ProofComplete

Singular Homology - Key Proof

We prove homotopy invariance, the cornerstone property establishing singular homology as a topological invariant.

Proof

Theorem: If f,g:XYf, g : X \to Y are homotopic maps via H:X×[0,1]YH : X \times [0,1] \to Y with H(x,0)=f(x)H(x, 0) = f(x) and H(x,1)=g(x)H(x, 1) = g(x), then f=g:Hn(X)Hn(Y)f_* = g_* : H_n(X) \to H_n(Y) for all nn.

Strategy: Construct a chain homotopy P:Cn(X)Cn+1(Y)P : C_n(X) \to C_{n+1}(Y) satisfying: P+P=g#f#\partial P + P\partial = g_\# - f_\# on chain groups. This algebraic relation implies f=gf_* = g_* on homology.

Step 1: Defining the prism operator

For a singular simplex σ:ΔnX\sigma : \Delta^n \to X, define the prism σ×I:Δn×[0,1]X×[0,1]\sigma \times I : \Delta^n \times [0,1] \to X \times [0,1] by (σ×I)(t,s)=(σ(t),s)(\sigma \times I)(t, s) = (\sigma(t), s).

Compose with HH to get H(σ×I):Δn×[0,1]YH \circ (\sigma \times I) : \Delta^n \times [0,1] \to Y. The prism Δn×[0,1]\Delta^n \times [0,1] can be subdivided into (n+1)(n+1)-simplices.

Step 2: Subdividing the prism

Write Δn×[0,1]\Delta^n \times [0,1] as a union of (n+1)(n+1)-simplices. Standard construction: if Δn\Delta^n has vertices v0,,vnv_0, \ldots, v_n, then Δn×[0,1]\Delta^n \times [0,1] has vertices (vi,0)(v_i, 0) and (vi,1)(v_i, 1). Subdivide using the (n+1)!(n+1)! simplices corresponding to orderings.

Define P(σ)P(\sigma) as the signed sum of these (n+1)(n+1)-simplices composed with H(σ×I)H \circ (\sigma \times I).

Step 3: Computing boundaries

The boundary of Δn×[0,1]\Delta^n \times [0,1] consists of:

  • Top face: Δn×{1}\Delta^n \times \{1\} (where HH equals gg)
  • Bottom face: Δn×{0}\Delta^n \times \{0\} (where HH equals ff)
  • Side faces: (n+1)(n+1)-simplices with vertices (vi,0)(v_i, 0) and (vi,1)(v_i, 1) omitted alternately

Computing P(σ)\partial P(\sigma) using the boundary formula: P(σ)=g(σ)f(σ)P(σ)\partial P(\sigma) = g(\sigma) - f(\sigma) - P(\partial \sigma)

Rearranging: P+P=g#f#\partial P + P\partial = g_\# - f_\# on Cn(X)C_n(X).

Step 4: Descending to homology

Let [α]Hn(X)[\alpha] \in H_n(X), so α=0\partial \alpha = 0. Then: g#(α)f#(α)=P(α)+P(α)=P(α)g_\#(\alpha) - f_\#(\alpha) = \partial P(\alpha) + P(\partial \alpha) = \partial P(\alpha)

Thus g#(α)f#(α)Bn(Y)g_\#(\alpha) - f_\#(\alpha) \in B_n(Y) is a boundary, so [g#(α)]=[f#(α)][g_\#(\alpha)] = [f_\#(\alpha)] in Hn(Y)H_n(Y).

Therefore g([α])=f([α])g_*([\alpha]) = f_*([\alpha]) for all [α]Hn(X)[\alpha] \in H_n(X), proving f=gf_* = g_*. ∎

Remark

The prism operator PP is a chain homotopy, the algebraic analogue of geometric homotopy. Chain homotopy equivalence is the correct notion of equivalence for chain complexes, and it descends to isomorphism on homology.

Remark

This proof works for any homology theory with chain-level structure. The key is constructing the homotopy operator PP from the geometric homotopy HH. For CW complexes, cellular chains simplify this construction significantly.