A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
A Delta-complex is a generalization of a simplicial complex that allows simplices to have self-identifications on their faces — for instance, a triangle can have two of its edges identified, which is impossible in a simplicial complex. This flexibility dramatically reduces the number of simplices needed to triangulate a space while preserving the ability to compute homology via chain complexes and boundary operators. Delta-complexes serve as a practical middle ground between the rigidity of simplicial complexes and the full generality of singular homology.
A Delta-complex (sometimes written Δ-complex) is a space built by attaching standard simplices via continuous maps on their faces, where the attaching maps are required to be order-preserving on vertices but are allowed to identify faces. More precisely, a Delta-complex structure on a space X specifies, for each n, a collection of continuous maps sigma_alpha : Deltan -> X (one for each n-simplex of the complex), such that: (1) the restriction of each sigma_alpha to the interior of Deltan is injective, with the interiors of all the simplices partitioning X; (2) the restriction of sigma_alpha to each face of Deltan is one of the (n-1)-dimensional maps sigma_beta (the face maps are themselves simplices of the complex); (3) a set A subset X is open if and only if sigma_alpha-1(A) is open in Deltan for every sigma_alpha.
The key difference from simplicial complexes is that a Delta-complex allows face identifications: two faces of the same simplex, or faces of different simplices, can be identified by the attaching maps. In a simplicial complex, the vertices of every simplex must be distinct, and two simplices can only intersect along a common face. Delta-complexes relax both conditions. A single edge can be a loop (both endpoints at the same vertex), and a triangle can have two or even all three edges identified. This gives Delta-complexes enormous flexibility in representing spaces with few cells.
The standard example is the torus represented as a square with opposite sides identified (the classic gluing diagram with edges labeled a, b, a, b). Cutting the square along a diagonal gives two triangles, and after the identifications, the Delta-complex has 1 vertex (all four corners of the square are identified), 3 edges (the two sides and the diagonal, after identifications), and 2 triangles. Compare this to the minimum of 7 vertices, 21 edges, and 14 triangles required for a simplicial triangulation. The boundary computation for this Delta-complex is straightforward: d_2 of each triangle gives an alternating sum of the three edges, with the identification maps determining which named edge each face becomes. The resulting chain complex is small enough to compute homology by hand.
The chain complex of a Delta-complex is defined exactly as for simplicial complexes: C_n is the free abelian group on the n-simplices, and the boundary operator is the alternating sum of face maps. The crucial property d compose d = 0 still holds, because it follows from the same combinatorial identity as in the simplicial case (the double alternating sum telescopes). The homology groups computed from a Delta-complex are isomorphic to the simplicial (and singular) homology groups of the underlying space. This isomorphism is proved by comparing both to singular homology or by using the simplicial approximation theorem to relate Delta-complex and simplicial complex structures.
Delta-complexes occupy a useful middle position in the hierarchy of cell-like structures: more flexible than simplicial complexes, more structured than CW complexes, and more concrete than singular chains. They appear prominently in Hatcher's "Algebraic Topology" as the primary vehicle for introducing homology, precisely because they combine computational tractability with geometric economy. Understanding Delta-complexes builds intuition for both the combinatorial foundations (how boundary operators encode topology) and the later generalization to CW complexes and singular homology.