1 Definition and basic examples

1.1 Chain complexes and chain maps (recall)

A chain complex over an abelian category (often modules over a ring) consists of objects \(C_n\) indexed by integers together with differentials \[ d_n : C_n \to C_{n-1} \] such that \(d_{n-1}\circ d_n=0\) for all \(n\). A chain map \(f:C_\bullet\to D_\bullet\) is a collection of morphisms \(f_n:C_n\to D_n\) commuting with differentials: \[ d^D_n \circ f_n = f_{n-1}\circ d^C_n \] for every degree \(n\). The index shift in the compatibility reflects that differentials lower degree in chain complexes.

1.2 Chain homotopy: formal definition

Let \(f,g:C_\bullet\to D_\bullet\) be chain maps. A chain homotopy from \(f\) to \(g\) is a family of morphisms \[ s_n : C_n \to D_{n+1} \] (degree \(+1\)) such that for each \(n\), \[ g_n - f_n = d^D_{n+1}\circ s_n + s_{n-1}\circ d^C_n. \] If such \(s_\bullet\) exists, \(f\) and \(g\) are chain homotopic, written \(f\simeq g\).

Equivalently, the difference \(g-f\) is “exact up to \(s\)” with respect to the differential structure on morphisms between complexes.

1.3 Degree conventions and sign conventions

The definition depends on two conventions:

  1. Chain vs. cochain: chain complexes use differentials \(d\) that lower degree, while cochain complexes use differentials that raise degree. For chain complexes, homotopy maps typically raise degree by \(1\).
  2. Sign placement: some authors write

\[ g_n-f_n = d^D_{n+1}s_n - s_{n-1}d^C_n \] or use different indexing for \(s_n\). These variants are equivalent after adopting a consistent sign rule. The essential idea remains: homotopy expresses \(g-f\) as a commutator-like expression built from the differentials and the degree-shifted family \(s_\bullet\).

1.4 Simple example: homotopy between maps of short complexes

Consider chain complexes supported in only two degrees, say \(C_1 \xrightarrow{d_1} C_0\) and \(D_1\xrightarrow{d_1} D_0\), with \(d_0=0\). Suppose \(f,g:C_\bullet\to D_\bullet\) are chain maps. A homotopy consists of a single map \(s_0:C_0\to D_1\) (since \(s_n\) must land in degree \(n+1\)). The homotopy identity at degree \(0\) reads \[ g_0-f_0 = d^D_1\circ s_0 + s_{-1}\circ d^C_0. \] Because \(d^C_0=0\) and \(s_{-1}\) is irrelevant here, this reduces to \(g_0-f_0=d^D_1 s_0\). At degree \(1\), the condition becomes \[ g_1-f_1 = d^D_2 s_1 + s_0 d^C_1, \] but \(D_2=0\) and \(s_1\) is absent, so it reduces to \(g_1-f_1=s_0 d^C_1\). Thus, in this short setting, a chain homotopy is determined by one “degree \(+1\)” map and must satisfy two straightforward component equations.

2 Properties of chain homotopy

2.1 Chain homotopy as an equivalence relation

Chain homotopy is an equivalence relation on chain maps between fixed chain complexes:

  • Reflexive: \(f\simeq f\) via the zero homotopy \(s_n=0\).
  • Symmetric: if \(f\simeq g\) via \(s\), then \(g\simeq f\) via \(-s\).
  • Transitive: if \(f\simeq g\) via \(s\) and \(g\simeq h\) via \(t\), then \(f\simeq h\) via \(s+t\), since the defining identity is linear in the homotopy maps.

2.2 Compatibility with addition and scalar multiples

If \(f_1\simeq g_1\) via \(s\) and \(f_2\simeq g_2\) via \(t\), then for any scalar \(\lambda\) (in the ring over which the category is linear), one has:

  • \(f_1+f_2\simeq g_1+g_2\) via \(s+t\),
  • \(\lambda f_1 \simeq \lambda g_1\) via \(\lambda s\),
  • and more generally, the homotopy relation behaves well with respect to forming linear combinations of chain maps.

This follows from the homotopy equation being additive in both \(f\) and \(g\), and linear in the homotopy maps.

2.3 Behavior under composition with chain maps

Chain homotopy respects composition. If \(f\simeq g:C_\bullet\to D_\bullet\) and \(u:D_\bullet\to E_\bullet\) is a chain map, then \(u\circ f\simeq u\circ g\) using the homotopy maps \(u_{n+1}\circ s_n\). Similarly, if \(v:B_\bullet\to C_\bullet\) is a chain map, then \(f\circ v\simeq g\circ v\) using \(s_n\circ v_n\) (with appropriate degree alignment). In short, homotopies are stable under pre- and post-composition by chain maps.

2.4 Homotopy between identity maps and contractible complexes

A key structural phenomenon is: when a complex is contractible (see later), its identity map is homotopic to the zero map. This immediately implies that any map out of or into such a complex becomes homotopically trivial under suitable composition, making contractible complexes behave like zero objects in the homotopy category.

3 Chain homotopy and induced maps on homology

3.1 Homology of a chain complex (recall)

For a chain complex \(C_\bullet\), the cycles in degree \(n\) are \[ Z_n=\ker(d_n), \] and the boundaries in degree \(n\) are \[ B_n=\operatorname{im}(d_{n+1}). \] The \(n\)-th homology group is \[ H_n(C_\bullet)=Z_n/B_n. \] A chain map \(f\) sends cycles to cycles and boundaries to boundaries, so it induces a map \(H_n(f):H_n(C_\bullet)\to H_n(D_\bullet)\).

3.2 Chain homotopic maps induce the same maps on homology

If \(f\simeq g\) via \(s_\bullet\), then for every degree \(n\), the induced homomorphisms on homology coincide: \[ H_n(f)=H_n(g). \] Intuitively, the difference \(g_n-f_n\) is built from the differentials, so on cycles it lands inside boundaries, and therefore vanishes after quotienting by \(B_n\).

3.3 Consequences for comparing resolutions

In homological algebra, one frequently replaces an object by a chain complex resolution, then computes derived invariants. Chain homotopy invariance implies that if two resolutions are related by maps that are homotopy equivalent in the appropriate sense, the induced maps on homology agree. As a result, computations are stable under changing the chosen resolution within the homotopy class.

3.4 Homotopy invariance principle (statement and intuition)

A standard principle can be summarized as: homotopic chain maps represent the same morphism between homology groups. More generally, chain homotopy treats complexes as “the same” for purposes of homology. This is the mechanism behind many constructions where one proves that a map is a homology isomorphism by exhibiting a homotopy equivalence or by reducing to contractible pieces.

4 Contractible and acyclic complexes

4.1 Contractible complexes: definition and characterization

A chain complex \(C_\bullet\) is contractible if its identity map \(\operatorname{id}_{C_\bullet}\) is chain homotopic to the zero map. Concretely, there exist morphisms \[ s_n:C_n\to C_{n+1} \] such that for every \(n\), \[ \operatorname{id}_{C_n}= d_{n+1}\circ s_n + s_{n-1}\circ d_n. \] This data is called a contracting homotopy.

4.2 Relationship between contractibility and acyclicity

Contractibility implies acyclicity: if \(C_\bullet\) is contractible, then all homology groups vanish, \[ H_n(C_\bullet)=0 \quad \text{for all } n. \] The reason is that the contracting identity provides an explicit way to write any cycle as a boundary. Contractible complexes are therefore stronger than merely acyclic ones.

4.3 Explicit contracting homotopies

A contracting homotopy turns the abstract statement “homology vanishes” into an explicit formula. Given a cycle \(z\in Z_n\), one can compute that \[ z = d_{n+1}(s_n(z)) \] using the contracting homotopy identity together with \(d_n(z)=0\). This explicitness is often exploited computationally, for example in verifying that certain complexes built from trivial extensions are indeed contractible.

4.4 Examples illustrating nontrivial homology vs contractibility

Two contrasts are common:

  • A complex may be acyclic without being contractible (depending on the category and additional structure). In such cases, homology vanishes but no contracting homotopy exists.
  • Conversely, a contractible complex necessarily has trivial homology, but its structure is “degenerate” in a way that is witnessed by a degree-shifted homotopy \(s_\bullet\).

These examples highlight that “acyclic” and “contractible” are related but not identical notions.

5 Homotopy category of chain complexes

5.1 From chain complexes to the homotopy category

The homotopy category \(K(\mathcal{A})\) of chain complexes over an additive category \(\mathcal{A}\) has:

  • objects: chain complexes,
  • morphisms: chain maps modulo chain homotopy.

Composition is induced from composition of chain maps; stability under composition (Section 2.3) ensures this is well-defined.

5.2 Morphisms modulo chain homotopy

Two chain maps \(f,g:C_\bullet\to D_\bullet\) represent the same morphism in \(K(\mathcal{A})\) exactly when they are chain homotopic. This quotient formalizes the idea that homological information cannot distinguish maps within a homotopy class.

5.3 Isomorphisms in the homotopy category

An isomorphism in \(K(\mathcal{A})\) corresponds to a chain homotopy equivalence: a chain map \(f:C_\bullet\to D_\bullet\) admitting a chain map \(g:D_\bullet\to C_\bullet\) such that both \(g\circ f\) and \(f\circ g\) are chain homotopic to the appropriate identities. Under such an equivalence, homology groups of \(C_\bullet\) and \(D_\bullet\) are naturally identified.

5.4 Functoriality and basic categorical viewpoint

Many constructions are natural at the level of the homotopy category: mapping cones, shifts, and derived functors (under additional conditions) are compatible with the quotient by homotopy. From a categorical perspective, the homotopy category is where chain homotopy becomes invisible and only the homotopy-invariant content remains.

6 Mapping cone and derived constructions

6.1 Mapping cone construction for a chain map

Given a chain map \(f:C_\bullet\to D_\bullet\), the mapping cone \(\mathrm{Cone}(f)_\bullet\) is a new chain complex defined by \[ \mathrm{Cone}(f)_n = D_n \oplus C_{n-1}, \] with differential (one standard convention) \[ d^{\mathrm{Cone}}_n = \begin{pmatrix} d^D_n & f_{n-1}\\ 0 & -d^C_{n-1} \end{pmatrix}. \] This construction packages the failure of \(f\) to be an isomorphism into a complex whose homology measures that failure in a precise way.

6.2 Mapping cone and homotopy equivalence

If \(f:C_\bullet\to D_\bullet\) is a chain homotopy equivalence, then \(\mathrm{Cone}(f)\) is contractible, hence has trivial homology. Conversely, contractibility of the cone is closely related to the existence of a homotopy inverse for \(f\). Thus, mapping cones provide a bridge between homotopy equivalence and vanishing of homology.

6.3 Triangles and exactness-like behavior in the homotopy category

Mapping cones give rise to distinguished triangles in the homotopy category (or in derived settings). While not “exact” in the classical sense of short exact sequences, these triangles reproduce the correct homological long exact sequences after passing to homology. In practice, many equivalences and comparisons are reduced to analyzing cones and their induced long exact sequences.

6.4 Applications: proving equivalences via cones

A common strategy is:

  1. Construct the mapping cone of a candidate map \(f\).
  2. Show the cone is contractible or has trivial homology for all degrees.
  3. Conclude that \(f\) is a homotopy equivalence (under appropriate hypotheses).

Because cones turn questions about maps into questions about complexes, this method is often effective, especially when the cone has a recognizable contracting homotopy.

7.1 Chain homotopy equivalence

A chain map \(f:C_\bullet\to D_\bullet\) is a chain homotopy equivalence if there exists a chain map \(g:D_\bullet\to C_\bullet\) such that \[ g\circ f \simeq \operatorname{id}_{C_\bullet} \quad\text{and}\quad f\circ g \simeq \operatorname{id}_{D_\bullet}. \] Such a map becomes an isomorphism in the homotopy category, and induces isomorphisms on all homology groups.

7.2 Chain homotopy vs quasi-isomorphism (conceptual comparison)

A quasi-isomorphism is a chain map that induces isomorphisms on homology groups. Every chain homotopy equivalence is a quasi-isomorphism, but the converse can fail: two complexes may have the same homology while not being homotopy equivalent as complexes. The distinction is important in derived-category methods, where quasi-isomorphisms are inverted rather than restricting to homotopy equivalences.

7.3 Comparison with cochain complexes (dual perspective)

For cochain complexes \(C^\bullet\) with differentials \(d^n:C^n\to C^{n+1}\), the dual notions use degree shifts in the opposite direction. The homotopy identity changes accordingly (the “homotopy maps” have degree \(-1\) in cochain conventions). Conceptually, the theory is the same: homotopy invariance ensures that homotopic maps induce the same maps on cohomology.

7.4 Null-homotopic maps and their characterizations

A chain map \(f:C_\bullet\to D_\bullet\) is null-homotopic if it is chain homotopic to the zero map. In terms of homotopy data, \(f\) is null-homotopic precisely when there exist maps \(s_n:C_n\to D_{n+1}\) such that \[ f_n = d^D_{n+1}\circ s_n + s_{n-1}\circ d^C_n. \] Null-homotopic maps induce the zero morphism on homology, since they vanish after passing to cycles modulo boundaries.

8 Computational techniques and verification

8.1 Constructing a chain homotopy from given data

In practice, one often knows how two maps behave on degrees and on boundaries, and then tries to build \(s_n\) degree by degree. For a fixed degree \(n\), the homotopy equation determines \(g_n-f_n\) in terms of \(s_n\) and \(s_{n-1}\). When the differentials are simple or when modules split, one can choose \(s_n\) to satisfy the relation iteratively.

8.2 Verifying the defining identity componentwise

To prove that a family \(s_\bullet\) is a chain homotopy, it suffices to check the defining equation in each degree: \[ g_n - f_n = d^D_{n+1}s_n + s_{n-1}d^C_n. \] This componentwise verification is often straightforward when complexes are truncated, or when one can compute images and kernels explicitly.

8.3 Using chain homotopies to simplify calculations

Chain homotopies allow replacements of maps by simpler ones without changing homological consequences. For example, one might modify a chain map by a null-homotopic term to arrange convenient formulas, triangular matrices, or vanishing components. Because induced maps on homology are unchanged, the simplification does not affect the final homological result.

8.4 Typical pitfalls with indices and signs

Common sources of error include:

  • shifting indices in the homotopy maps \(s_n:C_n\to D_{n+1}\),
  • mixing conventions between chain and cochain complexes,
  • using an inconsistent sign in the cone differential or in the homotopy equation.

A reliable approach is to test the formulas on a low-degree example where only a few terms survive, then generalize once the sign and indexing patterns are confirmed.