Topologically stable manifolds for index-$1$ singular dominated splittings
Sylvain Crovisier, Dawei Yang
TL;DR
This work develops a robust framework for topological stability and contraction in flows with index-1 singular dominated splittings, without requiring full hyperbolicity. By leveraging a nonlinear Poincaré flow, a local fibered flow model, identifications between nearby fibers, and a one-dimensional center-stable direction, the authors construct 2D topologically stable discs for μ-a.e. points under mild hypotheses on periodic orbits and toral dynamics. The approach combines plaque theory, δ-interval dynamics, and a lifted dominated-structure analysis to obtain measure-stability results and a Palis-density–type contraction conclusion for compact invariant sets. The results advance understanding of how stable manifolds can persist in singular, non-uniformly hyperbolic settings and pave the way for applications to three-dimensional vector fields and related conjectures.
Abstract
For $C^2$ vector fields, we study regular ergodic measures whose supports admit singular dominated splittings with one of the bundles having dimension $1$. For such a measure $μ$, we prove that if any periodic orbit within the support of $μ$ (when it exists) has at least one negative Lyapunov exponent, and if the dynamics on the support of $μ$ is not topologically equivalent to an irrational flow on a $2$-torus, then $μ$-almost every point $x$ admits a $2$-dimensional topologically stable manifold $V^s(x)$: we mean that $V^s(x)$ is an embedded disc such that the orbit any point within it converges to the orbit of $x$ up to a time-reparametrization. Note that we do not assume any hyperbolicity for $μ$. We also establish an analogous conclusion for compact invariant sets $Λ$ with a singular dominated splitting, assuming some mild contraction property (any regular ergodic measure properly supported in $Λ$ must have at least one negative Lyapunov exponent). This result will be used in our future work on the Palis density conjecture for three-dimensional vector fields.
