23.11.2012 Views

Discrete Holomorphic Local Dynamical Systems

Discrete Holomorphic Local Dynamical Systems

Discrete Holomorphic Local Dynamical Systems

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

320 Dierk Schleicher<br />

8. if f has a Baker wandering domain, then A( f ) and I( f ) are connected, and each<br />

component of A( f ) ∩ J( f ) is bounded; otherwise, each component of A( f ) ∩<br />

J( f ) is unbounded;<br />

9. every component of A( f ) is unbounded;<br />

10. wandering domains may escape slow enough for L( f ) and fast enough for A( f ).<br />

Item (1) comes out of Eremenko’s proof that I( f ) is non-empty (compare<br />

Theorem 1.8). In (2), invariance is built into the definition, and the inclusion<br />

comes from [BH99]. For Statement (3), note that ∂A( f ), ∂L( f ), and∂Z( f ) are all<br />

closed invariant sets, so they contain J( f ) by Theorem 1.7, and the claim reduces to<br />

the fact that any Fatou component that intersects A( f ), L( f ),orZ( f ) is contained in<br />

these sets. This follows from Lemma 4.5 and also implies (4) (see[RS00]). Statement<br />

(5) is from [RS00, BH99]. Statement (6) is due to Baker (see Lemma 4.5).<br />

Statements (7) and(9) are from [RS05a], while (8) is unpublished recent work by<br />

Rippon and Stallard. Wandering domains in L( f ) were given in Examples 2.9 and<br />

2.10, while Baker wandering domains are in A( f ); thisis(10).<br />

Recent work by Rippon and Stallard is on “spider’s webs”: these are subsets of<br />

I( f ) for which every complementary component is connected (such as the orbit of a<br />

Baker wandering domain, together with certain parts of I( f ) connecting the various<br />

wandering domains). Rippon and Stallard propose the idea that this feature may be<br />

no less common and prototypical for the dynamics of entire functions as Cantor<br />

Bouquets are nowadays often considered to be.<br />

Theorem 4.8 (Slow Escape Possible).<br />

For every real sequence Kn → ∞,thereisaz∈ I( f ) ∩ J( f ) so that | f ◦n (z)| < Kn for<br />

all sufficiently large n.<br />

This result is also due to Rippon and Stallard [RS].<br />

It had been observed by Fatou that I( f ) often contains curves to ∞. Thiswas<br />

shown for exponential maps in [DGH], and for more general entire functions having<br />

logarithmic tracts satisfying certain geometric conditions in [DT86]. This leads to<br />

the following.<br />

Definition 4.9 (Dynamic Ray).<br />

A dynamic ray tail is an injective curve γ : (τ,∞) → I( f ) so that<br />

• f ◦n (γ(t)) → ∞ as t → ∞ for every n ≥ 0, and<br />

• f ◦n (γ(t)) → ∞ as n → ∞ uniformly in t.<br />

A dynamic ray is a maximal injective curve γ : (0,∞) → I( f ) so that for every τ > 0,<br />

the restriction γ| (τ,∞) is a dynamic ray tail. An endpoint of a dynamic ray γ is a point<br />

a with a = limt↘τ γ(t).<br />

Remark 4.10. Dynamic rays are sometimes called “hairs”; we prefer the term “ray”<br />

in order to stress the similarity to the polynomial case.<br />

Eremenko asked whether every point in I( f ) can be connected to ∞ by a curve in<br />

I( f ) (compare Conjecture 4.2). This conjecture has been confirmed for exponential

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!