Vanishing Hessian, wild forms and their border VSP (Q2205579)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Vanishing Hessian, wild forms and their border VSP
scientific article

    Statements

    Vanishing Hessian, wild forms and their border VSP (English)
    0 references
    0 references
    0 references
    0 references
    20 October 2020
    0 references
    \textit{A. Iarrobino} and \textit{V. Kanev} introduced several notions of rank for a homogeneous polynomial \(F\in S = \mathbb C [x_0,\dots,x_n]\) in their study of the problem of decomposition into sums of powers of linear forms [Power sums, Gorenstein algebras, and determinantal loci, Berlin: Springer (1999; Zbl 0942.14026)], including border rank \(r (F)\), smoothable rank \(sr (F)\), and cactus rank \(cr (F)\). In general, \(r (F) \leq sr (F)\) and \(F\) is a \textit{wild form} if \(r (F) < sr (F)\). The dual ring \(T = \mathbb C [y_0, \dots, y_n]\) acts on \(S\) by differentiation via \(y_i \circ x_j = \delta_{i,j}\) and \(F\) is \textit{concise} if \(\text{Ann} (F) \subset T\) contains no linear forms. A concise form \(F\) satisfies \(n+1 \leq r (F)\) and has \textit{minimal border length} if \(r (F) = n+1\). The main theorem states that if \(F\) is a concise \(d\)-form of minimal border rank and \(d \geq 3\), then \(F\) is wild \(\iff cr (F) > r (F) \iff \text{Hess} (F) =0\), where \(\text{Hess} (F)\) is the determinant of the Hessian matrix. In particular, it follows from a result of \textit{T. Maeno} and \textit{J. Watanabe} [Ill. J. Math. 53, No. 2, 591--603 (2009; Zbl 1200.13031)] that for such \(F\), the Artinian Gorenstein \(\mathbb C\)-algebra \(T/\text{Ann} (F)\) does not have the strong Lefschetz property. They show by example that their equivalence fails for wild forms that are concise but not of minimal border rank. Their method of proof focuses on the finite length algebra \(A=T/\text{Ann} F\), showing that the Gorenstein property is equivalent to symmetry of the multiplication tensors. They also work out a few examples of the border variety of sums of powers \(\underline{\text{VSP}} (F,r)\) introduced by \textit{W. Buczynska} and \textit{J. Buczynski} [``Apolarity, border rank and multigraded Hilbert scheme'', Preprint, \url{arXiv:1910.01944}]. For a \(d\)-form \(F\), the \(\underline{\text{VSP}} (F,r)\) is essentially the subset in the main component of the Hilbert scheme \(\text{Hilb}^r (\mathbb P^n)\) consisting of zero dimensional schemes containing the subscheme defined by \(\text{Ann} (F)\). They give one infinite sequence of wild forms \(G_d\) of increasing degree in which \(\underline{\text{VSP}} (G_d,r (G_d))\) is a projective space and another sequence of wild cubics \(F_n\) for which \(\underline{\text{VSP}} (F_n,r (F_n))\) is reducible.
    0 references
    0 references
    0 references
    homogeneous polynomials
    0 references
    smoothable rank
    0 references
    border rank
    0 references
    cactus rank
    0 references
    concise forms
    0 references
    vanishing Hessian determinant
    0 references
    0 references
    0 references