11
$\begingroup$

This looks like a question that could be proved using the Seifert–Van Kampen theorem, but I am not sure if it is true.

Let $X=U \cup V$ be a topological space, where $U$, $V$ and $U\cap V$ are open, path-connected subspaces of $X$. Let $$i_U: \pi_1(U \cap V) \rightarrow \pi_1(U)\quad \text{and}\quad i_V: \pi_1(U \cap V) \rightarrow \pi_1(V)$$ be the inclusion maps. Suppose there exists a homotopically nontrivial element $\alpha \in \pi_1(U \cap V)$ such that $i_U(\alpha)$ is also nontrivial in $\pi_1(U)$. Is it true that if the inclusion map $i_V$ is injective, then $\alpha$ is nontrivial in $\pi_1(X)$?

Edit 1: I tried using the amalgamated product of fundamental groups of $U$ and $V$ to prove this result. Roughly speaking $\pi_1(X) = \langle \pi_1(U), \pi_1(V) : i_V(\alpha_i) = i_U(\alpha_i) \rangle$ where $\alpha_i$ are the generators of the intersection $\pi_1(U\cap V)$. By assumption none of the additional relations in the finite presentation of the fundamental group is of the form $\alpha = 1$. But I find it difficult to prove that the relations in the groups $\pi_1(U)$ and $\pi_1(V)$ cannot be used with the additional relations in the amalgamated product to yield $\alpha =1$.

Edit 2: Following the suggestion in the comments, the group-theoretic version of this question can be found here.

New contributor
Holomaniac is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
$\endgroup$
7
  • 1
    $\begingroup$ What have you tried? Please show us your work! $\endgroup$ Commented yesterday
  • $\begingroup$ @BenSteffan I tried using the amalgamated product of the fundamental groups of $U$ and $V$ presented as a free group with relations. Roughly speaking $\pi_1(X) = \langle \pi_1(U), \pi_1(V) \lvert i_V(\alpha_i) = i_U(\alpha_i) \rangle$ where $\alpha_i$ is in the intersection. By assumption none of the additional relations in the fundamental group is of the form $\alpha = 1$. But I find it difficult to prove that the relations defining the groups $\pi_1(U)$ and $\pi_1(V)$ cannot be used with the additional relations in the amalgamated product to yield $\alpha =1$. $\endgroup$
    – Holomaniac
    Commented yesterday
  • 1
    $\begingroup$ @JyrkiLantonen I am sorry if I am being slow here, but if $U\cup V$ is a "cap" (meaning a disk?), and $V$ is a belt, then shouldn't $U$ be a disk as well? One of the assumptions is that the fundamental group of $U$ is nontrivial. $\endgroup$
    – Holomaniac
    Commented yesterday
  • 5
    $\begingroup$ Perhaps it might be a good idea to ask the group theoretic question underpinning this separately: Given a span of groups $G' \xleftarrow{\phi} G \xrightarrow{\psi} G''$ where $\phi$ is injective and an element $0 \neq g \in G$, is the image of $g$ in $G' *_G G''$ non-zero provided that $\psi(g) \neq 0$? I'd expect you'd be much more likely to get an answer to that from the group theory people than the algebraic topologists around. $\endgroup$ Commented yesterday
  • 2
    $\begingroup$ The group-theoretic version (see "Edit 2" above) is now answered in the negative. $\endgroup$
    – Lee Mosher
    Commented yesterday

1 Answer 1

12
$\begingroup$

Here is an intuitive counterexample.

Consider two U-shaped curves in $\mathbb{R}^3$:

  • $K_1$, which extends upward, and
  • $K_2$, which extends downward and hooks around $K_1$,

as shown in the figure. Assume that one end of $K_1$ has finite length, while the other extends infinitely. Let $X$ be the complement of these curves,

$$ X = \mathbb{R}^3 \setminus (K_1 \cup K_2), $$

and define open subsets $U, V \subset X$ by

$$ U = \{(x, y, z) \in X \mid z > 0\}, \quad V = \{(x, y, z) \in X \mid z < 1\}. $$

In this setup, the fundamental groups of $U \cap V$, $U$, and $V$ are as follows:

  • $\pi_1(U \cap V) = \langle \alpha, \beta \rangle$.
  • $\pi_1(U)=\langle \alpha\rangle$. The natural map $i_U \colon \pi_1(U \cap V) \to \pi_1(U)$ is given by $\alpha \mapsto \alpha$, $\beta \mapsto 1$.
  • $\pi_1(V) = \langle \alpha, \gamma\rangle$. The natural map $i_V \colon \pi_1(U \cap V) \to \pi_1(V)$ is given by $\alpha \mapsto \alpha$, $\beta \mapsto \gamma\alpha\gamma^{-1}$, and is therefore injective.

However, from the figure, one can see that $\alpha$ can be contracted within $X$.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.