Numerical homotopies to compute generic points on positive dimensional algebraic sets (Q1594830)

The authors consider the space of solutions to a system of polynomial equations in \(n\) variables over the complex numbers. The problem is to determine the dimension of, and a ``generic'' point on, each component of this space by numerical methods. In a previous article [Numerical algebraic geometry. In: Renegar, James (ed.) et al., The mathematics of numerical analysis. 1995 AMS-SIAM summer seminar in applied mathematics, July 17--August 11, 1995, Park City, UT, USA. Providence, RI: American Mathematical Society. Lect. Appl. Math. 32, 749-763 (1996; Zbl 0856.65054)], the first author and \textit{C. Wampler} developed an algorithm for solving this problem based on slicing by ``generic'' (i.e., ``random'') linear spaces of different dimensions. Here, the authors embed the given system into a family of systems of polynomials in \(2n\) variables that depends on a large space of parameters. By choosing particular values of these parameters, they obtain \(n+1\) systems, \({\mathcal E}_0,\dots,{\mathcal E}_n\) and homotopies from \({\mathcal E}_i\) to \({\mathcal E}_{i-1}\), such that \({\mathcal E}_n\) has isolated nonsingular solutions and \({\mathcal E}_0\) is equivalent to the given system. Their algorithm uses polynomial continuation, as developed by \textit{A. P. Morgan} [Solving polynomial systems using continuation for engineering and scientific problems. Englewood Cliffs, NJ: Prentice-Hall (1987; Zbl 0733.65031)]. Examples are given to show that this new algorithm can be much more efficient than the earlier one employing slicing.