Effective cones of cycles on blowups of projective space (Q346722)

In this paper, the authors study the cones of higher codimension (pseudo)effective cycles on point blowups of projective space. Let \(\Gamma\) be a set of \(r\) distinct points on \({\mathbb{P}}^n\) and \(X^n_{\Gamma}\) be the blowup of \({\mathbb{P}}^n\) along \(\Gamma\). For a smooth variety \(Y\), let \(\overline{\text{Eff}}^k(Y)\) be the pseudoeffective cone of codimension-\(k\) cycles on \(Y\) and \({\overline{\text{Eff}}}_k(Y)\) be the pseudoeffective cone of dimension-\(k\) cycles. By induction on \(k\) and \(n\), the authors prove Theorem 3.1: if the \(r\) points in \(\Gamma\) are in linearly general position, and \(r\leq {\text{max}}(n+2, n+n/k)\), then \({\overline{\text{Eff}}}_k(X^n_{\Gamma})\) is linearly generated: it is the cone spanned by the classes of \(k\)-dimensional linear spaces in the exceptional divisors and the strict transforms of \(k\)-dimensional linear subspaces of \({\mathbb{P}}^n\), possibly passing through the points of \(\Gamma\). Let \(X_r^n\) be the blowup of \({\mathbb{P}}^n\) along \(\Gamma\) when \(\Gamma\) is a set of \(r\) very general points. The above theorem has the following applications. (1) If \(X_r^n\) is a Mori dream space, then \({\overline{\text{Eff}}}_k(X^n_r)\) is finitely generated. (2) The cone \({\overline{\text{Eff}}}_1(X^n_r)\) is linearly generated if and only if \(r\leq 2^n\). (3) If the SHGH conjecture holds for blowups of \({\mathbb{P}}^2\) at 10 points, then \({\overline{\text{Eff}}}^2(X^n_r)\) is not finitely generated if \(r\geq n+6\) and \(n\geq 3\). Here SHGH conjecture is Segre-Harbourne-Gimigliano-Hirschowitz conjecture: If \(r\geq 10\), \(m_1\geq m_2\geq ...\geq m_r\) and \(d>m_1+m_2+m_3\), then \[ H^0(X_r^2, dH_1-\sum^r_{i=1}m_iE_{i, 1}) =\frac{(d+1)(d+2)}{2}-\sum_{i=1}^r \frac{m_i(m_i +1)}{2}, \] where \(H_k\) is the class of a \(k\)-dimensional linear space in \({\mathbb{P}}^n\) and \(E_{i, k}\) is the class of a \(k\)-dimensional linear space contained in the exceptional divisor \(E_i\). The conclusion is that for (very) general points the higher codimension cones behave better than the cones of divisors since for the blowup \(X^n_r\) of \({\mathbb{P}}^n\), \(n>4\) at \(r\) very general points, the cone of divisors is not finitely generated when \(r>n+3\).