Homological algebra studies how algebraic structure fails to be exact—and turns that failure into information. Its central objects are chain complexes, homology groups, exact sequences, resolutions and derived functors such as Ext and Tor.
The subject grew from topology, where homology measures holes, but its machinery now runs through algebraic geometry, representation theory, number theory, category theory and modern geometry. The key idea is remarkably portable: arrange objects and maps into a sequence, identify what maps in, identify what maps out, and measure the mismatch.
Series route: Mathematics Learning Hub → How Mathematics Works → Homological Algebra. Useful foundations include Abstract Algebra, Category Theory, Commutative Algebra and Algebraic Topology.
1. Exactness is an information-balance condition
A sequence A→B→C is exact at B when the image of the first map equals the kernel of the second.
Everything arriving at B is exactly what becomes invisible when B maps onward.
2. Short exact sequences encode extensions
A short exact sequence 0→A→B→C→0 says A embeds into B, C is the corresponding quotient, and B is built as an extension of C by A.
The middle object may contain twisting information not visible from A and C separately.
3. Split exact sequences are untwisted
If the surjection B→C admits a compatible section, the sequence splits and B is isomorphic to A⊕C in the usual module or abelian-category setting.
Failure to split is therefore meaningful structure.
4. Chain complexes organise repeated maps
A chain complex is a sequence …→C₂→C₁→C₀→… with successive maps d satisfying d∘d=0.
The condition means every boundary is automatically a cycle.
5. Cycles are elements killed by the differential
At degree n, the cycle group Z_n is ker(d_n).
These are elements whose outgoing differential vanishes.
6. Boundaries are elements arriving from one degree higher
The boundary group B_n is im(d_{n+1}).
Because d²=0, every boundary lies inside the cycle group.
7. Homology measures cycles not explained by boundaries
The nth homology object is H_n=Z_n/B_n.
Exactness at C_n is equivalent to H_n=0.
Homology is therefore a precise measurement of failure of exactness.
8. Cochain complexes reverse the direction
A cochain complex is written C⁰→C¹→C²→… with coboundary maps increasing degree.
Its cohomology H^n is kernel modulo image in the same structural pattern.
9. Chain maps preserve differential structure
A chain map between complexes commutes with differentials.
It therefore sends cycles to cycles and boundaries to boundaries, inducing maps on homology.
10. Chain homotopies identify equivalent maps
Two chain maps can differ at the complex level while inducing the same map on homology if they are chain homotopic.
This mirrors topological homotopy: different detailed maps may carry the same homological information.
11. Categories make homological algebra portable
Homological algebra is naturally formulated in abelian categories, where kernels, cokernels, direct sums and exact sequences behave coherently.
Modules over a ring and abelian groups are standard examples.
12. Functors can preserve or destroy exactness
A functor is exact if it sends short exact sequences to short exact sequences.
Left-exact functors preserve exactness at the beginning of a sequence; right-exact functors preserve it at the end.
13. Derived functors measure failure of exactness
If a functor is not exact, its derived functors quantify what is lost.
This is one of the defining moves of homological algebra: failure is promoted from an obstacle into a new invariant.
14. Projective modules lift through surjections
A projective module P has the property that maps P→B can be lifted through suitable surjections A→B.
Free modules are projective, but projective modules need not be free over arbitrary rings.
15. Injective modules extend across embeddings
An injective module I has the dual extension property: maps into I extend across monomorphisms under the appropriate category setting.
Projectives support left-derived constructions; injectives support right-derived constructions.
16. Resolutions replace difficult objects by tractable ones
A projective resolution …→P₂→P₁→P₀→M→0 replaces M by an exact complex of projective objects.
Applying a functor to the resolution creates a new complex whose homology measures the functor’s failure to remain exact.
17. Resolutions are not unique, but derived invariants are
Different projective resolutions of the same module can look very different.
Comparison theorems show they are chain-homotopy equivalent in the relevant sense, so the resulting derived functors are canonically well defined up to natural isomorphism.
18. Tor measures failure of tensor product to be exact
Tensoring with a fixed module is generally right exact but not left exact.
The groups Tor_n^R(M,N) are its left-derived functors and measure the lost exactness.
19. Tor detects torsion interactions
For example, over Z, Tor_1^Z(Z/mZ,Z/nZ) records common torsion and is isomorphic to Z/gcd(m,n)Z.
The name Tor is therefore not accidental.
20. Ext measures extension data
The functor Hom_R(M,−) is left exact in its second argument.
Its right-derived functors Ext_R^n(M,N) measure higher extension information.
21. Ext¹ classifies short exact extensions
Under standard module settings, Ext¹_R(M,N) classifies equivalence classes of short exact sequences 0→N→E→M→0.
The zero class corresponds to split extensions.
22. Long exact sequences propagate information
A short exact sequence of complexes induces a long exact sequence in homology.
Connecting homomorphisms bridge neighbouring degrees, allowing unknown homology to be constrained from known pieces.
23. The snake lemma builds connecting maps
The snake lemma starts from a commutative diagram with exact rows and constructs an exact sequence linking kernels and cokernels.
It is a systematic machine for extracting hidden information from a diagram.
24. The five lemma transfers isomorphisms through exact diagrams
In a suitable commutative diagram of exact sequences, if surrounding vertical maps satisfy the required injectivity, surjectivity or isomorphism conditions, the middle map is forced to be an isomorphism.
Exactness turns local map information into a global conclusion.
25. Double complexes organise two interacting differentials
A double complex arranges objects in a grid with horizontal and vertical differentials.
Total complexes combine both directions and become the natural source of spectral sequences.
26. Spectral sequences calculate complicated homology in stages
A spectral sequence consists of successive pages E_r equipped with differentials, where each page’s homology produces the next.
Under convergence conditions, the stable information reconstructs graded pieces of the target object.
27. Spectral sequences are filters, not magic tables
Their usefulness comes from replacing one difficult calculation by a sequence of simpler approximations tied to a filtration or double complex.
Different spectral sequences are built for different decompositions of the same problem.
28. Derived categories treat quasi-isomorphisms as equivalences
A quasi-isomorphism is a chain map inducing isomorphisms on all homology groups.
The derived category formally inverts quasi-isomorphisms so complexes with the same homological information become equivalent for derived purposes.
29. Derived categories package resolutions conceptually
Instead of repeatedly choosing projective or injective resolutions by hand, derived categories provide a setting where derived functors can be understood as functors acting on complexes up to quasi-isomorphism.
This makes the machinery more structural and categorical.
30. Homological algebra powers algebraic geometry
Sheaf cohomology, derived pushforwards, Ext sheaves and spectral sequences are central tools for studying varieties and schemes.
Continue through Algebraic Geometry for the geometric side.
31. Homological algebra powers representation theory
Extensions classify ways representations fit together, projective resolutions compute group or algebra cohomology, and derived categories organise representation-theoretic equivalences.
Semisimple categories are special partly because higher extension data can vanish or simplify dramatically.
32. Homological algebra powers topology
Singular homology itself is built from chain complexes. Universal coefficient theorems and Künneth formulas involve Ext and Tor correction terms.
The abstract algebraic machinery therefore returns to the topological problems that originally motivated much of it.
33. A worked mechanism: homology of a short complex
Consider 0→Z –×2→ Z→0, with the first Z in degree 1 and the second in degree 0.
- The degree-1 differential sends n to 2n.
- Its kernel is 0, so H₁=0.
- At degree 0, every integer is a cycle because the outgoing differential is zero.
- The boundaries are the even integers 2Z.
- Therefore H₀=Z/2Z.
The homology records exactly what remains after the image of multiplication by 2 is quotiented out.
34. A worked mechanism: why Tor appears
Start with the exact sequence 0→Z –×m→ Z→Z/mZ→0 and tensor with Z/nZ.
- Tensor product preserves the right-hand exactness but may lose injectivity at the left.
- The lost kernel is measured by Tor₁.
- The result is controlled by elements of Z/nZ annihilated by multiplication by m.
- This produces Z/gcd(m,n)Z.
Tor is literally the algebraic record of how exactness failed after tensoring.
35. Common homological-algebra failure modes
- Exact=zero-map confusion: exactness concerns image equalling kernel, not maps individually being zero.
- Cycles=boundaries confusion: homology exists precisely because they can differ.
- Resolution dependence: assuming derived invariants depend on the chosen projective or injective resolution.
- Projective=free confusion: true over some rings, false in general.
- Ext/Tor symbol pushing: computing derived functors without identifying which failure of exactness they measure.
- Spectral-sequence certainty: ignoring convergence and extension problems between graded pieces.
- Derived-category over-abstraction: forgetting the concrete complexes and geometric or algebraic question being encoded.
36. Homological algebra as a mathematical machine
Objects and Maps → Complex → Kernel/Image → Homology → Resolution → Derived Functor → Long Exact/Spectral Sequence → Structural Invariant → Return to Algebra, Geometry or Topology.
37. What mastery looks like
- read exact sequences as information-flow statements;
- compute kernels, images, cycles, boundaries and homology;
- work with chain maps and homotopies;
- understand projective and injective objects through lifting and extension;
- build and compare resolutions;
- interpret Tor and Ext as derived measurements of non-exactness;
- use long exact sequences to propagate partial information;
- understand spectral sequences as staged calculations;
- recognise why derived categories invert quasi-isomorphisms;
- return abstract invariants to their original algebraic, geometric or topological problem.
38. Further reading
39. Conclusion
Homological algebra works by treating failure of exactness as a measurable signal. Chain complexes organise transformations. Homology records what survives between kernel and image. Resolutions replace difficult objects by tractable ones. Ext and Tor measure how functors fail to preserve exactness. Spectral sequences break large calculations into layers.
Category theory tells us how mathematical objects and maps fit together. Homological algebra adds a detector for what is lost, obstructed or twisted when those maps fail to fit exactly.
How Mathematics Works | Batch 14
- How Mathematics Works | Commutative Algebra
- How Mathematics Works | Algebraic Number Theory
- How Mathematics Works | Analytic Number Theory
- Homological Algebra — this article
Return to the Mathematics Learning Hub or the How Mathematics Works root.