Secant varieties to osculating varieties of Veronese embeddings of \(\mathbb P^n\) (Q1014588)

A well-known theorem by \textit{J. Alexander} and \textit{A. Hirschowitz} [J. Algebr. Geom. 4, No.~2, 201--222 (1995; Zbl 0829.14002)] states that all the higher secant varieties of \(V_{n,d}\), the \(d\)-tuple embedding of \(\mathbb{P}^n\), have the expected dimension, with few known exceptions. In this paper the authors study the same problem for the \(T_{n,d}\), the tangential variety to \(V_{n,d}\). A conjecture about secant varieties of tangential varieties was stated by \textit{M. V. Catalisano}, \textit{A. V. Geramita} and \textit{A. Gimigliano} [Proc. Am. Math. Soc. 130, No. 4, 975--985 (2002; Zbl 0990.14021)]: Conjecture 1: The secant variety \(\sigma_s(T_{n,d})\) has the expected dimension \(\min\{2sn+s-1,{n+d \choose n}-1\}\), unless {\parindent=6mm \begin{itemize}\item[(i)] \(d=2\), \(2 \leq 2s <n\) \item[(ii)] \(d=3\), \(s=n=2,3,4\). \end{itemize}} In the same paper the conjecture was proved for \(d=2\) (any \(s,n\)) and for \(s \leq 5\) (any \(d,n\)), while, \textit{E. Ballico} [J. Algebra 288, No. 2, 279--286 (2005; Zbl 1072.14533)] proved it for \(n=3\) (any \(s,d\)). As a first step in this paper, the authors prove that if Conjecture 1 holds for \(d=3\) then it holds also for \(d \geq 4\) (and any \(n,s\)). Their proof is based mainly on Horace differential lemma. A more general problem can be considered: Let \(O_{k,n,d}\) be the \(k\)-osculating variety to \(V_{n,d}\subset \mathbb{P}^N\), and study its \((s-1)\)th higher secant variety \(\sigma_s(O_{k,n,d})\). Again we want to determine all \(s\) for which \(\sigma_s(O_{k,n,d})\) is defective, i.e. for which its dimension is strictly less than the expected one. In a previous paper [Can. J. Math. 59, No. 3, 488--502 (2007; Zbl 1118.14060)], the authors found that \(\sigma_s(O_{k,n,d})\) is defective if and only if a certain 0-dimensional scheme \(Y \subset \mathbb{P}^n\) does not impose independent condition to forms of degree \(d\) in \(R:=k[ x_0, \dots, x_n ]\). The scheme \(Y=Z_1 \cup \dots \cup Z_s\) is supported at \(s\) generic points \(P_1, \dots, P_s \in \mathbb{P}^n\), and at each of them the ideal of the scheme \(Z_i\) is such that \(I_{P_i}^{k+2} \subset I_{Z_i} \subset I_{P_i}^{k+1}\). Then the authors state the following Conjecture 2: The secant variety \(\sigma_s(O_{k,n,d})\) is defective if and only if {\parindent=6mm \begin{itemize}\item[(i)] \(h^1(\mathcal{I}_X(d)) > \max \{0, deg(Y)-{d+n \choose n} \} \) or if \item[(ii)] \(h^0(\mathcal{I}_T(d)) > \max \{0, {d+n \choose n}-\deg(Y) \}\), \end{itemize}} where \(X=(k+1)P_1 \cup \dots \cup (k+1)P_s\) and \(T=(k+2)P_1 \cup \dots \cup (k+2)P_s\). As main result of the paper, the authors prove the previous conjecture for \(n=2\).