Symplectification of Rank 2 Distributions, Normal Cartan Connections, and Cartan Prolongations
Nicklas Day, Boris Doubrov, Igor Zelenko
TL;DR
The paper addresses local equivalence for bracket-generating rank $2$ distributions with a $5$-dimensional cube by integrating the Doubrov–Zelenko symplectification with Tanaka–Morimoto theory. It proves that the symplectified distribution $\mathrm{Symp}(D)$ carries a normal Cartan connection and is locally equivalent to the $(n-4)$th Cartan prolongation $\mathrm{pr}^{(n-4)}(D)$ at generic points for $n>5$, while the Tanaka symbols unify already at the $(n-5)$th prolongation. The $n=5$ case corresponds to a $G_2$-parabolic geometry, where normal Cartan connections exist for both $D$ and $\mathrm{Symp}(D)$. Moreover, lower prolongations fail to admit a linear invariant normalization condition, establishing that $(n-4)$ is the minimal prolongation supporting Tanaka–Morimoto normal Cartan geometry. Altogether, the results provide a uniform Cartan-theoretic framework for these distributions and clarify the role of Cartan prolongations in achieving canonical connections.
Abstract
We study the Doubrov--Zelenko symplectification procedure for rank $2$ distributions with $5$-dimensional cube -- originally motivated by optimal control theory -- through the lens of Tanaka--Morimoto theory for normal Cartan connections. In this way, for ambient manifolds of dimension $ n \geq 5 $, we prove the existence of the normal Cartan connection associated with the symplectified distribution. Furthermore, we show that this symplectification can be interpreted as the $(n-4)$th iterated Cartan prolongation at a generic point. This interpretation naturally leads to two questions for an arbitrary rank $2$ distribution with $5$-dimensional cube: (1) Is the $(n-4)$th iterated Cartan prolongation the minimal iteration where the Tanaka symbols become unified at generic points? (2) Is the $(n-4)$th iterated Cartan prolongation the minimal iteration admitting a normal Cartan connection via Tanaka--Morimoto theory? Our main results demonstrate that: (a) For $n > 5$, the answer to the second question is positive (in contrast to the classical $n = 5$ case from $G_2$-parabolic geometries); (b) For $n \geq 5$, the answer to the first question is negative: unification occurs already at the $(n-5)$th iterated Cartan prolongation.
