15.01.2013 Views

Simplicial Structures in Topology

Simplicial Structures in Topology

Simplicial Structures in Topology

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.

50 II <strong>Simplicial</strong> Complexes<br />

Let us rewrite the <strong>in</strong>dices of the elements of s(p) and s(q) to have the follow<strong>in</strong>g<br />

common elements:<br />

xr = y0,xr+1 = y1,...,xn = yn−r.<br />

Notice that the set s(p)∪s(q) has exactly m+r+1 common elements. Now consider<br />

the elements<br />

⎧<br />

⎨ xi, 0 ≤ i ≤ r − 1<br />

zi = xi = yi−r, r ≤ i ≤ n<br />

⎩<br />

yi−r, n + 1 ≤ i ≤ m + r<br />

together with the real numbers<br />

⎧<br />

⎨ −αi, 0 ≤ i ≤ r − 1<br />

γi = −αi + βi−r, r ≤ i ≤ n<br />

⎩<br />

βi−r, n + 1 ≤ i ≤ m + r .<br />

Notice that γi < 0fori = 0,...,r − 1andγi > 0fori = n + 1,...,m + r, because<br />

αi > 0foreveryi = 0,1,...,n and βi > 0fori = 1,...,m. Let us order the numbers<br />

γi <strong>in</strong> such a way that γ0 ≤ γ1 ≤ ... ≤ γm+r (if necessary, we make a permutation of<br />

the <strong>in</strong>dices). Let l be the largest <strong>in</strong>dex for which γl < 0 (because of the assumptions<br />

we made, such a set of <strong>in</strong>dices cannot be empty - thus l exists - nor can it be the set of<br />

all <strong>in</strong>dices - thus r ≤ l ≤ n); moreover, the vertices (viewed as functions) z0,z1,...,zl<br />

are summands of p (the numbers γi are negative), while the vertices zl+1,...,zm+r<br />

are part of q (the correspond<strong>in</strong>g numbers γi are non-negative). At this po<strong>in</strong>t, take<br />

λ = ∑ l i=0 γi < 0 and the two f<strong>in</strong>ite successions of real positive numbers<br />

The elements<br />

are <strong>in</strong> |K| because<br />

{ γ0<br />

λ ,...,γl<br />

γl+1<br />

} and {<br />

λ −λ ,...,γm+r<br />

−λ }.<br />

p ′ =<br />

l<br />

γi<br />

∑<br />

i=0<br />

l<br />

γi<br />

∑<br />

i=0<br />

λ zi and q ′ =<br />

λ =<br />

m+r<br />

∑<br />

i=l+1<br />

γi<br />

m+r<br />

∑<br />

i=l+1<br />

= 1;<br />

−λ<br />

γi<br />

−λ zi<br />

from what we proved above, it follows that s(p ′ ) ⊂ s(p) and s(q ′ ) ⊂ s(q). But<br />

s(p ′ ) ∩ s(q ′ )=/0 andso,byCase 1,<br />

The equalities<br />

d(| f |(p ′ ),| f |(q ′ )) ≤ � 2(l + 1)d(p ′ ,q ′ ).<br />

d(p ′ ,q ′ )= 1<br />

−λ d(p,q),<br />

d(| f |(p ′ ),| f |(q ′ )) = 1<br />

d(| f |(p),| f |(q)),<br />

−λ<br />

and the fact that � 2(l + 1) ≤ � 2(n + 1) allow us to conclude that

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

Saved successfully!

Ooh no, something went wrong!