Log canonical thresholds of binomial ideals (Q836903)

Let \(\mathfrak{a}\subseteq (x_1,\ldots, x_n)\) be an ideal in the polynomial ring \(k[x_1,\ldots, x_n]\) over a field \(k\) of characteristic 0. The log canonical threshold \(lct_0(\mathfrak{a})\) of \(\mathfrak{a}\) at the origin is an invariant defined via the log resolution of \(\mathfrak{a}\). It is a difficult question to compute the log canonical threshold directly from the definition. In the case of monomial ideals, or principal ideals generated by non-degenerate polynomials, \textit{J. A. Howald} [Trans. Am. Math. Soc. 353, No. 7, 2665--2671 (2001; Zbl 0979.13026) and Multiplier ideals of sufficiently general polynomials, \url{arXiv:math.AG/0303203} (2003; preprint)] computed the log canonical thresholds using linear programming. The main result of the paper under review is to show that, for a large class of binomial ideals, the log canonical thresholds are still computable by linear programming. This class of binomial ideals includes the complete intersection binomial ideals, and the defining ideals of monomial curves in \(\mathbb{A}^3_k\). The proof of this result uses two main techniques. The first technique is to use the summation formula of multiplier ideals to establish an inequality relating the log canonical threshold of the binomial ideal \(\mathfrak{a}\), and the multiplier ideal of the monomial ideals generated by the monomial terms of a set of generators of \(\mathfrak{a}\), for which the result of \textit{J. A. Howald} can be applied. The second technique is to use the result that \(lct_0(\mathfrak{a})\) can be computed as the limit of the \(F\)-pure threshold \(fpt(\mathfrak{a}_p)\) when \(p\) goes to infinity, where \(fpt(\mathfrak{a}_p)\) is the \(F\)-pure threshold of the reduction to characteristic \(p\) of the ideal \(\mathfrak{a}\).