Nagata's conjecture for a square or nearly-square number of points (Q996928)

It is conjectured by Harbourne and Hirschowitz, after Segre, that the dimension of the linear system of plane curves of degree \(d\) having \(n>9\) general points of multiplicity \(m\) is \(\max\{-1,\nu\}\), where \(\nu\) is the virtual dimension given by \(2\nu=d(d+3)-nm(m+1)\). This would imply, in particular, a conjecture of Nagata that no such curves exist if \(d<m\sqrt{n}\) and \(n>9\). Nagata proved his conjecture for the case where \(n=k^2\) is a perfect square. The main result of this paper is that the conjecture of Segre, Harbourne and Hirschowitz is also true if \(n=k^2\) is a square. The proof works by a degeneration argument. First, one allows \(\mathbb{ P}^2\) to degenerate by blowing up a line in the central fibre of the trivial family \(\mathbb{ P}^2\times \Delta\to \Delta\) over a disc. The result is a degeneration to \(\mathbb{ P}\cup\mathbb{ F}\), where \(\mathbb{ P}\cong \mathbb{P}^2\) and \(\mathbb{F}\) is a copy of the Hirzebruch surface \(\mathbb{ F}_1\), with the negative section \(R\) of \(\mathbb{ F}\) identified with a line in \(\mathbb{ P}\). The \(k^2\) points are then allowed to degenerate to \((k-1)^2\) points on \(\mathbb{ P}\) and \(2k-1\) points on \(\mathbb{ F}\), and \(\mathcal{O}(d)\) degenerates to \(\mathcal{O}_{\mathbb{P}}(d-m)\) and \({O}_{\mathbb{F}}(mR+dF)\), where \(F\) is the fibre of the ruled surface \(\mathbb{F}\). Imposing the multiple points now gives linear systems on \(\mathbb{P}\) and \(\mathbb{F}\), and the proof consists of an analysis of the possibilities for these systems and their restrictions to~\(R\). The same result is obtained, also by degeneration techniques, in preprints by \textit{L. Evain} [Computing limit linear series with infinitesimal methods, \url{arXiv:0407143}; cf. Ann. Inst. Fourier 57, No. 6, 1947--1974 (2007; Zbl 1134.14020)] and \textit{J. Roé}, ``Limit linear systems and applications'', \url{arXiv:0602213}] . The version here is more direct, but all three extend in different ways. This version, for example, also works of \(n=k^2\pm h\) with \(|h|\) small: in this paper the authors consider \(-9\leq h\leq 8\).