Skip to Content
Math4202Topology II (Lecture 23)

Math4202 Topology II (Lecture 23)

Algebraic Topology

Fundamental Theorem of Algebra

Recall the lemma g:S1R{0}g:S^1\to \mathbb{R}-\{0\} is not nulhomotopic.

g=hkg=h\circ k where k:S1S1k:S^1\to S^1 by zznz\mapsto z^n, k:π1(S1)π1(S1)k_*:\pi_1(S^1)\to \pi_1(S^1) is injective. (consider the multiplication of integer is injective)

and h:S1R{0}h:S^1\to \mathbb{R}-\{0\} where zzz\mapsto z. h:π1(S1)π1(R{0})h_*:\pi_1(S^1)\to \pi_1(\mathbb{R}-\{0\}) is injective. (inclusion map is injective)

Therefore g:π1(S1)π1(R{0})g_*:\pi_1(S^1)\to \pi_1(\mathbb{R}-\{0\}) is injective, therefore gg cannot be nulhomotopic. (nulhomotopic cannot be injective)

Theorem

Consider xn+an1xn1+an2xn2++a0=0x^n+a_{n-1}x^{n-1}+a_{n-2}x^{n-2}+\cdots+a_0=0 of degree >0>0.

Proof: part 1

Step 1: if an1+an2++a0<1|a_{n-1}|+|a_{n-2}|+\cdots+|a_0|<1, then xn+an1xn1+an2xn2++a0=0x^n+a_{n-1}x^{n-1}+a_{n-2}x^{n-2}+\cdots+a_0=0 has a root in the unit disk B2B^2.

We proceed by contradiction, suppose there is no root in B2B^2.

Consider f(x)=xn+an1xn1+an2xn2++a0f(x)=x^n+a_{n-1}x^{n-1}+a_{n-2}x^{n-2}+\cdots+a_0.

fB2f|_{B^2} is a continuous map from B2R2{0}B^2\to \mathbb{R}^2-\{0\}.

fS1=B2:S1R{0}f|_{S^1=\partial B^2}:S^1\to \mathbb{R}-\{0\} is nulhomotopic.

Recall that: Any map g:S1Yg:S^1\to Y is nulhomotopic whenever it extends to a continuous map G:B2YG:B^2\to Y.

Construct a homotopy between fS1f|_{S^1} and gg

H(x,t):S1R{0}xn+t(an1xn1+an2xn2++a0)H(x,t):S^1\to \mathbb{R}-\{0\}\quad x^n+t(a_{n-1}x^{n-1}+a_{n-2}x^{n-2}+\cdots+a_0)

Observer on S1S^1, xn=1,nN\|x^n\|=1,\forall n\in \mathbb{N}.

t(an1xn1+an2xn2++a0)=tan1xn1+an2xn2++a01(an1xn1+an2xn2++a0)=an1+an2++a0<1\begin{aligned} \|t(a_{n-1}x^{n-1}+a_{n-2}x^{n-2}+\cdots+a_0)\|&=t\|a_{n-1}x^{n-1}+a_{n-2}x^{n-2}+\cdots+a_0\|\\ &\leq 1(\|a_{n-1}x^{n-1}\|+\|a_{n-2}x^{n-2}\|+\cdots+\|a_0\|)\\ &=\|a_{n-1}\|+\|a_{n-2}\|+\cdots+\|a_0\|\\ &<1 \end{aligned}

Therefore H(s,t)>00<t<1H(s,t)>0\forall 0<t<1. is a well-defined homotopy between fS1f|_{S^1} and gg.

Therefore f=gf_*=g_* is injective, ff is not nulhomotopic. This contradicts our previous assumption that ff is nulhomotopic.

Therefore ff must have a root in B2B^2.

Proof: part 2

If an1+an2++a0<R\|a_{n-1}\|+\|a_{n-2}\|+\cdots+\|a_0\|< R has a root in the disk BR2B^2_R. (and R1R\geq 1, otherwise follows part 1)

Consider f~(x)=f(Rx)\tilde{f}(x)=f(Rx).

f~(x)=f(Rx)=(Rx)n+an1(Rx)n1+an2(Rx)n2++a0=Rn(xn+an1Rxn1+an2R2xn2++a0Rn)\begin{aligned} \tilde{f}(x) =f(Rx)&=(Rx)^n+a_{n-1}(Rx)^{n-1}+a_{n-2}(Rx)^{n-2}+\cdots+a_0\\ &=R^n\left(x^n+\frac{a_{n-1}}{R}x^{n-1}+\frac{a_{n-2}}{R^2}x^{n-2}+\cdots+\frac{a_0}{R^n}\right) \end{aligned} an1R+an2R2++a0Rn=1Ran1+1R2an2++1Rna0<1R(an1+an2++a0)<1R<1\begin{aligned} \left\|\frac{a_{n-1}}{R}\right\|+\left\|\frac{a_{n-2}}{R^2}\right\|+\cdots+\left\|\frac{a_0}{R^n}\right\|&=\frac{1}{R}\|a_{n-1}\|+\frac{1}{R^2}\|a_{n-2}\|+\cdots+\frac{1}{R^n}\|a_0\|\\ &<\frac{1}{R}\left(\|a_{n-1}\|+\|a_{n-2}\|+\cdots+\|a_0\|\right)\\ &<\frac{1}{R}<1 \end{aligned}

By Step 1, f~\tilde{f} must have a root z0z_0 inside the unit disk.

f(Rz0)=f~(z0)=0f(Rz_0)=\tilde{f}(z_0)=0.

So ff has a root Rz0Rz_0 in BR2B^2_R.

Deformation Retracts and Homotopy Type

Recall previous section, h:S1R{0}h:S^1\to \mathbb{R}-\{0\} gives h:π1(S1,1)π1(R{0},0)h_*:\pi_1(S^1,1)\to \pi_1(\mathbb{R}-\{0\},0) is injective.

For this section, we will show that hh_* is an isomorphism.

Lemma for equality of homomorphism

Let h,k:(X,x0)(Y,y0)h,k: (X,x_0)\to (Y,y_0) be continuous maps. If hh and kk are homotopic, and if the image of x0x_0 under the homotopy remains y0y_0. The homomorphism hh_* and kk_* from π1(X,x0)\pi_1(X,x_0) to π1(Y,y0)\pi_1(Y,y_0) are equal.

Proof

Let H:X×IYH:X\times I\to Y be a homotopy from hh to kk such that

H(x,0)=h(x),H(x,1)=k(x),H(x0,t)=y0 for all tI.H(x,0)=h(x), \qquad H(x,1)=k(x), \qquad H(x_0,t)=y_0 \text{ for all } t\in I.

To show h=kh_*=k_*, let [f]π1(X,x0)[f]\in \pi_1(X,x_0) be arbitrary, where f:IXf:I\to X is a loop based at x0x_0, so f(0)=f(1)=x0f(0)=f(1)=x_0.

Define

F:I×IY,F(s,t)=H(f(s),t).F:I\times I\to Y,\qquad F(s,t)=H(f(s),t).

Since HH and ff are continuous, FF is continuous. For each fixed tIt\in I, the map

sF(s,t)=H(f(s),t)s\mapsto F(s,t)=H(f(s),t)

is a loop based at y0y_0, because

F(0,t)=H(f(0),t)=H(x0,t)=y0andF(1,t)=H(f(1),t)=H(x0,t)=y0.F(0,t)=H(f(0),t)=H(x_0,t)=y_0 \quad\text{and}\quad F(1,t)=H(f(1),t)=H(x_0,t)=y_0.

Thus FF is a based homotopy between the loops hfh\circ f and kfk\circ f, since

F(s,0)=H(f(s),0)=h(f(s))=(hf)(s),F(s,0)=H(f(s),0)=h(f(s))=(h\circ f)(s),

and

F(s,1)=H(f(s),1)=k(f(s))=(kf)(s).F(s,1)=H(f(s),1)=k(f(s))=(k\circ f)(s).

Therefore hfh\circ f and kfk\circ f represent the same element of π1(Y,y0)\pi_1(Y,y_0), so

[hf]=[kf].[h\circ f]=[k\circ f].

Hence

h([f])=[hf]=[kf]=k([f]).h_*([f])=[h\circ f]=[k\circ f]=k_*([f]).

Since [f][f] was arbitrary, it follows that h=kh_*=k_*.

Last updated on