208 CHAPTER 10. BROUWER FIXED POINT THEOREM Rn∗

Definition 10.0.2 If S is an n simplex. Then it is triangulated if it is the union of smllersub-simplices, the triangulation, such that if S1,S2 are two simplices in the triangulation,with

S1 ≡[z1

0, · · · ,z1m], S2 ≡

[z2

0, · · · ,z2p]

thenS1∩S2 =

[xk0 , · · · ,xkr

]where

[xk0 , · · · ,xkr

]is in the triangulation and{

xk0 , · · · ,xkr

}={

z10, · · · ,z1

m}∩{

z20, · · · ,z2

p}

or else the two simplices do not intersect.

The following proposition is geometrically fairly clear. It will be used without commentwhenever needed in the following argument about triangulations.

Proposition 10.0.3 Say [x1, · · · ,xr] , [x̂1, · · · , x̂r] , [z1, · · · ,zr] are all r−1 simplices and

[x1, · · · ,xr] , [x̂1, · · · , x̂r]⊆ [z1, · · · ,zr]

and [z1, · · · ,zr,b] is an r+1 simplex and

[y1, · · · ,ys] = [x1, · · · ,xr]∩ [x̂1, · · · , x̂r] (10.0.1)

where{y1, · · · ,ys}= {x1, · · · ,xr}∩{x̂1, · · · , x̂r} (10.0.2)

Then[x1, · · · ,xr,b]∩ [x̂1, · · · , x̂r,b] = [y1, · · · ,ys,b] (10.0.3)

Proof: If you have ∑si=1 tiyi + ts+1b in the right side, the ti summing to 1 and nonneg-

ative, then it is obviously in both of the two simplices on the left because of 10.0.2. Thus[x1, · · · ,xr,b]∩ [x̂1, · · · , x̂r,b]⊇ [y1, · · · ,ys,b].

Now suppose xk = ∑rj=1 tk

j z j, x̂k = ∑rj=1 t̂k

j z j, as usual, the scalars adding to 1 and non-negative.

Consider something in both of the simplices on the left in 10.0.3. Is it in the right? Theelement on the left is of the form

r

∑α=1

sα xα + sr+1b =r

∑α=1

ŝα x̂α + ŝr+1b

where the sα , are nonnegative and sum to one, similarly for ŝα . Thus

r

∑j=1

r

∑α=1

sα tαj z j + sr+1b =

r

∑α=1

r

∑j=1

ŝα t̂αj z j + ŝr+1b (10.0.4)

Now observe that

∑j∑α

sα tαj + sr+1 = ∑

α

∑j

sα tαj + sr+1 = ∑

α

sα + sr+1 = 1.

208 CHAPTER 10. BROUWER FIXED POINT THEOREM R”™*Definition 10.0.2 If S is an n simplex. Then it is triangulated if it is the union of smllersub-simplices, the triangulation, such that if S,,S_ are two simplices in the triangulation,withS,= [z),°°- Zn]. Sy = [z9,°-- Z|thenSy OS. = [Xtgsto Xx, |where [Xx ee .Xx,| is in the triangulation and1 1 2 2{Xkgs0 Xx, } - {Zo Zin} V{Z00° Zp}or else the two simplices do not intersect.The following proposition is geometrically fairly clear. It will be used without commentwhenever needed in the following argument about triangulations.Proposition 10.0.3 Say [x,,--- ,x,],[%1,--- ,&,],[Z1,--+ ,Z,] are all r—1 simplices and[X1,°-° »X],[Ki,-°° | Cc [Z1,-°° ,Zy]and |21,+-+ ,Zy,b] isan r+ 1 simplex and[Yi,-°° Ys] = [X1,°°° »X;| M1 [&1,-°- x] (10.0.1)where{yi,--: sys} = {x1,-°- Xx} {R10 ,x,} (10.0.2)Then[X1,°°° »X;,b] a [Ki,-°° »X,,b] = [Yis--- »Ys,b] (10.0.3)Proof: If you have Y*_, tiv; +1;41b in the right side, the ¢; summing to | and nonneg-ative, then it is obviously in both of the two simplices on the left because of 10.0.2. Thus[X1,+°° »X;,b] a [Ri ,-°- »%,,b] 2 [Yi,0*° 5Ys,b].Now suppose x, = Li- 1 taj, x, = Y= fiz j, aS usual, the scalars adding to 1 and non-negative.Consider something in both of the simplices on the left in 10.0.3. Is it in the right? Theelement on the left is of the formr ry SaXq t+ S-41b= y SoXa +841ba=1 a=1where the sg, are nonnegative and sum to one, similarly for Sg. Thusr r rY sat%zj+s-ib= VY Saffaj+S,.1b (10.0.4)j=la=l a=1 j=1M:-Now observe thatVY sat? +541 = VY sat? +541 = Ysa +S541= 1.j @ a j a