Counting solutions to binomial complete intersections (Q870342): Difference between revisions

The paper studies binomial ideals in the polynomial ring in \(n\) variables over a field \(k\) of characteristic \(0\), i.e. ideals generated by a set of binomials \(p_j(x) = x^\alpha_j-c_jx^\beta_j,\alpha\) and \(\beta\) being \(n\)-tuples of integers. The main concern of the paper is deciding whether such an ideal has finitely many zeros (and at least one) in the algebraic closure \(\bar k\), which means that the corresponding variety is a complete intersection. This is done by using the matrix \(B= \left(\begin{smallmatrix} \alpha_1-\beta_1&\\ \cdots\\ \alpha_n-\beta_n \end{smallmatrix}\right)\). In particular, when \(\det(B) = 0\) it is necessary to introduce and discuss the notion of generic complete intersection (g.c.i.): the \(p_j(x)\) form a g.c.i. when the corresponding variety is a complete intersection for a generic choice of the coefficients \(c_j\). The paper then gives conditions that force the binomial ideal to be a g.c.i. One section of the paper is devoted to complexity questions, in particular to the problem of counting the number of zeros in polynomial and not exponential time.