Let X be the subspace of the unit cube I3 consisting of the union of the 6 faces and the 4 internal diagonals. Compute π1(X).
Solution:
2
Let X be an arbitrary topological space, and compute π1(ΣX).
Solution:
Write ΣX=U∪V where U=ΣX−(X×[0,1/2]) and U=ΣX−X×[1/2,1]). Then U∩V=X×{1/2}≅X, so π1(U∩V)=π1(X).
But both U and V can be identified by the cone on X, given by CX=X×IX×1, by just rescaling the interval with the maps:
iU:U→CX where (x,s)↦(x,2s−1) (The second component just maps [1/2,1]→[0,1]. )
iV:V→CX where (x,s)↦(x,2s). (The second component just maps [0,1/2]→[0,1])
But CX is contractible by the homotopy H:CX×I→CX where H((c,s),t)=(c,s(1−t)).
So \pi_1(U) = \pi_1(V) = 0.
By Van Kampen, we have \pi_1(X) = 0 \ast_{\pi_1(X)} 0 = 0.
3
Let X = S^1 \times S^1 and A\subset X be a subspace with A \cong S^1 \vee S^1. Show that there is no retraction from X to A.
Solution:
We have \pi_1(S^1 \times S^1) = \pi_1(S^1) \times\pi_1(S^1) since S^1 is path-connected (by a lemma from the problem sets), and this equals {\mathbf{Z}}\times{\mathbf{Z}}.
We also have \pi_1(S^1 \vee S^1) = \pi_1(S^1) \ast_{\left\{{pt}\right\}} \pi_1(S^1), which by Van-Kampen is {\mathbf{Z}}\ast {\mathbf{Z}}.
Suppose X retracts onto A, we can then look at the inclusion \iota: A \hookrightarrow X. The induced homomorphism \iota_*: \pi_1(A) \hookrightarrow\pi_1(X) is then also injective, so we’ve produced an injection from f: {\mathbf{Z}}\ast {\mathbf{Z}}\hookrightarrow{\mathbf{Z}}\times{\mathbf{Z}}.
This is a contradiction, because no such injection can exists. In particular, the commutator [a,b] is nontrivial in the source. But f(aba^{-1}b^{-1}) = f(a)f(b)f(a)^{-1}f(b)^{-1} since f is a homomorphism, but since the target is a commutative group, this has to equal f(a)f(a)^{-1} f(b)f(b)^{-1} = e. So there is a non-trivial element in the kernel of f, and f can not be injective - a contradiction.
4
Show that for every map f: S^2 \to S^1, there is a point x\in S^2 such that f(x) = f(-x).
Solution:
Suppose towards a contradiction that f does not possess this property, so there is no x\in S^2 such that f(x) = f(-x).
Then define g: S^2 \to S^1 by g(x) = {f(x) - f(-x)}; by assumption, this is a nontrivial map, i.e. g(x) \neq 0 for any x\in S^2.
In particular, -g(-x) = -{(f(-x) - f(x))} = {f(x) - f(-x)} = g(x), so -g(x) = g(-x) and thus g commutes with the antipodal map \alpha: S^2 \to S^2.
This means g is constant on the fibers of the quotient map p: S^2 \to{\mathbf{RP}}2, and thus descends to a well defined map \tilde g: {\mathbf{RP}}2 \to S^1, and since S^1 \cong {\mathbf{RP}}1, we can identify this with a map \tilde g: {\mathbf{RP}}2 \to{\mathbf{RP}}1 which thus induces a homomorphism \tilde g_*: \pi_1({\mathbf{RP}}2) \to \pi_1({\mathbf{RP}}1).
Since g was nontrivial, \tilde g is nontrivial, and by functoriality of \pi_1, \tilde g_* is nontrivial.
But \pi_1({\mathbf{RP}}2) = {\mathbf{Z}}_2 and \pi_1({\mathbf{RP}}1) = {\mathbf{Z}}, and \tilde g_*: {\mathbf{Z}}^2 \to{\mathbf{Z}} can only be the trivial homomorphism - a contradiction.
Alternate Solution
Use covering space {\mathbf{R}}\twoheadrightarrow S^1?
5
How many path-connected 2-fold covering spaces does S^1 \vee {\mathbf{RP}}2 have? What are the total spaces?
Solution:
First note that \pi_1(X) = \pi_1(S^1) \ast_{{\operatorname{pt}}} \pi_1({\mathbf{RP}}2) by Van-Kampen, and this is equal to {\mathbf{Z}}\ast {\mathbf{Z}}_2.
6
Let G = <a, b> and H \leq G where H = <aba^{-1}b^{-1},~ a^2ba^{-2}b^{-1},~ a^{-1}bab^{-1},~ aba^{-2}b^{-1}a>. To what well-known group is H isomorphic?
Solution: