Higher secant varieties of \(\mathbb P^n \times \mathbb P^n\) embedded in bi-degree \((1,d)\) (Q649850)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Higher secant varieties of \(\mathbb P^n \times \mathbb P^n\) embedded in bi-degree \((1,d)\)
scientific article

    Statements

    Higher secant varieties of \(\mathbb P^n \times \mathbb P^n\) embedded in bi-degree \((1,d)\) (English)
    0 references
    0 references
    0 references
    6 December 2011
    0 references
    If \(X \subset {\mathbb P}^N\) is a projective variety, let \(\sigma_s(X) = \overline{\cup _{P_1,\dots,P_s\in X}<P_1,\dots,P_s>}\) be its \(s^{th}\) higher secant variety. The expected dimension of \(\sigma_s(X)\) is \(\min \{N, s(\dim X )+s-1\}\) and it is a widely studied question to determine whether \(\sigma_s(X)\) has the expected dimension or not (in which case we say that \(X\) is \((s-1)\)-defective). In this paper it is studied the case when \(X\) is a Segre-Veronese variety of two factors, embedded in bidegree \((1,d)\); i.e. \(X\) is the embedding of \({\mathbb P}^n\times{\mathbb P}^m\) into \({\mathbb P}^N\) , \(N=(n+1)\times {m+d \choose d}-1\), defined via the sheaf \({\mathcal O}(1,d)\). The secant of Segre-Veronese varieties are of interest also because they parameterize partially symmetric tensors of given rank. The main result in the paper is to obtain two bounds \(s_1, s_2\) (\(s_1\leq s_2\)), such that \(\dim \sigma_s(X) = N\) for \(s\geq s_2\), and \(\dim \sigma_s(X) = s(n+m+1)-1\) for \(s\leq s_1\) (the expected dimensions). In this way the problems remains open only for \(s_1< s < s_2\) ; moreover \(s_1 , s_2\) are such that \(s_2-s_1\) is either 0 or \(n\). The main tool in the proof is to reduce the problem (via Terracini's Lemma and considering an affine open set and re-homogenizing in order to work in \( {\mathbb P}^{n+m}\) ) to the study of the Hilbert function of a subscheme of \( {\mathbb P}^{n+m}\) which is of type: \(dH_1+H_2+2P_1+\dots+2P_s\), where \(H_1, H_2\) are generic hyperplanes and \(P_1,\dots,P_s\) are generic points (here the schemes \(dH_1\), \(2P_i\) are defined by the homogeneous ideals \((I_{H_1})^d\), \((I_{P_i})^2\), respectively). The problem is solved by induction, making use of Castelnuovo inequality.
    0 references
    0 references
    Secant varieties
    0 references
    Segre varieties
    0 references
    Veronese varieties.
    0 references
    0 references
    0 references