1 Background and Motivation
1.1 Why “triangles” instead of exact sequences?
Classical homological algebra studies how modules or chain complexes fit together using short exact sequences. In many settings—especially after passing to derived or stable contexts—short exact sequences are either not available in a strict form or do not behave well under natural constructions such as homotopy, localization, or taking homological invariants.
A triangulated category replaces the need for literal short exact sequences by a weaker but highly structured surrogate: distinguished triangles. These triangles record how one object is “assembled” from another via a mapping process, while encoding the expected exactness properties only in the sense of how homological functors respond to the triangle.
1.2 The shift (suspension) operation
A key feature of triangulated categories is an autoequivalence, commonly denoted the shift or suspension functor. Conceptually, the shift accounts for the degree change that appears when translating exact-sequence behavior into statements about complexes. Without such a mechanism, the triangle formalism cannot consistently mimic the boundary maps and degree shifts that drive homological algebra.
1.3 Stable homological behavior and localization
Many constructions in algebraic topology and representation theory naturally force one to ignore certain morphisms or to identify objects up to a stable notion of equivalence. Localization can collapse data, and stabilization can systematically discard “inessential” parts while preserving homological information. Triangulated categories package these effects so that, even after stabilization or localization, there remains a coherent notion of exactness.
1.4 Relation to derived categories
Derived categories are the prototypical source of triangulated structures. When one starts from chain complexes and localizes at quasi-isomorphisms, the resulting homotopy-theoretic objects admit distinguished triangles corresponding to mapping cones of morphisms. This provides a systematic reason that triangles, rather than short exact sequences, govern the behavior of derived invariants.
2 Definition and Axioms
2.1 Basic data: objects, morphisms, and triangulation
A triangulated category consists of a class of objects and morphisms arranged in an additive (in particular, zero morphisms and biproducts exist) category, together with:
- an autoequivalence (the shift functor), and
- a specified class of distinguished triangles.
These distinguished triangles are triples of composable morphisms that serve as the analog of exact short sequences.
2.2 Distinguished triangles
A typical distinguished triangle has the form \[ A \to B \to C \to A[1], \] where \(A[1]\) is the shift of \(A\). The defining property is not merely the shape of the morphisms, but that the category’s axioms guarantee compatibility with rotation, mapping behavior, and a higher-order “gluing” rule.
In practice, distinguished triangles frequently arise from mapping cones: a morphism \(f:A\to B\) produces a cone object \(C\) so that \(A\), \(B\), and \(C\) form a distinguished triangle.
2.3 The shift functor and its functoriality
The shift functor \( [1] \) (or another notation) is an additive autoequivalence. Its role is twofold:
- it supplies the “degree shift” that completes a triangle, and
- it is required to interact functorially with the structure so that distinguished triangles can be transformed consistently.
Thus, if a triangle is distinguished, applying the shift should yield another distinguished triangle in a prescribed way, respecting morphisms.
2.4 Triangulated-category axioms (TR1–TR4)
2.4.1 Rotation of distinguished triangles
One axiom ensures that if a triangle is distinguished, then it remains distinguished after a cyclic reordering (“rotation”), with appropriate shifts on the involved objects. This expresses that exactness should not depend on which vertex is regarded as “the beginning.”
In effect, the triangle formalism is invariant under the expected homological rotation behavior: the same information can be presented from different perspectives.
2.4.2 Morphisms of triangles and compatibility
Morphisms of distinguished triangles are triples of compatible morphisms between the corresponding vertices. The axioms require that if certain squares commute and the source and/or target triangles are distinguished, then the induced morphisms behave appropriately and extension-like constructions exist.
This provides a stability principle: triangles can be compared and transported through morphisms in a controlled way.
2.4.3 The octahedral structure (TR4) at a high level
A deeper axiom captures how compositions of morphisms produce a web of triangles. At a conceptual level, it asserts that given composable maps \(A\to B\to C\), one can form distinguished triangles related by a pattern analogous to an “octahedron,” encoding how cones of morphisms fit together.
This is the mechanism behind the ability to construct higher homological relationships and is central for developing long exact sequences and spectral sequences.
3 Functors and Natural Constructions
3.1 Exact (triangulated) functors
A triangulated functor is one that preserves distinguished triangles and commutes with the shift functor up to specified compatibility. Such functors translate exactness information between triangulated categories without destroying the triangle structure that encodes homological behavior.
When \(F\) is triangulated, applying \(F\) to a distinguished triangle yields a distinguished triangle in the target category.
3.2 Adjoint functors in the triangulated setting
Adjoint functors interact strongly with triangulated structures. Under suitable hypotheses, a left or right adjoint of a triangulated functor can itself be triangulated, or at least can preserve distinguished triangles in a compatible manner.
This matters because adjunctions often arise from geometric or algebraic constructions (such as extension and restriction of scalars) and the triangulated framework ensures that derived or stable information transfers correctly.
3.3 Compatibility with the shift functor
Exactness in the triangulated sense is inseparable from how a functor treats the shift. If a functor does not respect \( [1] \), it may fail to preserve the “completion” of triangles, thereby breaking the exactness-like correspondence.
Accordingly, triangulated functors are required to come with coherence data specifying how they interact with the suspension functor.
3.4 Functoriality of distinguished triangles
Even when the definition emphasizes preservation of distinguished triangles, many constructions rely on stronger functorial behavior: cones and mapping constructions should be natural enough that triangle formation is compatible with morphism composition.
In well-behaved settings (notably derived and homotopy categories), these properties are realized by explicit cone constructions in the underlying model, making the triangle operations coherent.
4 Subcategories and Localization
4.1 Thick subcategories and closure properties
A thick subcategory is a full triangulated subcategory closed under taking direct summands. Closure under distinguished triangles and shifts allows thick subcategories to behave like robust “regions” of the category that remain stable under the homological operations encoded by the axioms.
This concept is important for organizing objects by homological complexity and for defining quotients that remove chosen parts of the category.
4.2 Verdier quotients
A Verdier quotient generalizes the idea of forming a quotient category by identifying morphisms that become isomorphisms after modding out a thick subcategory. The result is again triangulated, provided the quotient is formed using an appropriate triangulated substructure.
Verdier quotients are a standard tool for localizing categories in homological algebra, especially in derived contexts.
4.3 Localization by a class of morphisms
Localization replaces morphisms from a selected class by isomorphisms, making them invertible. In triangulated settings, such localizations often correspond to killing an associated subcategory of objects or morphisms and then passing to a quotient.
The triangulated structure ensures that the localized category still supports a meaningful triangle-based exactness theory.
4.4 Strategies for constructing quotient categories
Common approaches include:
- selecting a thick subcategory and applying a Verdier quotient,
- using universal properties of localization, and
- constructing the quotient through roofs or derived calculus of fractions in concrete models.
In practice, the main challenge is ensuring that the quotient inherits a well-defined shift functor and that distinguished triangles descend correctly.
5 Cohomological Viewpoints
5.1 Cohomological functors on triangulated categories
A cohomological functor translates distinguished triangles into long exact sequences in an abelian target. More precisely, applying the functor to a distinguished triangle yields a sequence of morphisms that behaves like exactness across degrees.
This is one of the main ways triangulated categories connect abstract triangle data with computable invariants.
5.2 Homological functors and long exact sequences
Similarly, homological functors produce long exact sequences, often indexed by shifts. The distinction between “cohomological” and “homological” depends on whether the functor reads triangles in covariant or contravariant ways and how degree shifts are assigned.
The long exact sequences arise formally from the axioms that control rotation and compatibility of distinguished triangles.
5.3 Representability and Yoneda-type principles
A central organizing principle is the relationship between morphisms in a triangulated category and the behavior of functors built from them. Representability results—analogous to the Yoneda lemma but adapted to the triangulated context—allow one to treat many cohomological functors as arising from hom-sets into or out of shifted objects.
This viewpoint turns triangle structures into universal patterns for constructing and reasoning about invariants.
5.4 Stability of exactness under shifts
Since triangles include a shift in their data, cohomological and homological exactness statements are designed to be stable under repeated suspension. This produces systematic degree-shifting in derived or stable computations and ensures that exactness patterns persist across the grading-like structure.
In effect, the shift functor is not decorative: it is the mechanism that keeps the exactness correspondence coherent.
6 Examples
6.1 Derived categories as canonical examples
In the derived category of an abelian category (or suitable exact category), objects are complexes, and morphisms are considered up to quasi-isomorphism. Distinguished triangles correspond to mapping cones: a morphism of complexes fits into a triangle whose third vertex is the cone.
This construction makes derived categories the standard model for triangulated theory and explains many axioms as formal consequences of cone behavior.
6.2 Homotopy categories
Homotopy categories of chain complexes (or more general objects such as differential graded structures in certain settings) often admit triangulated structures. Here, distinguished triangles arise from cone constructions in the homotopy category, reflecting how homotopies and mapping cones interact.
While homotopy categories can be less localized than derived categories, their triangle structure still organizes morphisms and homological invariants.
6.3 Stable module categories (general description)
In modular representation theory, one studies modules up to projective or injective behavior, producing a “stable” perspective. The resulting stable module category frequently carries a triangulated structure, where triangles reflect short exact sequences modulo projectives and are controlled by syzygies.
Although the details vary by setup, the common theme is that stabilization turns extension data into triangle data.
6.4 Triangulations coming from exact categories
Exact categories—where short exact sequences exist but may not be abelian—can also yield triangulated categories through derived or stable constructions. By taking appropriate quotient or localization procedures and then passing to homotopy-theoretic objects, one can build triangulated structures whose distinguished triangles reflect the given exactness.
This shows that triangulated categories are not limited to abelian or purely topological contexts.
7 Properties and Structural Results
7.1 Splitting criteria and idempotent completion
Not all additive categories are automatically complete with respect to splitting idempotents. Many triangulated constructions assume or enforce idempotent completeness (also called Karoubi closure), ensuring that when an object decomposes via an idempotent endomorphism, the decomposition exists within the category.
Splitting criteria help determine when such decompositions occur, and they affect how thick subcategories behave.
7.2 Compact objects and generation (overview)
Compact objects are those for which hom-functors commute with directed colimits. In many triangulated categories arising from geometry or algebra, compact objects provide a manageable “basis” that generates the whole category under colimits and triangulated operations.
Generation results describe when every object can be built from compact ones using shifts, cones, and colimits, enabling computational control in large categories.
7.3 Mapping cones and triangulated behavior
Mapping cones are the concrete device behind most distinguished triangles. Starting from a morphism \(f\), the cone produces an object that fits into a canonical triangle. The mapping cone construction supports:
- functorial behavior up to coherent equivalence,
- stability under composition (captured by the octahedral axiom), and
- systematic translation between algebraic and homological data.
In many models, the triangulated structure is defined so that cones generate all distinguished triangles.
7.4 t-structures and hearts (if applicable)
Some triangulated categories admit a t-structure, which provides an additional filtration-like structure separating “nonnegative” and “nonpositive” parts. The associated heart is an abelian category, and truncation functors allow one to relate derived or stable information back to abelian invariants.
When present, t-structures refine the triangle-based exactness into more classical exact sequences within the heart.
8 Advanced Topics (Conceptual Overview)
8.1 Monoidal and tensor triangulated structures (outline)
Triangulated categories can carry monoidal structures compatible with the triangulation. A tensor triangulated category has a tensor product that interacts with distinguished triangles in a controlled way, often making the tensor product exact in each variable.
This framework supports studying how “tensoring” affects homological phenomena and enables spectral tools tied to monoidal behavior.
8.2 Equivalences and derived equivalence (high level)
Two triangulated categories may be equivalent as triangulated categories, meaning there exists a functor preserving shifts and distinguished triangles with an inverse up to natural isomorphism. Such equivalences underlie derived equivalence ideas: algebraic structures can have equivalent derived or stable categories even if they differ at the level of abelian categories or rings.
These correspondences help transfer invariants between seemingly different contexts.
8.3 Spectral sequences from triangulated data
Spectral sequences often originate from filtrations or from repeated use of long exact sequences. Triangulated categories supply a conceptual source for many such sequences: distinguished triangles and their associated exactness yield successive approximations to homological invariants.
At an abstract level, one can view spectral sequences as organized bookkeeping of how information propagates through chains of triangles.
8.4 Bridgeland-style stability conditions (overview)
Stability conditions refine triangulated categories by assigning a geometric notion of “phases” to objects, typically using a central charge and a filtration-like structure compatible with the triangulated axioms. When defined, they allow classification of objects by stability and can lead to invariants counting stable objects.
This topic is an active area where triangulated structure interfaces with geometry and wall-crossing phenomena.