Hilbert-Kunz density function and asymptotic Hilbert-Kunz multiplicity for projective toric varieties (Q1632473)

Let \(R\) be a Noetherian ring of characteristic \(p > 0\) of dimension \(d\) and \(I \subseteq R\) be an ideal of finite colength. Let \(M\) be a finitely generated \(R\)-module. The Hilbert-Kunz multiplicity of \(M\) with respect to \(I\) is defined as \[ e_{H \, K}(M,I) = \lim_{n\rightarrow \infty} \frac{\ell(M/I^{[q]}M)}{q^{d}}, \] where \(q = p^{n}\), \(I^{[q]}\) is the \(n\)-th Frobenius power of \(I\), and \(\ell(M/I^{[q]}M)\) denotes the length of the \(R\)-module \(M/I^{[q]}M\). The Hilbert-Kunz density function is defined as a compactly supported continuous function \(\mathrm{HKd}(M,I) : [0, \infty) \rightarrow [0, \infty)\) such that \[ e_{H \, K}(M,I) = \int_{0}^{\infty}\mathrm{HKd}(M,I)(x) dx. \] In the paper under review, the authors study Hilbert-Kunz density functions and the asymptotic Hilbert-Kunz multiplicity for toric pairs \((X,D)\), where \(X\) is a projective toric variety of dimension \(d-1 \geq 1\) over an algebraically closed field \(K\) of characteristic \(p>0\) and \(D\) is a very ample torus invariant Cartier divisor. It is well-known that if \((X,D)\) is a toric pair, then we have an associated lattice polytope \(P_{D} \subset \mathbb{R}^{d-1}\). The first result of the paper gives an interpretation of the Hilbert-Kunz density function for a toric pair \((X,D)\) in terms of \(\mathcal{P}_{D}\). Theorem A. Let us denote by \(C_{D}\) the convex rational polyhedral cone spanned by \(P_{D} \times 1\) in \(\mathbb{R}^{d}\). Let \[ \mathcal{P}_{D} = \{p \in C_{d} \, | \, p \not \in (u_{j}, 1) + C_{d} \text{ for every } u_{j} \in P_{D} \cap \mathbb{Z}^{d-1}\}. \] Then the Hilbert-Kunz density function \(\mathrm{HKd}(X,D)\) for a toric pair \((X,D)\) is given by the sectional volume of the closure of \(\overline{\mathcal{P}_{D}}\), i.e., \[ \mathrm{HKd}(X,D)(\lambda) = \mathrm{Vol}_{d-1}(\overline{\mathcal{P}_{D}} \cap \{z = \lambda\}) \] for some \(\lambda\geq 0\). In order to get the main result of the paper, we also need the following (technical) proposition. Proposition. Let \((X,D)\) be a toric pair, then for \(\lambda \geq 0\) we have \[ \mathrm{HKd}(X,kD)(\lambda +1) = \frac{e_{0}(X,D)k^{d-1}}{(d-1)!} \phi_{kD}(\lambda) + O(k^{d-2}), \] where \(\phi_{kD}(\lambda) \rightarrow [0,1]\) is the compactly supported continous function given by \[ \phi_{kD}(\lambda) = \mathrm{Vol}_{d-1}([W_{v} \times \{z=\lambda\}] \setminus \bigcup_{u \in \mathbb{Z}^{d-1}} [(u,1)+C_{kD}]), \] where \(e_{0}(\cdot)\) denotes the Hilbert-Samuel multiplicity, and for any vertex \(v \in \mathbb{Z}^{d-1}\) we denote by \(W_{v} \subset \mathbb{R}^{d-1}\) the \(d-1\) dimensional unit cell at the vertex \(v\). Combining the above two results we get the main theorem of the paper under review. Main Result. Let \(X\) be a projective toric variety with a very ample torus invariant Cartier divisor \(D\) with the associated homogeneous ring \((R,m)\). We have \[ \lim_{k \rightarrow \infty} \frac{e_{H\, K}(R,m^{k}) - e_{0}(R,m^{k})/d!}{k^{d-1}} = \frac{e_{0}(R,m)}{(d-1)!} \int_{0}^{\infty} \phi_D{}(\lambda) d\lambda, \] where \(\phi_{D} : [0,\infty) \rightarrow [0,\infty)\) is defined as above (with \(k=1\)) and is solely determined by the shape of the polytope \(P_{D}\) associated with the toric pair \((X,D)\).