5
$\begingroup$

I'm trying to find references to approximations of topological dynamical systems in the following sense:

A topological dynamical system $(X, f)$ consists of a topological space (typically compact Hausdorff, or even metrisable) and a continuous $f : X \to X$. We can topologise $C(X, X)$ in various ways, for instance with pointwise / compact-open / uniform topologies. Given such a topology $\tau$, we denote the topological space by $C_\tau(X, X)$.

Are there theorems along the lines of "Given a sequence $(f_n) \subseteq C_\tau(X, X)$, if $(f_n) \to f$ then the dynamical systems $(X, f_n)$ 'approximate' $(X, f)$"? Perhaps $f_n \in C_\tau(X_n, X_n)$ for some subspace $X_n$ and we can consider 'approximations' from subsystems? Or $\{f \in C_\tau(X, X) : (X, f) \text{ has property } P\} \subseteq \overline{\{f \in C_\tau(X, X) : (X, f) \text{ has property } Q\}}$ for dynamical properties $P$ and $Q$?

$\endgroup$
  • 1
    $\begingroup$ You may be looking for the notion of Structural Stability: en.wikipedia.org/wiki/Structural_stability $\endgroup$ – Lasse Rempe-Gillen Aug 21 '15 at 13:39
  • 1
    $\begingroup$ Of course, there are situations where you may not have structural stability, but still have "dynamical convergence" along certain sequence. For example, as pointed out below, the Julia set of a quadratic polynomial is not continuous at c=1/4. However, if you approach through the interval $c\in (0,1/4)$, then the Julia sets will converge (and the maps are even topologically conjugate on the Julia set - where they are topologically just the doubling map on the circle). $\endgroup$ – Lasse Rempe-Gillen Aug 21 '15 at 13:42
  • $\begingroup$ This seems like the right concept, although I'm probably asking for something weaker than being 'locally topologically conjugate'. But this is a nice direction to look in. Thank you! $\endgroup$ – Giraffro Aug 21 '15 at 13:55
  • 1
    $\begingroup$ Another situation that comes to mind is the convergence of families of unicritical polynomials, parameterised as $z\mapsto (1+z/d)^d+c$, to the exponential family $z\mapsto e^z+c$. The convergence seems "dynamical" in a certain sense. There is an old preprint by Devaney-Goldberg-Hubbard, now published in two parts with additional co-authors. Also, additional results in a similar spirit by Kriete, Krauskopf and others; see arxiv.org/abs/0910.0743 by my former student Helena Mihaljević-Brandt. However, I am not sure this "dynamical convergence" has ever been fully formalised as a concept. $\endgroup$ – Lasse Rempe-Gillen Aug 24 '15 at 11:11
3
$\begingroup$

In general, dynamical properties of dynamical systems are often quite unstable. The Mandelbrot set makes one nice parameter: the base equation $z \mapsto z^2 + c$ is varies continuously with respect to any reasonable topology on continuous functions. The Mandelbrot set contains large open regions (for instance, the central cardoid, or all the disks touching it). In the interior of these, the dynamics is stable: nearby maps are actually topologically conjugate. But outside of these almost all properties of the dynamics will depend quite sensitively on where you are. For instance, look up plots of Julia sets of the function $z \mapsto z^2 + c$ for $c$ close to $1/4$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.