๐Ÿงฎ Brain Teaser

The Quotient That Closes Up: [0,1]ร—[0,1][0,1] \times [0,1] Mod Its Boundary

Consider the unit square I2=[0,1]ร—[0,1]I^2 = [0,1]\times[0,1] with its usual topology. Let โˆ‚I2\partial I^2 denote its boundary (the four edges).

Define the quotient space X=I2/โˆ‚I2X = I^2 / \partial I^2, where all points of โˆ‚I2\partial I^2 are identified to a single point โˆ—*.

Prove that XX is homeomorphic to S2S^2, the 2-sphere.

Hint: Think geometrically. Can you describe a continuous bijection explicitly?

quotient spaceshomeomorphismcompactnesssphereclosed-map lemma

Answer: Collapsing Square Boundary Gives Sphere

Key Idea / Intuition

The square is homeomorphic to the closed disk D2D^2 (just round the corners). When you collapse the boundary circle of a disk to a single point, you are essentially "pinching" the boundary to create a bubble โ€” which is exactly a sphere. The formal argument finds an explicit homeomorphism, or uses the universal property of quotient maps together with a compactness argument.

There are two clean approaches:

  1. Via the disk: First show I2โ‰…D2I^2 \cong D^2, then show D2/S1โ‰…S2D^2/S^1 \cong S^2 via a geometric map.
  2. Direct map: Write down an explicit surjection I2โ†’S2I^2 \to S^2 that collapses โˆ‚I2\partial I^2 to one point, then invoke the closed-map lemma.

We'll use approach 2 since it's the most illuminating.


Formal Proof / Solution

Step 1: The closed-map lemma (key tool).

If f:Xโ†’Yf: X \to Y is a continuous bijection and XX is compact and YY is Hausdorff, then ff is a homeomorphism.

Step 2: Set up the quotient.

Let q:I2โ†’X=I2/โˆ‚I2q: I^2 \to X = I^2/\partial I^2 be the quotient map. The space XX has one special point โˆ—=q(โˆ‚I2)* = q(\partial I^2) and otherwise qq is injective on the interior int(I2)\text{int}(I^2).

Step 3: Describe the explicit map I2โ†’S2I^2 \to S^2.

Identify S2โŠ‚R3S^2 \subset \mathbb{R}^3 as the unit sphere. Use spherical coordinates: parametrize S2S^2 minus the north pole N=(0,0,1)N = (0,0,1) and south pole S=(0,0,โˆ’1)S=(0,0,-1) via latitude ฯ•โˆˆ(0,ฯ€)\phi \in (0,\pi) and longitude ฮธโˆˆ[0,2ฯ€)\theta \in [0, 2\pi).

Define f:I2โ†’S2f: I^2 \to S^2 by mapping (s,t)โ†ฆ(sinโก(ฯ€t)cosโก(2ฯ€s),โ€‰sinโก(ฯ€t)sinโก(2ฯ€s),โ€‰cosโก(ฯ€t))(s,t) \mapsto (\sin(\pi t)\cos(2\pi s),\, \sin(\pi t)\sin(2\pi s),\, \cos(\pi t)).

That is, set ฮธ=2ฯ€s\theta = 2\pi s and ฯ•=ฯ€t\phi = \pi t:

f(s,t)=(sinโก(ฯ€t)cosโก(2ฯ€s),โ€…โ€Šsinโก(ฯ€t)sinโก(2ฯ€s),โ€…โ€Šcosโก(ฯ€t)).f(s,t) = \bigl(\sin(\pi t)\cos(2\pi s),\; \sin(\pi t)\sin(2\pi s),\; \cos(\pi t)\bigr).

Step 4: Verify ff collapses exactly โˆ‚I2\partial I^2.

  • At t=0t = 0 (bottom edge): f(s,0)=(0,0,1)=Nf(s,0) = (0,0,1) = N for all ss. โœ“
  • At t=1t = 1 (top edge): f(s,1)=(0,0,โˆ’1)=Sf(s,1) = (0,0,-1) = S for all ss. โœ“
  • At s=0s = 0 and s=1s = 1 (left/right edges): cosโก(0)=cosโก(2ฯ€)=1\cos(0) = \cos(2\pi) = 1, sinโก(0)=sinโก(2ฯ€)=0\sin(0)=\sin(2\pi)=0, so both vertical edges map to the same curve โ€” they agree since ฮธ=0\theta = 0 and ฮธ=2ฯ€\theta = 2\pi are the same angle. โœ“

So the entire boundary โˆ‚I2\partial I^2 maps to either NN, SS, or the same meridian (ฮธ=0)(\theta=0)... wait โ€” actually, NN and SS are two distinct points! We need ff to collapse all of โˆ‚I2\partial I^2 to one point. Let us reconsider.

Corrected cleaner approach โ€” via the disk:

Step 3' (better): First note that I2โ‰…D2I^2 \cong D^2 (the closed unit disk) via any homeomorphism ฯˆ:D2โ†’I2\psi: D^2 \to I^2 (e.g., radial rescaling to a square). So:

X=I2/โˆ‚I2โ‰…D2/S1.X = I^2/\partial I^2 \cong D^2 / S^1.

Now define g:D2โ†’S2g: D^2 \to S^2 by the formula: for z=(x,y)โˆˆD2z = (x,y) \in D^2 with r=โˆฅzโˆฅโ‰ค1r = \|z\| \le 1,

g(x,y)=(2x1โˆ’r2,โ€…โ€Š2y1โˆ’r2,โ€…โ€Š2r2โˆ’1).g(x,y) = \bigl(2x\sqrt{1-r^2},\; 2y\sqrt{1-r^2},\; 2r^2 - 1\bigr).

Verify gg is well-defined and continuous: clear from the formula.

Verify gg is surjective: For any (a,b,c)โˆˆS2(a,b,c) \in S^2, set r2=1+c2r^2 = \tfrac{1+c}{2}, so r=1+c2โˆˆ[0,1]r = \sqrt{\tfrac{1+c}{2}} \in [0,1], and (x,y)=(a,b)21โˆ’r2(x,y) = \tfrac{(a,b)}{2\sqrt{1-r^2}} when r<1r < 1 (the south pole c=โˆ’1c=-1 gives r=0r=0; the north pole c=1c=1 collapses the boundary). One checks surjectivity.

Verify gg collapses exactly S1=โˆ‚D2S^1 = \partial D^2: On r=1r = 1 (the boundary), 1โˆ’r2=0\sqrt{1-r^2} = 0, so:

g(x,y)โˆฃr=1=(0,โ€‰0,โ€‰1)=N,g(x,y)\big|_{r=1} = (0,\, 0,\, 1) = N,

a single point โ€” the north pole. Interior points r<1r < 1 are mapped injectively (distinct (x,y)(x,y) give distinct (a,b,c)(a,b,c)). โœ“

Step 4': Apply the universal property.

Since g:D2โ†’S2g: D^2 \to S^2 is continuous and collapses S1S^1 to a single point, it factors through the quotient:

g~:D2/S1โ†’S2,g~โˆ˜q=g,\tilde{g}: D^2/S^1 \to S^2, \qquad \tilde{g} \circ q = g,

and g~\tilde{g} is a continuous bijection.

Step 5': Invoke the closed-map lemma.

  • D2/S1D^2/S^1 is compact (quotient of a compact space).
  • S2S^2 is Hausdorff.
  • g~\tilde{g} is a continuous bijection.

Therefore g~\tilde{g} is a homeomorphism. โ– \blacksquare

Conclusion:

I2/โˆ‚I2โ‰…D2/S1โ‰…S2.I^2/\partial I^2 \cong D^2/S^1 \cong S^2.

The beautiful idea is that collapsing the boundary of a disk to a point is the topological operation of "inflating a balloon" โ€” and the closed-map lemma turns a geometric intuition into a rigorous proof with minimal work.

Source: Munkres, Topology, Chapter 2; classic topology folklore

Type: topologySource: Munkres, Topology, Chapter 2; classic topology folkloreEdit on GitHub โ†—