Numerical Solution of a Class of Deficient Polynomial Systems

From MaRDI portal





We consider polynomial systems represented by \[ (1)\quad p_ 1(x_ 1,...,x_ n)=0,...,p_ n(x_ 1,...,x_ n)=0, \] where \(p_ i\) is a polynomial of degree \(d_ i\) in the n variables, for \(1\leq i\leq n\). Bezout's theorem states loosely that, except on a set of measure zero, such systems have \(d_ 1d_ 2...d_ n\) solutions. In fact, homotopy continuation methods can find all solutions to (1) by following \(d_ 1d_ 2...d_ n\) paths. Nonetheless, many cases of (1) arising in applications are deficient in the sense that they have less than \(d_ 1d_ 2...d_ n\) solutions. However, general homotopy continuation methods still need to follow \(d_ 1d_ 2...d_ n\) paths, and are thus inefficient. In this paper, homotopies are presented for which continuation methods do not need to follow extra paths. The algebraic eigenvalue problem can be formulated as a member of the class of deficient systems treated. The paper also contains an example consisting of load-flow equations for an electrical network. The paper is mathematically sound, and results are rigorously proven. \textit{A. Morgan} and \textit{A. Sommese} [Preprint, GMR-5437, General Motors Research Labs. (1986)] present results with the same goal, but for a different class of deficient systems.




Cited in
(32)








This page was built for publication: Numerical Solution of a Class of Deficient Polynomial Systems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3028213)