is practicable except perhaps in very low dimensions because of the exploding computational costs.











This page was built for publication: A simplicial mapping method for locating the zeros of a function

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3340040)