Curves having one place at infinity and linear systems on rational surfaces (Q995626): Difference between revisions

Given \(r+1\) general points \(p_0,\dots,p_r\) in the projective plane, it is not known whether the linear system of plane curves of a given degree \(d\), having multiplicities \(\geq M_i\) at each \(p_i\) has the expected dimension (i.e. \(\max \{0, {d+2\choose 2} - \sum _{i=0}^r {m_i+1\choose 2}\}\)). A well known conjecture due to many authors (the first version was given by B. Segre in 1961) states that such linear system has the expected dimension unless there is a (-1)-curve (i.e. a curve whose strict transform on the blow-up of the plane at the \(p_i\)'s is an exceptional divisor) which is a fixed multiple component for the curves in the system. In this paper, which provides also a quite detailed summary of the known results on the subject, the conjecture is proved for a wide range of choices for the vector \((m_0,\dots,m_r)\) (and any choice of \(d\)). The result is based on the study of certain blow-ups of the plane, namely considering curves \(C\) which are rational, smooth except for one point \(p\) on the line \(H\) ``at infinity'', and such that \(C\cap H = p\) and \(C\) is unibranched at \(p\); a particular surface \(X_C\) is then considered (which comes from repeted blowings up at \(p\)) and the speciality of linear systems on \(X_C\) is studied, and then related (via semicontinuity) to the linear systems we are interested in. Given any vector \((m_0,\dots,m_r)\), a bound for \(d\) in order to have that the linear system in question is non-special is given, and for certain families of such vectors it is shown that the bound is sharp, thus proving the conjecture in those cases.