1

[Hatcher, Problem 19, Section 1.2]

Show that the subspace $X$ of $\mathbb{R}^3$ that is the union of the spheres of radius $\frac1n$ with center $(\frac{1}{n},0,0)$ for $n=1,2,3 \dots$ is simply connected.

I understand that the solution directly follows from an application of Van Kampen Theorem but will this work?

any path in $X$ is a compact subset of $X$, so it cannot lie in infinitely many spheres(I don't know how to reason this)

Now any loop starting at $(0,0,0)$ can be written as composition of loops $\gamma_i$ where $\gamma_i$ completely lies in the sphere of radius $\frac{1}{i}$. Now in each of the sphere the loop $\gamma_i$ is homotopic to the constant path at $(0,0,0)$ and thus their join, that is our orignal path will also be homotopic to the constant path at $(0,0,0)$ and this proves our result.

0-Is the fact that each path on $X$ will lie in a finite number of spheres?

1-Is it necessary to show that the loop $\gamma$ can be written as join of finitely many paths?

2-If so, how can I formalize this arguement?

Paul Frost
  • 87,968
  • 3
    If I were you, I'd check the hypotheses of Van Kampen's Theorem very carefully before concluding that it applies in this situation. – Lee Mosher Dec 22 '22 at 23:31
  • 2
    And as for whether a path can lie in infinitely many spheres, how about the path $p : [0,1] \to X$ for which $p(0)=(0,0,0)$, and for $n \ge 1$ the subpath $p \mid [1/(n+1),1/n]$ is a reparameterization of the path whose restriction to $[0,1]$ goes once around the circle where the $n^{\text{th}}$ sphere intersects the $x,y$ plane. – Lee Mosher Dec 22 '22 at 23:34
  • 4
    Finally, I suspect that where you wrote $\mathbb R^2$, you instead intended to write $\mathbb R^3$; and where you wrote $(\frac{1}{n},0)$ you instead intended to write $(\frac{1}{n},0,0)$. – Lee Mosher Dec 22 '22 at 23:37
  • I would try to prove the second part of your thesis, the composition of simpler loops. But as @LeeMosher pointed out, the amount of factors need not be finite. – Laz Dec 23 '22 at 02:26

1 Answers1

4

Any path in $X$ is a compact subset of $X$, so it cannot lie in infinitely many spheres (I don't know how to reason this).

This is wrong. $H = \{(x_1,x_2,x_3) \in X \mid x_3 = 0\}$ is a copy of the Hawaiian earring. Using Fundamental group of Hawaiian earring is uncountable you will see that there are loops not contained in a finite number of circles, thus not in a finite number of spheres.

Therefore 0 is wrong and 1 and 2 do not make much sense.

I understand that the solution directly follows from an application of van Kampen Theorem, but will this work?

I do not see a direct application of the theorem. This would require to represent $X$ as the union of open subspaces containing the basepoint. But whatever the choice of the basepoint may be, one of these open sets $U$ must contain $x_0 = (0,0,0)$ and thus contain all but finitely many spheres. This does not help us because determining its fundamental group means essentially to know what $\pi_1(X,x_0)$ is (since $U$ retracts to a copy of $X$).

Here is a proof not based on van Kampen.

Let $S^2_n$ be the sphere of radius $\frac1n$ with center $(\frac{1}{n},0,0)$. Define $$X_n = \bigcup_{m=n}^\infty S^2_m .$$

Lemma. Let $u : I = [0,1] \to X$ be a loop based at $x_0$ such that $u(I) \subset X_n$. Then there exists a homotopy $H : u \simeq u'$ rel. $\{0,1\}$ such that $H(I \times I) \subset X_n$ and $u'(I) \subset X_{n+1}$.

Proof. The set $U = u^{-1}(X \setminus X_{n+1})$ is open in $I$ and does not contain the boundary points $0,1$. The connected components of $U$ are pairwise disjoint open intervals $J_\alpha, \alpha \in A$. Let $x_n = (\frac2n,0,0)$ and $A' = \{ \alpha \in A \mid x_n \in u(J_\alpha)\}$. We claim that $A'$ is finite.

Assume that $A'$ is infinite. Then there exists a sequence of pairwise distinct $\alpha_i \in A'$, $i \in \mathbb N$. Pick $s_i \in J_{\alpha_i}$ such that $u(s_i) = x_n$. Note that by construction each $J_\alpha$ with $\alpha \in A$ contains at most one $s_i$. The sequence $(s_i)$ has cluster point $\sigma \in I$. Since $u$ is continuous, we have $u(\sigma) = x_n$. Thus $\sigma \in U$. Let $\alpha^* \in A$ be the index such that $\sigma \in J_{\alpha^*}$. Since $J_{\alpha^*}$ is an open neighborhood of $\sigma$, infinitely many $s_i$ must be contained in $J_{\alpha^*}$. This is a contradiction.

So let $J_i = (a_i,b_i)$, $i=1,\ldots, k$ be the finitely many components of $U$ whose image under $u$ contains $x_n$. We have $u(a_i) = u(b_i) = x_0$, hence we get loops $u_i : [a_i,b_i] \to S^2_n$ based at $x_0$. Note that the $[a_i,b_i]$ can intersect only in boundary points. The $u_i$ are homotopic in $S^2_n$ rel. $\{a_m,b_m\}$ to the constant loops based at $x_0$. The corresponding homotopies and the stationary homotopy on $I \setminus \bigcup_{i=1}^k J_i$ fit together to a homotopy $G$ rel. $\{0,1\}$ from $u$ to a loop $v$ based at $x_0$ with image in $X_n \setminus \{x_n\}$. Clearly $G(I \times I) \subset X_n$.

$X_{n+1}$ is a strong deformation retract of $X_n \setminus \{x_n\}$, thus we get a homotopy $K : v \simeq u'$ rel. $\{0,1\}$ such that $u'(I) \subset X_{n+1}$ and $K(I \times I) \subset X_n$. Let $H$ be the homotopy obtained by taking first $G$ and then $K$. By construction $H : u \simeq u'$ and $H(I \times I) \subset X_n$.

Theorem. $\pi_1(X,x_0) = 0$.

Proof. Let $u : [0,1] \to X$ be a loop based at $x_0$. Starting with $u_1 = u$, the lemma allows to construct inductively a sequence of loops $u_n : I \to X$ based at $x_0$ and homotopies $H_n : u_n \simeq u_{n+1}$ rel. $\{0,1\}$ such that $u_n(I) \subset X_n$ and $H_n(I \times I) \subset X_n$. Identify $[0,1]$ with $[0,\infty]$ and define $$H : [0,1] \times [0,\infty] \to X, H(s,t) = \begin{cases} H_n(s,t-n+1) & t \in [n-1,n]\\ x_0 & t = \infty \end{cases}$$ This map is well-defined and obviously continuous on $I \times [0,\infty)$. It is also continuous in all $(s,\infty)$. In fact, let $U \subset X$ be an open neighborhood of $x_0 = H(s,\infty)$ in $X$. It contains $X_n$ for $n \ge n_0$. By construction the $H_n$ take place inside $X_n$, thus $H(I \times(n,\infty]) \subset U$. Clearly $I \times(n,\infty]$ is an open neighborhood of $(s,\infty)$ in $I \times [0,\infty]$.

We have $H(s,0) = H_1(s,0) = u(s)$ an $H(s,\infty) = x_0$.

This shows that $u$ represents the trivial element in $\pi_1(X,x_0)$.

Remark.

The above proof is based on the fact that the inclusion $i_n : X_{n+1} \to X_n$ induces an isomorphism $$(i_n)_* : \pi_1(X_{n+1}) \to \pi_1(X_n) .$$

This was proved in the lemma. It also follows from the more general theorem in the answer of Under what conditions $\pi_1(X \vee Y) \approx \pi_1(X) * \pi_1(Y)$?

Paul Frost
  • 87,968