Bounding the number of connected components of a real algebraic set (Q1174716): Difference between revisions

This article presents upper bounds on the number of connected components (NCC) of a real affine algebraic set --- a natural `` complexity measure'' of this set. The first approach to this problem is to estimate NCC by a recursive function of the dimension of the real affine space, of the number of equations defining the algebraic set and of the maximal degree of these equations. Here the authors add two new bounds to the known Milnor-Thom's bound. To get them the authors use modifications of known ideas: (1) to regard NCC as a number of real solutions of some system of real algebraic equations, and (2) to estimate NCC by Betti numbers of a complexification of the real algebraic set by means of Smith's theory. --- The second approach is to give an estimate on NCC in terms of Newton polyhedra of defining equations. Here the authors use Khovansky's formula, which expresses the Euler characteristic of an algebraic set as the sum of mixed volumes of Newton polyhedra. It should be noted that these bounds give new information only for sufficiently large dimensions and degrees.