03.12.2012 Views

PERCOLATION AND DISORDERED SYSTEMS Geoffrey GRIMMETT

PERCOLATION AND DISORDERED SYSTEMS Geoffrey GRIMMETT

PERCOLATION AND DISORDERED SYSTEMS Geoffrey GRIMMETT

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Percolation and Disordered Systems 199<br />

B(n) 010101<br />

R<br />

00 11 00 11 01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

00 11 00 11 00 11 00 11 00 11 00 11 01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

00 11<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

010101010101<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

01<br />

Fig. 7.4. An illustration of Lemma 7.17. The hatched regions are copies of B(m)<br />

all of whose edges are p-open. The central box B(m) lies within some (dotted) region<br />

R. Some vertex in R is joined by a path to some vertex in ∂B(n), which is in turn<br />

connected to a seed lying on the surface of B(n).<br />

and the claim of the lemma follows by (7.13). �<br />

Having constructed open paths from B(m) to K(m, n), we shall need to<br />

repeat the construction, beginning instead at an appropriate seed in K(m, n).<br />

This is problematic, since we have discovered a mixture of information, some<br />

of it negative, about the immediate environs of such seeds. In order to overcome<br />

the effect of such negative information, we shall work at edge-density<br />

p + δ rather than p. In preparation, let (X(e) : e ∈ E d ) be independent random<br />

variables having the uniform distribution on [0, 1], and let ηp(e) be the<br />

indicator function that X(e) < p; recall Section 2.3. We say that e is p-open<br />

if X(e) < p and p-closed otherwise, and we denote by P the appropriate<br />

probability measure.<br />

For any subset V of Z 3 , we define the exterior boundary ∆V and exterior<br />

edge-boundary ∆eV by<br />

∆V = {x ∈ Z 3 : x �∈ V, x ∼ y for some y ∈ V },<br />

∆eV = {〈x, y〉 : x ∈ V, y ∈ ∆V, x ∼ y}.<br />

We write EV for the set of all edges of L 3 joining pairs of vertices in V .<br />

We shall make repeated use of the following lemma 5 , which is illustrated<br />

in Figure 7.4.<br />

5 This lemma is basically Lemma 6 of [164], the difference being that [164] was addressed<br />

at site percolation. R. Meester and J. Steif have kindly pointed out that, in the case of<br />

site percolation, a slightly more general lemma is required than that presented in [164].<br />

The following remarks are directed at the necessary changes to Lemma 6 of [164], and they<br />

use the notation of [164]. The proof of the more general lemma is similar to that of the<br />

original version. The domain of β is replaced by a general subset S of B(n) \ T(n), and<br />

G is the event {there exists a path in B(n) from S to K(m, n), this path being p-open in<br />

B(n) \ S and (β(u) + δ)-open at its unique vertex u ∈ S}. In applying the lemma just<br />

after (4.10) of [164], we take S = ∆C2 ∩ B(n) (and similarly later).

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

Saved successfully!

Ooh no, something went wrong!