Skip to Content
Math4202Topology II (Lecture 31)

Math4202 Topology II (Lecture 31)

Algebraic Topology

Before this chapter, we have a big chapter about free groups, be

Seifert-Van Kampen Theorem

The Seifert-Van Kampen Theorem (classical version)

There is an isomorphism between π1(U,x0)π1(V,x0)/N\pi_1(U,x_0)* \pi_1(V,x_0)/N and π1(UV,x0)=π1(X,x0)\pi_1(U\cup V,x_0)=\pi_1(X,x_0).

Note

H is a normal subgroup of GG if gG\forall g\in G, hH,ghg1Hh\in H, ghg^{-1}\in H

Consider the element i1(g)1i2(g)i_1(g)^{-1}i_2(g), where ggπ(UV,x0)g\in g\in \pi_(U\cap V,x_0). Since i1(g)π1(U,x0)i_1(g)\in \pi_1(U,x_0) and i2(g)π1(V,x0)i_2(g)\in \pi_1(V,x_0), then i1(g)1i2(g)π1(U,x0)π1(V,x0)i_1(g)^{-1}i_2(g)\in \pi_1(U,x_0)* \pi_1(V,x_0). (the element is in the free product)

Let NN be the smallest normal subgroup containing all elements of the form i1(g)1i2(g)i_1(g)^{-1}i_2(g) for gπ1(UV,x0)g\in \pi_1(U\cap V,x_0). Such group is constructed by taking the intersection of NαN_\alphas.

  1. π1(U,x0)π1(V,x0)\pi_1(U,x_0)* \pi_1(V,x_0) is a normal subgroup satisfying i1(g)1i2(g)Ni_1(g)^{-1}i_2(g)\in N for all gπ1(UV,x0)g\in \pi_1(U\cap V,x_0).
  2. Take the intersection of all such normal subgroups to get NN.

Proof

Let H=π1(U,x0)π1(V,x0)H=\pi_1(U,x_0)* \pi_1(V,x_0) H=π1(U,x0)π1(V,x0)/NH=\pi_1(U,x_0)* \pi_1(V,x_0)/N (we must quotient by NN; in the raw free product i1(g)1i2(g)i_1(g)^{-1}i_2(g) is in general a non-trivial word, so the relation we want to enforce below only holds after passing to the quotient).

ϕ1:π1(U,x0)H\phi_1:\pi_1(U,x_0)\to H and ϕ2:π1(V,x0)H\phi_2:\pi_1(V,x_0)\to H are defined as the inclusion into the free product followed by the quotient map q:π1(U,x0)π1(V,x0)Hq:\pi_1(U,x_0)* \pi_1(V,x_0)\to H.

We need to verify that ϕ1i1=ϕ2i2\phi_1\circ i_1=\phi_2\circ i_2.

That is i1(g)=i2(g)Hi_1(g)=i_2(g)\in H for all gπ1(UV,x0)g\in \pi_1(U\cap V,x_0). This is equivalent to show that i1(g)1i2(g)=eHi_1(g)^{-1}i_2(g)=e\in H.

Motivation: By construction, NN is the normal closure of {i1(g)1i2(g):gπ1(UV,x0)}\{i_1(g)^{-1}i_2(g):g\in\pi_1(U\cap V,x_0)\}, so every such element lies in kerq\ker q. Hence q(i1(g)1i2(g))=eHq(i_1(g)^{-1}i_2(g))=e_H, giving the commutativity needed to invoke the universal property.

By Seifert-Van Kampen Theorem, Φ:π1(X,x0)H\exists \Phi:\pi_1(X,x_0)\to H.

And ϕ1(U,x0)π1(X,x0)\phi_1(U,x_0)\to \pi_1(X,x_0) by j1j_1 and ϕ2(V,x0)π1(X,x0)\phi_2(V,x_0)\to \pi_1(X,x_0) by j2j_2.

And j:π1(U,x0)π1(V,x0)π1(X,x0)j:\pi_1(U,x_0)* \pi_1(V,x_0)\to \pi_1(X,x_0).

/Track lost/

Incentive to continue: To obtain the classical isomorphism we must exhibit a two-sided inverse to Φ\Phi. The natural candidate is the map j~:Hπ1(X,x0)\tilde j:H\to\pi_1(X,x_0) induced by jj; it is well-defined because j1i1=j2i2j_1\circ i_1=j_2\circ i_2 forces j(N)={e}j(N)=\{e\}. Verifying j~Φ=idπ1(X,x0)\tilde j\circ\Phi=\mathrm{id}_{\pi_1(X,x_0)} (by uniqueness in the universal property) and Φj~=idH\Phi\circ\tilde j=\mathrm{id}_H (by checking on classes coming from π1(U)\pi_1(U) and π1(V)\pi_1(V), which generate HH) completes the proof — this is carried out in Lecture 32.

Corollary of the Seifert-Van Kampen Theorem (classical version)

Assume U,V,UVU,V,U\cap V are path connected, and UVU\cap V is simply connected.

Then π1(UV,x0)=π1(U,x0)π1(V,x0)\pi_1(U\cup V,x_0)=\pi_1(U,x_0)* \pi_1(V,x_0). (NN is the trivial subgroup)

Example

The figure eight shape have intersection with trivial subgroup.

So π1(UV,x0)=π1(U,x0)π1(V,x0)=π1(S1,x0)π1(S1,x0)=ZZ\pi_1(U\cup V,x_0)=\pi_1(U,x_0)* \pi_1(V,x_0)=\pi_1(S^1,x_0)*\pi_1(S^1,x_0)=\mathbb Z*\mathbb Z (Free product of Z\mathbb Z). sttsst\neq ts.

Another Corollary of the Seifert-Van Kampen Theorem (classical version)

Assume U,V,UVU,V,U\cap V are path connected, and VV is simply connected.

Then π1(UV,x0)=π1(U,x0)/N\pi_1(U\cup V,x_0)=\pi_1(U,x_0)/N.

Example

Consider the punctured torus UU union with a spherical cap VV.

π1(V,x0)={e}\pi_1(V,x_0)=\{e\}.

UVS1U\cap V\cong S^1.

And π1(UV,x0)=Z=g\pi_1(U\cap V,x_0)=\mathbb Z=\langle g\rangle.

Consider i1(g)1i_1(g)^{-1} in ZZ\mathbb Z*\mathbb Z.

i1(g)1=st1s1t=1i_1(g)^{-1}=st^{-1}s^{-1}t=1. So in the quotient, st=tsst=ts.

π1(UV,x0)=π1(U,x0)/N=Z×Z\pi_1(U\cup V,x_0)=\pi_1(U,x_0)/N=\mathbb Z\times \mathbb Z.

Last updated on