5
$\begingroup$

Given a continuous group action $G \times X \rightarrow X$ on a topological space $X$, is there a standard term for the subsets $K \subset X$ for which

  1. Every open neighborhood of $K$ intersects every group orbit;
  2. $K$ is minimal with respect to the previous property ?

Examples: points of $X$ if the action is transitive or ergodic, a meridian in $S^2$ if the action is rotating the sphere around the poles, the limit cycle plus the fixed point in the Van der Pol oscillator.

Edit. Maybe "minimality" is not the right term. In the examples above, you can replace the meridian for a dense set of points on the meridian and the same with the limit cycle in the Van der Pol oscillator (i.e., the examples are not really examples or, rather, they are examples of the phenomenon I wish to describe, but have failed to do).

$\endgroup$
5
  • 2
    $\begingroup$ The 'any' in 1. means 'every', right? $\endgroup$
    – LSpice
    Commented Apr 29, 2021 at 14:16
  • $\begingroup$ If $X$ is good enough, will not minimality force that $K$ contains a single point from each orbit? $\endgroup$ Commented Apr 29, 2021 at 14:48
  • $\begingroup$ Not necessarily. If you have an ergodic flow then every point will satisfy the condition, but there is more than just one orbit. $\endgroup$ Commented Apr 29, 2021 at 16:10
  • 5
    $\begingroup$ In the study of single homeomorphisms on a Cantor set, that is ${\bf Z}$-actions, a closed set satisfying property 1 is known as a hitting set. So minimal hitting set. It is possible to construct a minimal hitting set that hits every orbit at most once; this is called a basic set (although this is not completely standard). $\endgroup$ Commented Apr 29, 2021 at 16:47
  • $\begingroup$ @DavidHandelman, I checked and the term is quite used in computer science (look for "hitting set problem"), but the meaning is a bit different. In their terminology I am looking for sets such that every one of their open neighborhoods is a hitting set for the set of group orbits (plus being minimal). $\endgroup$ Commented Apr 30, 2021 at 18:18

0

You must log in to answer this question.

Browse other questions tagged .