🧮 Brain Teaser

The Torus Minus a Point Deformation Retracts onto a Wedge

Let T=S1×S1T = S^1 \times S^1 be the torus. Show that T{p}T \setminus \{p\} (the torus with one point removed) is homotopy equivalent to S1S1S^1 \vee S^1.

Hint: Think about the standard CW structure on the torus.

CW complexdeformation retractfundamental grouptoruswedge sum

Answer: Torus Minus a Point Deformation Retracts onto Wedge

Key Idea / Intuition

The torus has a beautiful CW structure: one 0-cell, two 1-cells (the longitude and meridian circles), and one 2-cell whose boundary is glued according to aba1b1aba^{-1}b^{-1}. When you remove a point from the interior of the 2-cell, that 2-cell — now punctured — can be collapsed: a disk minus an interior point deformation retracts onto its boundary circle. So the whole space collapses onto the 1-skeleton, which is exactly S1S1S^1 \vee S^1.

The key insight is purely combinatorial/topological: removing a point from the top-dimensional cell allows that cell to retract to its boundary, leaving only the lower-dimensional skeleton.


Formal Proof / Solution

Step 1: CW structure of TT.

Give TT the standard CW structure:

  • One 0-cell: e0={p0}e^0 = \{p_0\}
  • Two 1-cells: ea1e^1_a and eb1e^1_b (representing the two generating loops)
  • One 2-cell: e2e^2 attached via the word aba1b1aba^{-1}b^{-1}

The 1-skeleton is e0ea1eb1S1S1e^0 \cup e^1_a \cup e^1_b \cong S^1 \vee S^1.

Step 2: Remove a point from the interior of e2e^2.

Choose pInt(e2)p \in \mathrm{Int}(e^2), i.e., a point in the open 2-cell. The resulting space is: T{p}=(S1S1)(e2{p})T \setminus \{p\} = (S^1 \vee S^1) \cup (e^2 \setminus \{p\})

Step 3: The punctured 2-cell retracts to its boundary.

The open 2-cell e2e^2 is homeomorphic to an open disk D2D^2. Removing a point pp from its interior gives a space homeomorphic to D2{0}D^2 \setminus \{0\}, which deformation retracts onto D2=S1\partial D^2 = S^1 by the straight-line retraction: H(x,t)=(1t)x+txx(1t)x+txx(scaled appropriately)H(x, t) = \frac{(1-t)x + t \cdot \frac{x}{|x|}}{|(1-t)x + t \cdot \frac{x}{|x|}|} \cdot \text{(scaled appropriately)}

More precisely, define r:D2{0}S1r: D^2 \setminus \{0\} \to S^1 by r(x)=x/xr(x) = x/|x|, and the deformation retraction H(x,t)=xxtH(x,t) = \frac{x}{|x|^t} which at t=0t=0 is the identity and at t=1t=1 is rr.

Step 4: The retraction is compatible with the attaching map.

Since the retraction HH fixes D2\partial D^2 pointwise (because x=1|x| = 1 on the boundary), it is compatible with the attaching map φ:D2S1S1\varphi: \partial D^2 \to S^1 \vee S^1. Therefore, the deformation retraction of e2{p}e^2 \setminus \{p\} onto e2\partial e^2 induces a deformation retraction of the entire space T{p}T \setminus \{p\} onto the 1-skeleton.

Step 5: Conclusion.

We have a deformation retraction: T{p}    S1S1T \setminus \{p\} \;\simeq\; S^1 \vee S^1

In particular, π1(T{p})π1(S1S1)ZZ\pi_1(T \setminus \{p\}) \cong \pi_1(S^1 \vee S^1) \cong \mathbb{Z} * \mathbb{Z}, the free group on two generators — a dramatic contrast with π1(T)Z×Z\pi_1(T) \cong \mathbb{Z} \times \mathbb{Z}, which is abelian. Removing just one point from the torus makes the fundamental group non-abelian!

Source: Introduction to Topological Manifolds, John M. Lee (standard result); also Hatcher Algebraic Topology §1.1

Type: topologySource: Introduction to Topological Manifolds, John M. Lee (standard result); also Hatcher Algebraic Topology §1.1Edit on GitHub ↗