Binomial generation of the radical of a lattice ideal (Q615825)

To every lattice \(L \subset \mathbb{Z}^m\) there corresponds to a binomial ideal \(I_L\), called the lattice ideal, and a rational polyhedral cone \(\sigma_L\) generated by a vector configuration \(A\) in \(\mathbb{Z}^n\). A fundamental problem is to find the minimal number of polynomial equations whose common zeroes is the algebraic set \(Z(I_L)\) defined by \(I_L\), which is called the arithmetical rank of \(I_L\). A related problem is to determine the binomial arithmetical rank of \(I_L\), the minimal number of binomials needed to generate \(I_L\) up to radical. The main result of the paper under review is a necessary and sufficient condition to decide whether or not a given set of binomials generates \(I_L\) up to radical. These new conditions explain why there are lattice ideals whose binomial arithematical rank is huge compared to their arithmetical rank. They also explain why there are lattices whose ideals are set theoretic complete intersection in only one positive characteristic. As applications, the authors give another characterization of complete intersection lattice ideals and generalize to lattice ideals a theorem stating that a toric ideal with a full vector configuration is either a set theoretic complete intersection or almost set theoretic complete intersection in characteristic zero while it is always a set theoretic complete intersection in positive characteristic.