Hölder regularity of Grobman-Hartman theorem for dynamic equations on measure chains (Q723590)

It is well-known that the conjugation in the Hartman-Grobman theorem is continuous, but already Hartman gave an example that it is not even Lipschitz. Later, one was able to show at least Hölder-continuity. This paper constructs such a Hölder-regular conjugation between a semilinear dynamical equation on a time scale (measure chain) \[ x^\Delta = A_1(t)x+f(t,x,y),\quad y^\Delta=A_2(t)y \] and its linearization \[ x^\Delta = A_1(t)x,\quad y^\Delta=A_2(t)y, \] provided \(x^\Delta=A_1(t)x\) possesses an exponential dichotomy, while \(f\) has a small Lipschitz constant and satisfies a boundedness assumption. It seems some invertibility condition in the unstable bundle is missing in the definition of an exponential dichotomy. Moreover, regressivity of the (nonlinear) dynamic equation appears crucial, since the Hartman-Grobman theorem fails otherwise. Furthermore, the claim in the introduction that \textit{C. Pötzsche} [J. Differ. Equations 245, No. 5, 1210--1242 (2008; Zbl 1190.34120)] ``extended the main results in'' [\textit{Y. Xia} et al., J. Differ. Equations 235, No. 2, 527--543 (2007; Zbl 1126.34030)] ``to a more general form. He investigated the behavior of the topological conjugacy under parameter variation.'' is misleading. Indeed, independent from, and generalizing [loc. cit.], a topolocical linearization result in the presence of a center manifold was given.