Geometric aspects of rank-3 vector bundles over surfaces and 2-plane distributions on 5-manifolds
Brandon P. Ashley, Michael T. Schultz · 2025
Read Geometric aspects of rank-3 vector bundles over surfaces and 2-plane distributions on 5-manifolds by Brandon P. Ashley, Michael T. Schultz (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-25918.txt.
Introduction
Brandon P. Ashley and Michael T. Schultz connect the geometry of surfaces in projective space with distributions on five dimensional spaces. They study rank three vector bundles over a surface, meaning three dimensional vector spaces attached at each point, equipped with a connection that defines horizontal directions. Removing the zero section leaves a five dimensional total space with a field of two dimensional planes.
Ashley and Schultz show that any surface in three dimensional projective space can be tied to such a structure. They build a dictionary that translates features of the surface, in the language of projective differential geometry, into the growth pattern of the plane field. The growth pattern records how repeated brackets of allowed motions fill new directions.
Catalog record
| CREATOR | Brandon P. Ashley, Michael T. Schultz |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 90 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-25918
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Geometric aspects of rank-3 vector bundles over surfaces and 2-plane distributions on 5-manifolds Brandon P. Ashley, Michael T. Schultz Abstract . We study geometric aspects of horizontal 2-plane distributions on the 5-dimensional total space of a rank-3 vector bundle equipped with connection over a surface. We show that any surface in 3-dimensional projective space can be associated to such a geometric structure in 5-dimensions, and establish a dictionary between the projective differential geometry of the surface and the growth vector of the 2-plane distribution. 2020 Mathematics Subject Classification 35A30, 53A20, 53B15 1. Introduction In differential geometry, the role of a connection on the tangent bundle of a manifold is of central importance to the intrinsic geometry of the space. Gauge theoretic considerations asks the question of how more general geometric structures may be associated to a given manifold. One interaction between these two areas can be seen in the induced extrinsic geometry obtained by immersing the manifold into an ambient space and restricting the ambient geometry onto the image. The classical differential geometry of surfaces in 3 -dimensional euclidean space \mathbb{R}^{3} illustrates this interaction. Let \Sigma be a smooth surface and suppose that \psi:\Sigma\to\mathbb{R}^{3} is an immersion. Then there is a natural orthogonal splitting of the rank-3 vector bundle B=\psi^{*}T\mathbb{R}^{3}\to\Sigma as B=T\Sigma\oplus N_{\Sigma} induced by the first fundamental form of \Sigma imparted by the immersion \psi . Here N_{\Sigma} is the normal bundle of \Sigma\subset\mathbb{R}^{3} , which we identify with its image \psi(\Sigma) by working locally, and may regard N_{\Sigma}\to\Sigma and B\to\Sigma as being a trivial vector bundles. The fibres of B carry the obvious \mathrm{GL}(3) representation, and we may view B as being an associated bundle of a left principal \mathrm{GL}(3) -bundle over \Sigma . The first fundamental form induces a canonical symmetric metric connection on T\Sigma , the Levi-Civita connection, and a the normal bundle N_{\Sigma} inherits a connection determined by the second fundamental form. These two connections fit together to form a \mathrm{GL}(3) -connection \nabla on the rank-3 bundle B as follows. Let (x,y) be local coordinates on \Sigma . Then the immersion \psi can be locally represented by (1.1) (x,y)\mapsto z(x,y)=\left(z_{1}(x,y),z_{2}(x,y),z_{3}(x,y)\right)\in\mathbb{R}^{3}\,. Then p=z_{x} and q=z_{y} locally span the tangent bundle T\Sigma , and we may complete to a frame e=(\sigma,p,q) of B by choosing an appropriate section \sigma of the normal bundle N_{\Sigma} with unit norm. It is well known that z satisfies the second order system of differential equations (1.2) \displaystyle z_{xx}=\Gamma^{1}_{11}z_{x}+\Gamma^{2}_{11}z_{y}+l\sigma\,, \displaystyle z_{xy}=\Gamma^{1}_{12}z_{x}+\Gamma^{2}_{12}z_{y}+m\sigma\, \displaystyle z_{yy}=\Gamma^{1}_{22}z_{x}+\Gamma^{2}_{22}z_{y}+n\sigma\,, where \Gamma^{i}_{jk} are the Christoffel symbols defining the Levi-Civita connection, and l,m,n are the coefficients of the second fundamental form. Equation ( 1.2 ), known classically as the Gauss formula, yields a \mathfrak{gl}(3) -valued 1-form \omega on \Sigma given by (1.3) \omega=\left(\begin{array}[]{ccc}0&dx&dy\\ l\,dx+m\,dy&\Gamma^{1}_{11}\,dx+\Gamma^{1}_{12}\,dy&\Gamma^{2}_{11}\,dx+\Gamma^{2}_{12}\,dy\\ m\,dx+n\,dy&\Gamma^{1}_{12}\,dx+\Gamma^{1}_{22}\,dy&\Gamma^{2}_{12}\,dx+\Gamma^{2}_{22}\,dy\\ \end{array}\right)\,, encoding the connection 1-forms of \nabla . This connection naturally contains information on the geometry of \Sigma\subset\mathbb{R}^{3} . For example, a simple calculation of the curvature tensor \mathbf{R}=d\omega-\omega\wedge\omega reveals that the classical Gauss-Codazzi equations (1.4) \displaystyle m_{x}-l_{y}=l\Gamma^{1}_{12}+m(\Gamma^{2}_{12}-\Gamma^{1}_{11})-n\Gamma^{2}_{11}\,, \displaystyle n_{x}-m_{y}=l\Gamma^{1}_{22}+m(\Gamma^{2}_{22}-\Gamma^{1}_{12})-n\Gamma^{2}_{12} are simply the vanishing of the certain components of the curvature tensor; it is a basic result that \mathbf{R}=0 if \Sigma\subset\mathbb{R}^{3} is a plane. One may obtain further identities by utilizing the differential Bianchi identity d_{\nabla}\mathbf{R}=0 , where d_{\nabla} is the exterior covariant derivative. On the other hand, it is useful to insist that the frame e be orthonormal in B , which is equivalent to a reduction of the structure group of B from \mathrm{GL}(3) to \mathrm{SO}(3) . This leads to the well known Darboux frame away from umbilic points of \Sigma (see e.g., [ 13 , Ch. 2] ), which leads to an analogous connection that is canonically flat. Other geometric information can be seen by examining covariant derivatives of the curvature \mathbf{R} with respect to \nabla , or equivalently by studying the infinitesimal holonomy of \nabla . In this case, a coarse invariant - the growth vector - can be obtained by studying the derived flag of the exterior differential system (EDS) of the connection 1-forms pulled back to the 5-dimensional total space M^{5}=\mathrm{Tot}(B) of the vector bundle B . Then it is easy to see that the differential ideal \Omega:=\langle\pi^{*}\omega\rangle_{\mathrm{diff}} generated by three 1-forms on M^{5} (1.5) \displaystyle\omega_{\sigma} \displaystyle=d\sigma-pdx-qdy\,, \displaystyle\omega_{x} \displaystyle=dp-(l\sigma+p\Gamma^{1}_{11}+q\Gamma^{2}_{11})dx-(m\sigma+p\Gamma^{1}_{12}+q\Gamma^{2}_{12})dy\,, \displaystyle\omega_{y} \displaystyle=dq-(m\sigma+p\Gamma^{1}_{12}+q\Gamma^{2}_{12})dx-(n\sigma+p\Gamma^{1}_{22}+q\Gamma^{2}_{22})dy\,, determine a rank-3 EDS \Omega=\langle\omega_{\sigma},\omega_{x},\omega_{y}\rangle_{\mathrm{diff}} on M^{5} from the connection 1-forms on \Sigma . Dualizing, \Omega determines a rank-2 distribution \Delta=\Omega^{\perp}\subset TM^{5} , i.e., \Delta is defined as the rank-2 sub-bundle of TM such that \Omega=\mathrm{ann}(\Delta) is the annihilator of \Delta . The vector fields generating \Delta are easily computed as (1.6) \displaystyle X \displaystyle=\partial_{x}+p\partial_{\sigma}+(l\sigma+p\Gamma^{1}_{11}+q\Gamma^{2}_{11})\partial_{p}+(m\sigma+p\Gamma^{1}_{12}+q\Gamma^{2}_{12})\partial_{q}\,, \displaystyle Y \displaystyle=\partial_{y}+q\partial_{\sigma}+(m\sigma+p\Gamma^{1}_{12}+q\Gamma^{2}_{12})\partial_{p}+(n\sigma+p\Gamma^{1}_{22}+q\Gamma^{2}_{22})\partial_{q}\,. Generically, the distribution is bracket generating , meaning that the derived flag of \Delta - the collection of all sub-bundles of TM^{5} generated by iterated Lie brackets of \Delta=\mathrm{span}\,\{X,Y\} - generates all of TM^{5} . In this generic case 1 1 1 There is another bracket generating case, that of the contact distribution , which has growth vector (2,3,4,5) . This however, is not generic behavior. , Cartan showed [ 6 ] that such distributions have growth vector (2,3,5) , meaning that \mathrm{rank}(\Delta^{\prime})=\mathrm{rank}([\Delta,\Delta])=3 and \mathrm{rank}(\Delta^{(2)})=\mathrm{rank}([\Delta^{\prime},\Delta^{\prime}])=5 . It is a basic consequence of the connection \nabla that there is a splitting TM^{5}=H\oplus V into horizontal and vertical distributions, where the rank-2 distribution H=\pi^{*}T\Sigma\cong\Delta and V is the rank-3 distribution tangent to the fibres of B . Then the curvature \mathbf{R} of \nabla vanishes if and only if \Delta is integrable in the sense of Frobenius, that is \Delta^{\prime}=\Delta . In the case of nonvanishing curvature, we symbolically have \Delta^{\prime}\subset V generated by \mathbf{R} and \Delta^{(2)}\subset V generated by the covariant derivative \nabla\mathbf{R} . Cartan demonstrated in loc. cit. that the local equivalence problem associated to generic bracket generating 2-plane distributions on a given M^{5} exhibits a surprising connection to the (split-real form of the) simple Lie algebra \mathfrak{g}_{2} , which arises as the maximal local symmetry algebra of the distribution. Implicit in Cartan’s construction, and later elucidated by Nurowski [ 20 ] , is the existence of a conformal structure of split signature (3,2) on TM^{5} in which \Delta is totally null, and whose Weyl tensor determines the local obstructions to maximal \mathfrak{g}_{2} symmetry. The conformally flat case is necessary and sufficient for maximal symmetry. The geometry of such distributions continue to be of interest into modernity [ 21 , 36 , 39 , 25 , 24 , 26 ] , and have famously arisen in connection to the geometry of surfaces rolling without slipping or twisting [ 5 , 40 , 10 ] , geometry of twistor distributions on split signature semi-riemannian 4-manifolds [ 2 , 22 ] , and Darboux integrability of certain nonlinear PDE in the plane [ 35 , 34 , 3 ] . In this article we initiate a study of relationships between two subjects shaped by the work of E. Cartan: on the one hand, his work [ 7 ] from 1920 on the differential geometry of surfaces in 3-dimensional projective space - in particular, those that admit projective deformations; on the other hand, his work [ 6 ] from 1910 on 2-plane distributions in 5-dimensions. To set the stage for our work, we utilize in § 2 the well known horizontal and vertical distributions