1 Motivation and background

Derived categories were introduced to organize homological information in a way that is insensitive to many choices made when working with resolutions and chain-level models. Instead of comparing algebraic objects only through exact sequences, the derived perspective treats chain complexes as primary and identifies those with the same homology. This approach has become a standard language in areas where cohomological methods are central.

1.1 Homological algebra

Homological algebra studies algebraic structures by assigning sequences of groups or modules to them and then measuring failure of exactness. It developed from attempts to understand extensions, resolutions, and cohomology theories in a systematic way. The derived category refines this viewpoint by allowing one to work directly with complexes while retaining the ability to extract classical invariants.

1.2 Chain complexes

A chain complex is a sequence of objects connected by morphisms whose successive compositions vanish. Such complexes encode both algebraic data and the relations among that data. They are the basic objects from which derived categories are built.

1.2.1 Homology and cohomology

The homology of a chain complex measures the extent to which the differential fails to make the complex exact. Cohomology is the dual formulation, often obtained from cochain complexes. These invariants are stable under many standard constructions and are the principal quantities preserved by quasi-isomorphisms.

1.2.2 Quasi-isomorphisms

A quasi-isomorphism is a morphism of complexes that induces isomorphisms on all homology groups. In many settings, such maps should be regarded as equivalences because they preserve the essential cohomological content. The derived category is obtained by formally treating quasi-isomorphisms as invertible.

1.3 From abelian categories to derived constructions

An abelian category provides a setting in which kernels, cokernels, and exact sequences behave well. Derived constructions begin with the category of complexes in such a category and then localize at quasi-isomorphisms. This makes it possible to compare objects not merely up to isomorphism but up to homological equivalence.

2 Construction of the derived category

The derived category is constructed from chain complexes by forcing quasi-isomorphisms to become isomorphisms. This process preserves enough structure to support homological calculations while eliminating distinctions that are invisible to cohomology. The resulting category is larger and more flexible than the original abelian category.

2.1 The category of chain complexes

Given an abelian category, one can form a category whose objects are chain complexes and whose morphisms are chain maps. This category reflects the differential structure explicitly. However, chain maps that induce the same effect on homology may still differ significantly at the level of complexes, so a further identification is needed.

2.2 Localization at quasi-isomorphisms

Localization is the formal process of adjoining inverses to a chosen class of morphisms. In the derived setting, the morphisms to be inverted are precisely the quasi-isomorphisms. The localized category captures homological equivalence rather than strict chain-level equality.

2.2.1 Formal inversion of morphisms

To invert a morphism formally means to create a new category in which that morphism behaves as an isomorphism, even if no inverse existed originally. This requires altering the morphism sets in a controlled way. In the derived category, this operation is applied to all quasi-isomorphisms at once.

2.2.2 Universal properties

The derived category is characterized by a universal property: any functor from complexes that sends quasi-isomorphisms to isomorphisms factors uniquely through it. This makes it a canonical construction. Universal properties are especially useful because they determine the derived category up to unique equivalence.

2.3 Roofs and fractions

Morphisms in the derived category are often represented by diagrams called roofs or fractions. A typical roof consists of a complex mapping by a quasi-isomorphism to an intermediate object, which then maps to the target. This representation provides a practical way to describe maps after localization.

2.4 Triangulated structure

Derived categories carry an additional structure called a triangulated structure. It packages exactness information into distinguished triangles rather than short exact sequences. This framework is well suited to tracking how complexes fit together and how homological data changes under mapping cones and shifts.

3 Fundamental types of derived categories

Derived categories come in several forms depending on boundedness conditions and the class of complexes allowed. These variants are chosen to suit particular problems and to control technical issues such as convergence and completeness. The most common versions differ by how far complexes may extend in either direction.

3.1 Bounded derived categories

A bounded derived category is built from complexes that are nonzero only in finitely many degrees. This setting is often manageable and is frequently used in algebraic geometry and representation theory. It captures finite-length homological information in a particularly compact form.

3.2 Unbounded derived categories

The unbounded derived category allows complexes that extend indefinitely in both positive and negative directions. It is more flexible and better suited to modern homological methods. Because it includes a broader range of objects, it requires more care in definitions and constructions.

3.3 Derived categories of bounded complexes

Sometimes one begins with the category of bounded complexes before passing to localization. This intermediate step is useful when the underlying abelian category has finiteness properties that make boundedness natural. It can simplify computations and clarify the relationship with classical homological algebra.

3.4 Variants for different exactness conditions

Different exactness assumptions lead to different derived frameworks. For example, one may work with bounded below or bounded above complexes, or impose projective or injective hypotheses on resolutions. These variants adapt the derived category to the technical needs of a given theory.

4 Derived functors

Derived functors extend ordinary functors to settings where exactness fails. They are among the most important reasons derived categories were developed. By replacing objects with suitable resolutions, one can define functors that retain homological information otherwise lost in ordinary constructions.

4.1 Right derived functors

Right derived functors are used for left exact functors, such as global sections or Hom. They are computed using injective resolutions or other appropriate replacements. The derived category provides a natural environment for defining and interpreting them.

4.2 Left derived functors

Left derived functors extend right exact functors, such as tensor product. They are typically computed using projective or flat resolutions. In derived language, they express the corrected behavior of the original functor on complexes.

4.3 Ext and Tor

Ext and Tor are classical derived functors that measure extensions and torsion phenomena. Ext is associated with Hom-like constructions, while Tor arises from tensor products. Both are central examples of how derived methods recover subtle information about modules and sheaves.

4.3.1 Computation via resolutions

These functors are commonly computed by replacing objects with projective, injective, or flat resolutions. The choice of resolution does not affect the final result up to canonical isomorphism. Derived categories formalize this independence and make the computation conceptual rather than ad hoc.

4.3.2 Relationship to homology

Ext and Tor can be interpreted as homology groups of suitable complexes. This relationship illustrates how derived functors generalize ordinary cohomological invariants. It also shows why homological algebra naturally leads to derived categories.

4.4 Derived pushforward and pullback

In geometric contexts, pushforward and pullback functors often fail to preserve exactness. Their derived versions correct this defect and are indispensable in sheaf theory and algebraic geometry. They allow one to compare spaces through the cohomology of objects defined on them.

5 Triangulated category aspects

The derived category is a standard example of a triangulated category. Its triangulated structure organizes exactness data through triangles and shift operations. This perspective is especially useful when dealing with long exact sequences and mapping cones.

5.1 Distinguished triangles

Distinguished triangles are the analogues of short exact sequences in the derived setting. They relate three objects and encode how one object is built from another. Many fundamental constructions in the derived category are expressed through such triangles.

5.2 Shift and suspension functors

The shift functor moves a complex one degree and changes the sign convention of the differential in the appropriate way. It is a basic symmetry of the derived category. The suspension viewpoint emphasizes that shifting changes the grading while preserving the underlying homological structure.

5.3 Long exact sequences

From a distinguished triangle, one obtains a long exact sequence in cohomology. This is a key mechanism for extracting numerical information from derived data. Long exact sequences remain one of the most effective tools for computations.

5.4 Cones and mapping cones

The mapping cone of a morphism measures how far that morphism is from being a quasi-isomorphism. It produces an object that fits into a distinguished triangle with the source and target. Cones are fundamental in defining triangles and understanding homotopical behavior.

6 t-structures and hearts

A t-structure adds an additional layer of organization to a triangulated category. It separates objects into positive and negative parts and makes it possible to recover an abelian category inside the derived setting. This bridge between triangulated and abelian worlds is one of the most powerful features of derived theory.

6.1 Standard t-structure

The standard t-structure on a derived category is defined by vanishing conditions on cohomology in positive or negative degrees. It is the most familiar example and serves as the model for many others. Its truncation functors are closely related to ordinary cohomological truncation.

6.2 Heart of a t-structure

The heart is the subcategory consisting of objects concentrated in degree zero with respect to the t-structure. It is abelian, even though the ambient category is triangulated. This allows one to recover familiar algebraic objects from a derived framework.

6.3 Cohomological truncation

Truncation functors cut off complexes above or below a given degree while preserving enough structure to induce exact triangles. They are essential in the construction of t-structures. Through truncation, one can build objects degree by degree and analyze their cohomology systematically.

Perverse sheaves are objects in a derived category defined by a nonstandard t-structure. They play an important role in geometry and topology because they encode refined local and global data. Their construction illustrates how changing the t-structure changes the resulting heart and hence the associated abelian theory.

7 Enhancements and higher categorical viewpoints

Derived categories admit richer models that retain more information than the triangulated category alone. These enhancements are important because triangulated categories often forget higher-order homotopies. Modern approaches use differential graded, model categorical, or infinity-categorical frameworks to recover that information.

7.1 DG categories

A differential graded category enriches morphisms to chain complexes rather than mere sets. It provides a natural enhancement of derived categories and often retains finer structure. Many derived constructions are easiest to formulate in this language.

7.2 Stable infinity-categories

Stable infinity-categories offer a higher-categorical framework in which exactness and homotopy coherence are built in. They generalize derived categories while avoiding some limitations of triangulated categories. This setting has become central in contemporary homotopical algebra.

7.3 Model category approaches

Model categories provide a formalism for homotopy theory using cofibrations, fibrations, and weak equivalences. Derived categories can be obtained from suitable model structures on chain complexes. This approach gives concrete tools for constructing and comparing derived objects.

7.4 Enhancements and uniqueness issues

An enhancement records additional homotopical data underlying a derived category. Uniqueness questions ask whether a given triangulated category comes from a single, essentially canonical enhancement. Such issues matter because different enhancements can produce the same triangulated shadow while differing in higher structure.

8 Applications

Derived categories are widely used because they unify many constructions that arise in geometry, algebra, and topology. They provide a common framework for comparing objects through cohomological methods. Their flexibility has made them indispensable in modern mathematical practice.

8.1 Algebraic geometry

In algebraic geometry, derived categories are used to study sheaves, morphisms of varieties, and cohomological invariants. They help express geometric information in a form that is stable under many natural operations. This has led to deep connections between geometry and homological algebra.

8.1.1 Derived categories of coherent sheaves

The derived category of coherent sheaves on a scheme or variety is one of the most important examples. It records information about coherent sheaves together with their extensions and higher cohomology. This category often reflects subtle geometric properties of the underlying space.

8.1.2 Derived equivalences

Two geometric objects are derived equivalent if their derived categories are equivalent as triangulated categories. Such an equivalence can reveal surprising similarities between spaces that are not obviously alike. Derived equivalence has become a major theme in modern geometry.

8.2 Representation theory

Representation theory uses derived categories to study modules over algebras and the relationships among them. Derived methods clarify extension phenomena and the behavior of homological invariants under equivalence. They also provide a language for comparing different algebraic presentations.

8.2.1 Derived equivalence of algebras

Two algebras are derived equivalent when their derived categories of modules are equivalent. This relation is often stronger than classical Morita equivalence but still preserves many important invariants. It is a powerful tool for classifying algebraic structures.

8.2.2 Tilting theory

Tilting theory produces derived equivalences from special objects or modules with strong homological properties. A tilting object can generate an entire derived category and induce an equivalence with the derived category of another algebra. This makes tilting a central technique in representation theory.

8.3 Topology and geometry

Derived categories also appear in topology, where they organize cohomological constructions and sheaf-theoretic methods. They help describe global invariants in terms of local data. In geometry, they provide a flexible bridge between algebraic and topological viewpoints.

8.3.1 Spectral sequences

Spectral sequences are computational tools that relate successive approximations to homological invariants. Derived categories often provide the natural setting in which these sequences arise and converge. They are especially useful for tracking how complex filtrations behave.

8.3.2 Sheaf cohomology

Sheaf cohomology is naturally expressed using derived functors of global sections. The derived category makes this construction systematic and conceptually transparent. It also unifies sheaf-theoretic cohomology with other cohomological theories.

Derived categories are part of a broader landscape of categorical and homological ideas. They are closely connected to triangulated categories, abelian categories, and homotopy theory. Many modern developments in geometry and algebra are built on these relationships.

9.1 Triangulated categories

A triangulated category is an abstract category equipped with a shift functor and distinguished triangles. Derived categories are the prototypical examples. However, not every triangulated category comes from a derived category, so the notion is broader.

9.2 Abelian categories

An abelian category is a setting in which exact sequences and homological constructions behave well. Derived categories are built from complexes in abelian categories and then localized. The heart of a t-structure often recovers an abelian category from a derived one.

9.3 Homotopy categories

The homotopy category identifies chain maps that differ by homotopy. It is an intermediate quotient between the category of complexes and the derived category. While useful, it does not yet invert all quasi-isomorphisms.

9.4 Derived algebraic geometry

Derived algebraic geometry extends classical algebraic geometry by incorporating derived and homotopical methods into the geometry of spaces and schemes. It uses derived categories and related higher-categorical structures to encode intersections, deformation theory, and infinitesimal behavior. This field illustrates the broad reach of derived ideas beyond ordinary cohomology.