Star flows and multisingular hyperbolicity (Q2039588)

This paper analyzes flows on manifolds of arbitrary dimension and looks at the robustness under perturbation of certain dynamical properties, such as transitivity and various forms of hyperbolicity. For discrete-time dynamical systems, many properties can be studied by looking at the action of the derivative on the tangent bundle. For flows, in contrast, the situation is more complicated and there are multiple possible approaches. One is to consider the tangent flow, defined on the entire tangent bundle \(TM\) of \(M\). Another approach is to look at the linear Poincaré flow, which is defined on the normal bundle. In fact, the linear Poincaré flow quotients out any shearing that occurs along the direction of the flow and only captures the effect of the dynamics on vectors perpendicular to the direction of the flow. If the flow has no singularities, then the linear Poincaré flow is a better tool to analyze the dynamical properties of the flow. If the flow has singularities (points where the vector field defining the flow is zero), then the linear Poincaré flow is not even defined at these singularities. In the paper, the authors define an \emph{extended linear Poincaré flow.} Away from the singularities, this is the same as the linear Poincaré flow. At a singularity \(\sigma\), the flow is defined on a carefully defined subset of the projectivization of the tangent space \(T_{\sigma}M\). This only partial resolves the problem, however, as quotienting has removed the effect of shearing along orbits. Away from singularities, the shearing does not effect the qualitative behaviour of the dynamics. In a neighbourhood of a singularity, however, the shearing is important and cannot be ignored. To overcome this, the authors define multiplicative cocycles associated to the flow which, together with the extended linear Poincaré flow, retain all of the important dynamical information. Using both the extended linear Poincaré flow and these cocycles, they define the notion of \emph{multisingular hyperbolicity.} This is a generalization to higher dimensions of singular hyperbolicity in dimension 3. One of the major results of the paper concerns star flows. A vector field is called a \emph{star flow} if every periodic orbit of any vector field \(C^1\)-close to it is hyperbolic. In dimension 3, the Lorenz attractor is the motivating example for both singular hyperbolicity and star flows. The authors show in any dimension that every multisingular hyperbolic flow is a star flow and that an open and dense subset of star flows are multisingular hyperbolic. Readers of the paper should note that many of the novel concepts are introduced without full details in Sections 1 and 2, but that this is followed by very detailed, rigorous, and complete definitions in Section 3.