Accurate solution of polynomial equations using Macaulay resultant matrices (Q4821046)

This is a very interesting original paper investigating numerical analysis questions in the context of real algebraic geometry. In the vein of the pioneering book of \textit{H. J. Stetter} [Numerical polynomial algebra. SIAM (2004; Zbl 1058.65054)], the authors are interested in the numerical aspects of solving systems of polynomial equations. Because of the considerable difficulties of extending linear numerical analysis (in the spirit of Wilkinson) to nonlinear (polynomial) problems, the authors focus on the very specific problem of solving a system of two polynomial equations in two variables, using a particular resultant technique proposed in 1902 by Macaulay. Already in this seemingly simple case, the reader can get a flavor of the sophisticated mathematical arsenal needed to overcome many technical problems. The authors honestly state that they are still far from a comprehensive understanding of the problem. In particular, they could obtain only a very crude bound on the forward error, and their proposed variation around Macaulay's original algorithm is not backward stable. In my opinion, this paper and similar studies will eventually pave the way for a consolidated numerical polynomial algebra eventually leading to the development of efficient and reliable computer-aided software for real algebraic geometry, with many potential applications in engineering and applied mathematics.