Pointed shape and global attractors for metrizable spaces (Q617726): Difference between revisions

The authors deal with dynamical (resp. semidynamical) systems on a metric space \(X, \;(X, \mathbb R, \pi)\)), resp. \((X, \mathbb R_+, \pi),\;\pi: X \times \mathbb R_+ \to X\) and different types of attractors \(K\): global, strong global and local attractors, which are assumed to be compact, \(i: (K,x_0 ) \subset (X,x_0)\). Such a (semidynamical) system can be described by the mapping \(\{ S_t: X \to X \mid t\in \mathbb R_+\}\), \(S_t(x)=\pi(x,t)\). A local attractor is one which behaves as an attractor only for some suitable neighborhood \(W\) of \(K\), where the maximal such \(W\) is called the region of attraction \(A(K)\) of \(K\) One of the central questions in this paper is to develop conditions under which \(i\) is a shape equivalence and the key theorem, from which all other results follow is: Let \(X\) be a metric \(ANR\) and \(\{S_t\}\) a semidynamical system with global attractor \(K\). Then for any basepoint the inclusion \(i\) is a pointed shape equivalence, a pointed weak and a pointed strong shape equivalence. Here a notational awkwardness appears because the authors use the terminolgy of K. Borsuk's book, who introduces W-shape and S-shape, while in modern shape theory strong shape has a totally different meaning. Borsuk started from the observation, that for non-compact spaces both notions (W-shape and S-shape) are not equivalent anymore. On the other hand ist would be interesting to know if this assertion also holds for this presently used concept of strong shape. In this case one would immediately get isomorphisms for all kinds of homology, cohomology and homotopy groups. Among the corollaries of this theorem is the assertion that for a local attractor \(K \in ANR\) the inclusion \(i: (K,x_0)\to (A(K),x_0)\) is a shape equivalence as before \(i: (K,x_0)\to (X,x_0)\).