next up previous contents
Next: Chain Recurrence Up: Asymptotic Behavior and Recurrence Previous: Invariant Sets

Limit Sets: tex2html_wrap_inline11245 , tex2html_wrap_inline11247 , and Nonwandering

 

We begin by introducing the tex2html_wrap_inline11247 -limit set , which starts us down the road toward defining an attractor. Let p be a point of a map tex2html_wrap_inline14784 of the manifold M. Then the tex2html_wrap_inline11247 -limit set of p is

displaymath4464

Conversely, by going backwards in time we get the tex2html_wrap_inline11245 -limit set of p ,

displaymath4470

These limit sets are the closure of the ends of the orbits. For example, if p is a periodic point, then tex2html_wrap_inline14800 . It is not difficult to show that tex2html_wrap_inline14802 is a closed subset of M and that it is invariant under f, i.e., tex2html_wrap_inline14808 . Finally, a point tex2html_wrap_inline14810 is called recurrent  if it is part of the tex2html_wrap_inline11247 -limit set, i.e., tex2html_wrap_inline14814 .

The forward limit set , tex2html_wrap_inline14816 , is defined as the union of all tex2html_wrap_inline11247 -limit sets; likewise the backward limit set , tex2html_wrap_inline14820 , is defined as the union of all tex2html_wrap_inline11245 -limit sets:

displaymath4481

The forward and backward limit sets are not necessarily closed. This observation motivates yet another useful notion of recurrence, the nonwandering set. A point tex2html_wrap_inline14810 wanders  if there exist a neighborhood U of p and an m ;SPMgt; 0 such that tex2html_wrap_inline14832 for all n ;SPMgt; m. A nonwandering point  is one that does not wander. This brings us to the nonwandering set , tex2html_wrap_inline13998 , which is a closed, invariant (under f) subset of M:

displaymath4493

The dynamical decomposition of a set into its wandering and nonwandering parts separates, in mathematical terms, a dynamical system into its transient behavior --the wandering set--and long-term or asymptotic behavior --the nonwandering set.

The tex2html_wrap_inline11247 -limit set and the nonwandering set do not address the question of the stability of an asymptotic motion. To get to the idea of an attractor, we begin with the idea of an attracting set. A closed invariant set tex2html_wrap_inline14844 is an attracting set  if there is some neighborhood U of A such that for all tex2html_wrap_inline14850 and all tex2html_wrap_inline12912 ,

displaymath4500

Moreover, the domain  or basin of attraction  of A is given by

displaymath4507

The attracting set can consist of a collection of different sets that are dynamically disconnected. For instance, a single attracting set could consist of two separate periodic orbits. To overcome this last difficulty, we will say that an attractor  is an attracting set that contains a dense orbit. Conversely, a repeller  is defined as a repelling set with a dense orbit.

Although this is a reasonable definition mathematically, we will see that this is not the most useful definition of an attractor for physical applications or numerical simulations. In these circumstances it will turn out that a more useful (albeit mathematically naive) definition of an attractor or repeller is the closure of a certain collection of periodic orbits. The proper definition of an attractor is yet another hot spot in the creative tension between the rigor demanded by a mathematician and the utility required by a physicist.


next up previous contents
Next: Chain Recurrence Up: Asymptotic Behavior and Recurrence Previous: Invariant Sets

Nicholas B. Tufillaro
Mon Mar 3 01:58:02 PST 1997