Abstract. We study dynamics of continuous maps on compact metrizable spaces containing a free interval (i.e., an open subset homeomorphic to an open interval). A special attention is paid to relationships between topological transitivity, weak and strong topological mixing, dense periodicity and topological entropy as well as to the topological structure of minimal sets. In particular, a trichotomy for minimal sets and a dichotomy for transitive maps are proved.
IntroductionOne-dimensional dynamics became an object of wide interest in the middle of 1970's, some 10 years after Sharkovsky's theorem, when chaotic phenomena were discovered by Li and Yorke in dynamics of interval maps. As general references one can recommend the monographs [CE80], [BC92], [ALM00] and [dMvS93] (several motivations for studying one-dimensional dynamical systems are discussed in the introduction of [dMvS93]). The interval and circle dynamics are well understood. To extend/generalize the results to graph maps is sometimes quite easy, sometimes extremely difficult (for instance the characterization of the set of periods for graph maps is known only in very special cases). A good, more than 50 pages long survey of some topics in the dynamics of graph maps can be found in an appendix in [ALM00] (the second edition). Also the paper [Bl84] and the paper [Bl86] with its two continuations under the same title are a must for everybody who wishes to study the dynamics of graph maps.When working in one-dimensional topological dynamics it is natural to try to extend results from the interval to more general spaces. One usually extends them (or slight modifications of them) first to the circle/trees and then to general graphs (then perhaps also to special kinds of dendrites or to other one-dimensional continua; in the case of general dendrites often a counterexample can be found). The present paper suggests that sometimes another approach can be more fruitful -we show that some important facts from the topological dynamics on the interval/circle work on much more general spaces than graphs, namely on spaces containing an open part looking like an interval (we will call it a free interval ). It seems that the first result indicating that the presence of a free interval might have important dynamical consequences for the whole space was obtained as early as 1988 in [Ka88], for two other results see [AKLS99] and [HKO11]. However, they were isolated in a sense and apparently have not attracted much attention; a systematic study of the influence of a free interval on dynamical properties of a space has not been done yet. Based on the main results of the present paper, we believe that namely the class of spaces with a free interval is a natural candidate for possible extension of classical results of one-dimensional topological dynamics. Of course, not all of them can be carried over from the interval/circle to spaces with a free interval (and then trees/graphs naturally enter the scene as candidates for possible extension). However, 2010 Mathematics Subj...