Comment on \(GL(2, \mathbb R)\) geometry of fourth-order ODEs (Q1009685)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Comment on \(GL(2, \mathbb R)\) geometry of fourth-order ODEs
scientific article

    Statements

    Comment on \(GL(2, \mathbb R)\) geometry of fourth-order ODEs (English)
    0 references
    3 April 2009
    0 references
    \textit{R. Bryant} [in: Complex geometry and Lie theory, Proc. Symp., Sundance/UT (USA) 1989, Proc. Symp. Pure Math. 53, 33--88 (1991; Zbl 0758.53017)], associated a type of G-structure to each fourth order scalar ODE. His \(G\)-structure is invariant under contact transformations. He calculated the lowest order invariants of such an ODE under contact transformations, in terms of his \(G\)-structure. Suppose that the space of solutions of the ODE is Hausdorff. Bryant then demonstrated that vanishing of a certain collection of those invariants is precisely the condition under which the space of solutions inherits an affine connection with a certain ``exotic'' holonomy group. In the article under review, Paweł Nurowski reconsiders Bryant's invariants, and makes them explicit in local coordinates. This greatly clarifies Bryant's more abstract paper. He also expresses the vanishing of Bryant's invariants in terms of the vanishing of the Ricci tensor of a certain connection. Assuming that Bryant's vanishing condition is satisfied, Nurowski explicitly writes out in coordinates the structure equations for a principal bundle \(P\) with connection over the space of solutions of the equation. He writes an explicit map of \(P\) to the frame bundle of the space of solutions. The connection on \(P\) is precisely Bryant's exotic holonomy connection. The map of \(P\) to the frame bundle is the reduction of the frame bundle. This reduction is invariant under parallel transport, giving the exotic holonomy. The expressions in Nurowski's article are very long, but allow direct explicit computation of examples more easily than the more implicit description of Bryant's article. Nurowski then employs his explicit formulas to compute some examples. In particular, he characterizes the ODEs \[ y^{(4)} = 0 \] and \[ y^{(4)} = \frac{4 y^{(3)}}{3 y^{(2)}} \] by the curvature tensor of their induced connections on their solution spaces. He also finds inhomogeneous examples of ODE satisfying Bryant's vanishing condition. Reviewer's remark: The reader should be careful of two very small problems. The first problem: The space of solutions of a fourth order ODE will not necessarily be a smooth manifold. If it is, the ODE is called amenable. Roughly speaking, amenability means that the dynamics of the ODE are topologically trivial. The second problem: Nurowski's ``principal bundle'' \(P\) is a principal bundle if and only if the induced coframing on each leaf of his Pfaffian system \(\mathcal{V}\) is complete. Even should \(P\) turn out to be a principal bundle, it would still have non-Hausdorff base space unless the ODE is amenable. Theorems 2.1, 4.1 and 4.5 are incorrect as stated, and should be read with these two problems in mind. We can avoid both problems by saying that Bryant's vanishing conditions occur just when the structure equations on the manifold \(P\) are locally isomorphic to the structure equations of a \(\text{GL}\left(2,\mathbb{R}\right)\)-geometry on a principal bundle over a 4-manifold, given by Nurowski's explicit expressions. We cannot necessarily identify that 4-manifold with the space of solutions of the ODE.
    0 references
    0 references
    0 references
    0 references
    0 references
    geometrization of ODEs
    0 references
    Wilczynski invariants
    0 references
    Tresse invariants
    0 references
    0 references
    0 references
    0 references