Some defective secant varieties to osculating varieties of Veronese surfaces (Q2492120)

In the last years several authors have studied problems related to the (classical) question of determining the dimension of the higher secant varieties to a projective variety \(X\). In particular, the case when \(X\) parameterizes forms (or tensors) is of interest. In the present paper \(X\) is the \(k\)-osculating variety \({\mathcal O}_{k,d}\) to the \(d^{th}\)-Veronese surface (the \(d\)-ple embedding of \({\mathbb P}^2\)), which parameterizes forms of type \(L^{d-k}F\), where \(L\in S_1\), \(F\in S_k\), \(S=K[x_0,x_1,x_2]\). Let \({\mathcal O}^s_{k,d}\) be the variety of the \(s\)-secant \((s-1)\)-planes to \({\mathcal O}_{k,d}\); via Terracini's Lemma and inverse systems one can relate the computation of \(\dim {\mathcal O}_{k,d}^s\) to the one of the Hilbert function (in degree \(d\)) of a 0-dimensional scheme \(Z\subset {\mathbb{P}}^2\) which is supported to \(s\)-generic points. Such Hilbert function is not easy to determine, but with a clever use of quadratic transformations and Horace method (the scheme \(Z\) is ``quite close'' to a scheme of fat points) the authors are able to completely solve the problem when \(s=4,5,6\) or 9 (and they say that also cases \(s=7,8\) could be delt with in the same way). It turns out that for \(s=4,9\) the Hilbert function of \(Z\) is always the expected one, hence the same happens for \(\dim {\mathcal O}_{k,d}^s\), while for \(s=5,6\) all values of \(k,d\) for which \({\mathcal O}_{k,d}^s\) is defective are described. The results confirm a conjecture stated by the authors and others in a previous paper (to appear in Can. J. of Math., see \texttt{math.AG/0403132}).