Poincare Duality

Core Idea

Poincare duality states that for a closed, connected, oriented n-manifold M, there is an isomorphism H^k(M; Z) = H_{n-k}(M; Z) for all k. This isomorphism is given by the cap product with the fundamental class [M] in H_n(M): the map alpha -> alpha cap [M] is an isomorphism H^k(M) -> H_{n-k}(M). Poincare duality is one of the deepest theorems in algebraic topology, revealing a perfect symmetry between the homology and cohomology of manifolds and connecting the dimensions k and n-k in a precise way.

Explainer

Poincare duality is one of the crown jewels of algebraic topology, first conjectured by Poincare in 1895 and proved rigorously using the machinery of singular homology in the mid-20th century. The theorem states: if M is a closed (compact without boundary), connected, oriented n-manifold, then for every k, there is an isomorphism D : H^k(M; Z) -> H_{n-k}(M; Z) given by the cap product with the fundamental class: D(alpha) = alpha cap [M]. The fundamental class [M] is the unique generator of H_n(M; Z) = Z compatible with the orientation.

The cap product cap : H^k(M; Z) x H_n(M; Z) -> H_{n-k}(M; Z) is defined at the chain/cochain level: for a k-cochain f and an n-chain sigma, (f cap sigma) evaluates f on the front k-face of sigma and returns the back (n-k)-face. Formally, if sigma : Deltaⁿ -> M is a singular n-simplex, then f cap sigma = f(sigma|_{[v_0,...,v_k]}) * sigma|_{[v_k,...,v_n]}. This descends to homology/cohomology and provides a well-defined pairing. The Poincare duality isomorphism alpha -> alpha cap [M] is obtained by capping all of the fundamental class with the cohomology class alpha.

Poincare duality reveals a perfect symmetry in the Betti numbers of closed oriented manifolds: b_k = b_{n-k} for all k. For a closed oriented surface of genus g (n = 2): b_0 = b_2 = 1 and b_1 = 2g — the symmetry b_0 = b_2 is visible. For a closed oriented 3-manifold: b_0 = b_3 and b_1 = b_2. For a closed oriented 4-manifold: b_0 = b_4, b_1 = b_3, and b_2 is unconstrained (it equals itself). This symmetry in Betti numbers is a consequence of the duality and provides an immediate consistency check on homology computations for manifolds.

The hypothesis of orientability is essential. A non-orientable closed n-manifold has H_n(M; Z) = 0 (there is no Z-fundamental class), and the Poincare duality symmetry fails with integer coefficients. The real projective plane RP² illustrates this: H_0 = Z, H_1 = Z/2Z, H_2 = 0, which is not symmetric. However, every closed manifold — orientable or not — has a fundamental class with Z/2Z coefficients, and Poincare duality holds with Z/2Z coefficients universally: H^k(M; Z/2Z) = H_{n-k}(M; Z/2Z). This is why Z/2Z coefficients play a special role in the topology of non-orientable manifolds.

Poincare duality has profound consequences throughout topology and geometry. It implies that the Euler characteristic of an odd-dimensional closed oriented manifold is zero (since the Betti numbers pair up and cancel in the alternating sum). It gives rise to the intersection form on middle-dimensional homology of 4-manifolds, which is a central invariant in 4-manifold topology (Donaldson's theorem, Freedman's classification). In differential topology, Poincare duality connects to the Hodge star operator and de Rham cohomology. The duality is the topological manifestation of a deep geometric principle: on a manifold, k-dimensional submanifolds can be "dualized" to (n-k)-dimensional quantities, and this duality is captured algebraically by the cap product with the fundamental class.