A relation between one-point and multi-point Seshadri constants (Q1883011)

Let \(X\) be an \(n\)-dimensional projective variety over a field \(k\) with \(k=\overline k\). For any ample divisor \(L\) on \(X\) and \(r\) points \(p_1,\dots,p_r\in X\), we define the \(d\)-dimensional Seshadri constant \(\varepsilon _d(L,p_1,\dots,p_r)\) to be the number \(\root d\of {\inf_Z\left\{ \frac{L^d \cdot Z}{ \sum \text{mult}_{p_i}Z}\right\}}\), where \(Z\) runs over all positive \(d\)-dimensional cycles of \(X\). It is not difficult to prove that we always have: \(\varepsilon _d(L,p_1,\dots,p_r) \leq \root n\of {\frac{L^n}{r}}\). For \(p_1,\dots,p_r\) very general in \(X\) we just write \(\varepsilon_d (L,r)\), and we omit \(d\) if \(d=1\), the most studied case. The idea for the Seshadri constant comes from Nagata's conjecture, which can be expressed as: If \(r\geq 9\), then \(\varepsilon({\mathcal O}_{\mathbb{P}^2}(1),r) = {1\over \sqrt{r}}\), which amounts to say that for a very general set of \(r\) points in the plane the Seshadri constant is maximal. A generalization of the conjecture, due to Szemberg, says that for every \(n\)-dimensional \(X\), it exists \(r_0\) such that \(\forall r\geq r_0\): \(\varepsilon (L,r) = \root n\of {\frac{L^n}{r}}\). The main result in the paper relates the \((n-1)\)-dimensional Seshadri constant of a \(n\)-dimensional variety with the case of points in a projective \(n\)-space: Let \(n=\dim X\geq 2\), then for every smooth point \(p\in X\) we have \[ \varepsilon _{(n-1)}(L,r) \geq \varepsilon _{(n-1)}(L,p)\varepsilon_{(n-1)}({\mathcal O}_{\mathbb{P}^n}(1),r). \] This allows to get many corollaries by using known results for the projective space case; in particular, it is proved that Nagata's conjecture implies Szemberg's one for smooth surfaces \(X\) when \(L\) generates \(NS(X)\) and \(L^2\) is a square.