The groupoid version of van Kampen's theorem states that if $X$ is a topological space and $X = U\cup V$ where $U,V$ are open, then the fundamental groupoid $\tau_{\leq 1}(X)$ of $X$ is the limit of the diagram $$\tau_{\leq 1}(U)\leftarrow \tau_{\leq 1}(U\cap V)\rightarrow \tau_{\leq 1}(V)$$ where the morphisms between fundamental groupoids above are those induced by the inclusions of $U\cap V$ into $U$ and $V$.

If $S = S^1$, we can choose $U$ and $V$ to be homeomorphic to copies of $(0,1)$, so that $U\cap V = (0,1)\sqcup (0,1)$. We can't use the usual version of van Kampen's theorem here because $U\cap V$ isn't path connected. But is it doable to show that $\pi_1(S^1)=\mathbb Z$ using the groupoid version? If so, can I see a proof? Thanks.


One really needs to set up the Seifert-van Kampen theorem for the fundamental groupoid $\pi_1(X,S)$ on a set of base points chosen according to the geometry. One sees the circle as obtained from the unit interval $[0,1]$ by identifying $0$ and $1$. So we need the fundamental groupoid on the set $S= \{0,1\}$. Thus $\pi_1(X)$ can be too large; $\pi_1(X,x)$ can be too small; but $\pi_1(X,S)$ can be just right.

For example we may have $X= U \cup V$, with $X,U,V$ path connected space but $U \cap V$ may have $1,024$ path components. It is also often the case that choosing a single base point will neglect any symmetry in the geometric situation.

Full details are in the book Topology and Groupoids. The first step is to prove a pushout form of the SvKT, when $S$ meets each path component of $U,V, W =U \cap V$. Assume further that $V$ say is path connected, and $x \in W \cap S$. Then one can set up a composite of pushouts:

$$\begin{matrix}\pi_1(W,S) & \to & \pi_1(V,S) & \to & \pi_1(W,x)\\ \downarrow & & \downarrow & &\downarrow\\ \pi_1(U,S) & \to & \pi_1(X,S)& \to & \pi_1(X,x) \end{matrix} $$ which is itself a pushout.

In the case of the circle this becomes a composite of the form

$$\begin{matrix} \{0,1\} &\to & \{ 0 \}\\ \downarrow & & \downarrow\\ \pi_1([0,1], \{0,1\}) & \to & \pi_1(S^1,0). \end{matrix} $$

Since $[0,1]$ is convex, $\pi_1([0,1], \{0,1\}) \cong \mathcal I$, where the latter is the groupoid with two objects $0,1$ and two non identity arrows $\iota:0 \to 1$ and its inverse $ \iota^{-1}:1 \to 0$. This is a model of the "unit interval" in the category of groupoids. From these facts we can determine $\pi_1(S^1,0)$. Note that within $[0,1]$ we have no special reason to choose $0$ or $1$ as base point, so we choose them both.

Categorically speaking, the groupoid $\mathcal I$ has similar properties in the category of groupoids to those of the integers $\mathbb Z$ in the category of groups. So although it looks "trivial", it should not be ignored.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.