A lifted square formulation for certifiable Schubert calculus (Q507165)

From MaRDI portal
scientific article
Language Label Description Also known as
English
A lifted square formulation for certifiable Schubert calculus
scientific article

    Statements

    A lifted square formulation for certifiable Schubert calculus (English)
    0 references
    0 references
    0 references
    3 February 2017
    0 references
    An \(m\times n\) matrix M with \(m\geq n\) is rank-deficient if and only if all of its \(n \times n\) minors vanish. This occurs if and only if there is a nonzero vector \(v \in \mathbb C^n\) with \(Mv = 0\). There are \({m\choose n}\) minors and each is a polynomial of degree \(n\) in the \(mn\) entries of \(M\). In local coordinates for \(v\), the second formulation gives \(m\) bilinear equations in \(mn+n-1\) variables, and the map \((M, v)\rightarrow M\) is a bijection over an open dense set of matrices of rank \(4n-1\). The set of rank-deficient matrices has dimension \((m+1)(n-1)\), which shows that the second formulation is a complete intersection, while the first is not if \(m >n\). The principle at work here is that adding extra information may simplify the description of a degeneracy locus. Schubert varieties in the flag manifold are universal degeneracy loci [\textit{W. Fulton}, Duke Math. J. 65, No. 3, 381--420 (1992; Zbl 0788.14044)]. The authors explain how to add information to a Schubert variety to simplify its description in local coordinates. This formulates membership in a Schubert variety as a complete intersection of bilinear equations and formulates any Schubert problem as a square system of bilinear equations. This lifted formulation is both different from and typically significantly more efficient than the primal-dual square formulation in the literature. The motivation comes from numerical algebraic geometry which uses numerical analysis to represent and manipulate algebraic varieties on a computer. It does this by solving systems of polynomial equations and following solutions along curves. For numerical stability, low degree polynomials are preferable to high degree polynomials. More essential is that Smale's \(\alpha\)-theory enables the certification of computed solutions to square systems of polynomial equations, and therefore efficient square formulations of systems of polynomial equations are desirable. Square formulations of Schubert problems also enable the certified computation of monodromy. Since general degeneracy loci are pullbacks of Schubert varieties, these square formulations may lead to formulations of more general problems involving degeneracy loci as square systems of polynomials.
    0 references
    0 references
    Schubert calculus
    0 references
    certification
    0 references
    square systems
    0 references
    0 references
    0 references
    0 references