Simplicial Homology Groups

Research Depth 28 in the knowledge graph I know this Set as goal
Unlocks 23 downstream topics
algebraic-topology simplicial-homology homology-groups cycles boundaries

Core Idea

The n-th simplicial homology group H_n(K) = ker(d_n)/im(d_{n+1}) measures the n-dimensional "holes" in a simplicial complex K. Elements of H_n are equivalence classes of n-cycles (chains with zero boundary) modulo n-boundaries (chains that bound (n+1)-dimensional regions). H_0 counts connected components, H_1 detects loops that do not bound surfaces, and H_2 detects enclosed cavities. Homology groups are computable topological invariants — different triangulations of the same space yield isomorphic groups.

Explainer

Given a simplicial complex K with its chain complex ... -> C_2(K) -d_2-> C_1(K) -d_1-> C_0(K) -> 0, the n-th simplicial homology group is the quotient H_n(K) = ker(d_n) / im(d_{n+1}). The kernel ker(d_n), called the group of n-cycles and denoted Z_n, consists of all n-chains whose boundary is zero. The image im(d_{n+1}), called the group of n-boundaries and denoted B_n, consists of all n-chains that are boundaries of (n+1)-chains. Since d_n compose d_{n+1} = 0, every boundary is a cycle (B_n is a subgroup of Z_n), and the homology group H_n = Z_n / B_n measures how much larger the cycle group is than the boundary group — cycles that are not boundaries represent genuine "holes."

The zeroth homology H_0(K) counts connected components. A 0-chain is a formal sum of vertices, and its boundary is zero (there is no d_{-1}), so every 0-chain is a cycle. A 0-boundary is d_1 of some 1-chain — for an edge [a,b], d_1([a,b]) = b - a. Two vertices are homologous (represent the same element of H_0) if and only if they are connected by a path of edges. Thus H_0(K) is a free abelian group with one generator per connected component. For a connected complex, H_0 is simply Z.

First homology H_1(K) detects 1-dimensional holes — loops that cannot be filled. A 1-cycle is a chain of edges forming a closed loop (every vertex appears as many times as a head as it does as a tail). A 1-boundary is the boundary of a 2-chain — a sum of triangles whose boundary edges form the loop. If the loop bounds a filled region, it is a boundary and represents zero in H_1. If the loop surrounds a hole (a missing interior), it is a cycle but not a boundary, representing a nontrivial homology class. For example, the triangulated torus has H_1 isomorphic to Z direct sum Z, with generators being the two fundamental loops (around the hole and through the tube) — neither loop can be filled by triangles in the torus.

Higher homology works analogously. H_2(K) detects 2-dimensional "cavities" — closed surfaces that do not bound solid regions. The triangulated 2-sphere has H_2 isomorphic to Z: the entire sphere is a 2-cycle (a closed surface with no boundary edges), but it does not bound any 3-chain (there is no solid ball in the complex). The Betti numbers b_n = rank(H_n) count the number of independent n-dimensional holes: b_0 is the number of components, b_1 the number of independent tunnels or loops, b_2 the number of enclosed cavities. Torsion elements (elements of finite order) in H_n detect more subtle phenomena related to non-orientability: the real projective plane has H_1 isomorphic to Z/2Z because its central loop, traversed twice, bounds a disk in the projective plane.

The remarkable fact about simplicial homology — and what makes it a central tool in topology — is its topological invariance: homeomorphic spaces have isomorphic homology groups, regardless of how they are triangulated. This means homology genuinely measures the shape of a space, not an artifact of its combinatorial presentation. Computing simplicial homology reduces to linear algebra: write down the boundary matrices, compute their kernels and images (via row reduction over Z), and take the quotient. This algorithmic computability, combined with topological invariance, makes simplicial homology one of the most powerful and practical invariants in all of topology.

Practice Questions 5 questions

Prerequisite Chain

Longest path: 29 steps · 105 total prerequisite topics

Prerequisites (2)

Leads To (3)