An asymptotic version of Dumnicki's algorithm for linear systems in \({\mathbb{CP}^2}\) (Q956702)

Let \(\mathbb{P}^2\) be the complex projective plane. For \(r\) general points on \(\mathbb{P}^2\) and any positive integer \(m\) denote by \({\mathcal L}_d(m^r)\) the linear system of degree \(d\) plane curves with multiplicity at least \(m\) in those \(r\) points. A basic problem on \({\mathcal L}_d(m^r)\) is that of their dimension. In particular one can try to detect those of the spaces \({\mathcal L}_d(m^r)\) such that their dimension is not the expected one, i.e. they are special. This is related, see the paper under review for references, with the celebrated Nagata conjecture whis is stated (in terms of Seshadri constants) as follows: for \(r>9\), on the blowing up of \(\mathbb{P}^2\) in general \(r\) points the divisor \(H-\sqrt{{1 \over r}}E\) is nef, being \(H\) the pullback of a line in \(\mathbb{P}^2\) and \(E\) the total exceptional divisor. In a previous paper the author proved an asymptotic result (see Thm 1.1 in the current paper) showing that for a sequence of pairs \((d_i,m_i)\) (\(i\) a positive integer) such that \({d_i^2 \over rm_i^2}\) converges to \(1\) (when \(i\) tends to \(\infty\)) non speciality of \({\mathcal L}_{d_i}((m_i+1)^r)\) implies Nagata conjecture. Then, asymptotic results of non speciality of linear systems seem of interest and should provide lower bounds for certain Seshadri constants. A non-speciality criterion is given by Dumnicki (see Prop. 1.2) whose main idea is to consider not the complete linear system \({\mathcal L}_d(m^r)\) but the subsystem defined by imposing furthermore that the monomials of the equation of a curve in the linear system belongs to a particular subset of degree \(d\)-monomials. The paper under review develops an asymptotic version of Dumnicki's criterion and uses it to provide lower bounds on Seshadri constants. In particular, for \(r=10\) the bound \(4/13\) is obtained.