Special homogeneous linear systems on Hirzebruch surfaces (Q5961900)

From MaRDI portal
scientific article; zbMATH DE number 5786264
Language Label Description Also known as
English
Special homogeneous linear systems on Hirzebruch surfaces
scientific article; zbMATH DE number 5786264

    Statements

    Special homogeneous linear systems on Hirzebruch surfaces (English)
    0 references
    0 references
    16 September 2010
    0 references
    When considering a linear system of plane curves with given multiplicities at a fixed set of generic points in \({\mathbb P}^2\), a well-known conjecture describes what the dimension of such linear system should be; namely, let \(P_1,\dots,P_r\in {\mathbb P}^2\) be generic points and \(m_1,\dots,m_r\in {\mathbb N}\), then the linear system \({\mathcal L}(d;m_1,\dots,m_r)\) of curves of degree \(d\) having multiplicity at least \(m_i\) at \(p_i\) has the expected dimension \(\max \{{d+2\choose 2}- \sum _{i=1}^r {m_i+1\choose 2}-1,\;-1 \}\), unless there is a fixed multiple divisor \(nC\), \(n\geq 2\), for the divisors in the system, and \(C\) is a (-1)-divisor, i.e. its strict transform \(\widetilde{C}\) on the blow-up of \(X\) at the \(P_i\)'s is an exceptional divisor. This conjecture (which has been proved in many cases but remains open in general) is generalized in this paper to the case of linear systems on a Hirzebruch surface \({\mathbb F}_n\). Let \({\mathbb F}_n={\mathbb P}({\mathcal O}_{{\mathbb P}^1}\oplus {\mathcal O}_{{\mathbb P}^1}(n))\), then Pic\({\mathbb F}_n={\mathbb Z}F_n\oplus {\mathbb Z}H_n\), where \(F_n\) is the class of a fiber and \(H_n\) the class of a section with \(F_n\cdot H_n=1\) and \(H_n^2=n\). We will write \({\mathcal L}_n(a,b)\) for the complete linear system \(aF_n+bH_n\); for any \(a,b\in {\mathbb Z}\) the \(\dim {\mathcal L}_n(a,b)\) is known. Then let \({\mathcal L}_n(a,b;m_1,\dots,m_r)\) be the sublinear system of \({\mathcal L}_n(a,b)\) given by the curves with multiplicity at least \(m_i\) at \(P_i\), where \(P_1,\dots,P_r\) is a generic set of points in \({\mathbb F}_n\). Moreover, we say that \(E={\mathcal L}_n(a,b;m_1,\dots,m_r)\) is a (-1)-system, and \(E\) is a (-1)-divisor, if \(a,b>0\), \(\dim E =0\), which is also the expected dimension, \(E^2=-1\) and the divisor \(E\) is irreducible. Then the conjecture stated here for the linear systems \({\mathcal L}_n(a,b;m_1,\dots,m_r)\) on \({\mathbb F}_n\) is that they are special (i.e. they are effective and their dimension is bigger than expected) if and only if their curves contain a fixed multiple (-1)-divisor. The main result in the paper shows that the conjecture is true for \(m_1=\dots=m_r\leq 10\). The first idea for the proof id to consider linear systems \({\mathcal L}(d;m_1,\dots,m_r,\overline{k_1,\dots,k_s})\) in \( {\mathbb P}^2\), where the multiplicity are consider with respect to \(r\) generic points and \(s\) points on a generic line \(L\subset {\mathbb P}^2\). It is shown that \(\dim {\mathcal L}_n(a,b;m_1,\dots,m_r)=\dim {\mathcal L}(a+(n+1)b; a+nb,m_1,\dots,m_r,\overline{b,\dots,b})\), where there are \(n+1\) collinear points, thus the problem on \({\mathbb F}_n\) reduces to a problem on \({\mathbb P}^2\). The main theorem is proved by using previous known results and via an efficient (but quite technical) inductive procedure by ``diagrams and reductions'', moreover it is indicated how this procedures could be used to deal algorithmically with many more cases, for which the obstacle to give an answer comes from computational complexity.
    0 references
    0 references
    linear systems
    0 references
    fat points
    0 references
    Harbourne-Hirscowitz conjecture
    0 references
    Hirzebruch surface
    0 references
    0 references
    0 references