Fundamental form and the Cartan tensor of \((2,5)\)-distributions coincide (Q867457)

(From the Introduction): In a previous paper [Differ. Geom. Appl. 24, No. 3, 235--259 (2006; Zbl 1091.58002)], the author constructed a special differential invariant, the fundamental form of generic rank-2 vector distributions \(D\) on an \(n\)-dimensional manifold (\(n \geq 5\)). The construction is based on the new variational approach to differential invariants of geometric structures on manifolds, which is completely different from the Cartan method of equivalence (geometric structures mean submanifolds of the tangent bundles; for example, a Riemannian metric is defined by a spherical bundle of the tangent bundle). This approach can be briefly described as follows. The main objects are extremals of the geometric structure, which are special curves in the cotangent bundle. To any extremal, one can assign a special curve in some Lagrange Grassmannian called the Jacobi curve. The Jacobi curve can be considered as a generalization of the spaces of Jacobi fields along Riemannian geodesics. Symplectic invariants of the Jacobi curve (i.e., invariants w.r.t. the natural action of the linear symplectic group on the Lagrange Grassmannian, where this Jacobi curve lives) are automatically invariants of the geometric structure. In this way, the problem of finding invariants of geometric structures can be reduced to a much more concrete problem of finding symplectic invariants of curves in Lagrange Grassmannians. Both the Riemannian curvature tensor and the Weyl projective tensor in Riemannian geometry can be obtained as a realization of some basic symplectic invariants of Jacobi curves (parametrized in the first case and unparametrized in the second case) of the Riemannian extremals (the natural lifts of Riemannian geodesics to the cotangent bundle). The classical Weyl conformal tensor of (pseudo-)Riemannian structure can also be obtained in this way if one works with extremals and their Jacobi curves of the geometric structure given by the field of cones (complexified in the Riemannian case), defining the conformal structure (in the pseudo- Riemannian case, these extremals are natural lifts of null geodesics to the cotangent bundle). An application of this point of view to vector distributions leads to the fundamental form, constructed in [op. cit.], which, therefore, can be considered as an analog of the Weyl projective or conformal tensors in the Riemannian geometry. In the case \(n = 5\), this fundamental form can be realized as an invariant homogeneous polynomial of degree 4 on each plane \(D(q)\). This realization is called the tangential fundamental form. The tangential fundamental form has the same algebraic nature as the Cartan tensor, and the natural problem is to compare these invariants. In the present paper the author proves that his fundamental form coincides (up to constant factor) with the Cartan tensor. Since these two invariants were constructed in completely different ways, the comparison of them turned out to be not so easy task. This result gives also a new geometric explanation for the existence of the Cartan tensor, which is the existence of a special degree-four differential on a curve in the Lagrange Grassmannian (or, more general, in Grassmannians of half-dimensional subspaces). It also allows to find natural analogs of the Cartan tensor in the Riemannian geometry in terms of the Weyl projective tensor or the Weyl conformal tensors. The paper is organized as follows. In Section 2, the construction of the fundamental form of (2,5)-distribution with the small growth vector (2, 3, 5) is described. In Section 3 the author describes the main steps of the construction of the Cartan tensor, rewriting all formulas required for the comparison of our and Cartan's invariants and formulate the main theorem of the paper (Theorem 2). In Section 4, he gives a frame naturally adapted to the distribution and derives the formula for the fundamental form in terms of its structural functions. This formula is important not only in the proof of the main theorem: in many cases, it is more efficient from the computational point of view than the method given in [op. cit.]. In Section 5, the main theorem is proved.