1 Definition and exactness criteria

1.1 Abelian categories and exact sequences

Exactness of a functor is usually formulated inside abelian categories, where kernels and cokernels exist and where monomorphisms and epimorphisms align with categorical notions of injectivity and surjectivity. A short exact sequence in an abelian category has the form \[ 0 \to A \xrightarrow{f} B \xrightarrow{g} C \to 0 \] and encodes that \(f\) identifies \(A\) with the kernel of \(g\), while \(g\) identifies \(C\) with the cokernel of \(f\).

In such settings, exact sequences are not merely formal patterns: they are equivalent to statements about images, kernels, and cokernels. This makes it possible to define exact functors by requiring these structural ingredients to be preserved.

1.2 Exactness in terms of kernels and cokernels

Let \(F:\mathcal{A}\to\mathcal{B}\) be a functor between abelian categories. The functor is called left exact if it preserves kernels, meaning that for every morphism \(u:X\to Y\) in \(\mathcal{A}\), the induced comparison map gives an isomorphism \[ F(\ker u)\cong \ker(Fu). \] Similarly, \(F\) is right exact if it preserves cokernels, i.e. \[ F(\operatorname{coker} u)\cong \operatorname{coker}(Fu) \] for all \(u\).

A functor is exact (two-sided exact) when it is both left exact and right exact, so it preserves both kernels and cokernels, equivalently preserving the exactness of short exact sequences.

1.3 Exactness preserving short exact sequences

A standard criterion is: a functor \(F\) is exact if and only if it takes every short exact sequence in \(\mathcal{A}\) to a short exact sequence in \(\mathcal{B}\). Concretely, if \[ 0\to A\to B\to C\to 0 \] is short exact, then \[ 0\to F(A)\to F(B)\to F(C)\to 0 \] is short exact as well.

This criterion is often adopted as the definition in abelian contexts because it directly captures “no information loss” in the sense that the kernel–image–cokernel structure remains intact after applying the functor.

1.4 Left, right, and two-sided exact functors

Because preservation of kernels and cokernels can fail independently, exactness is typically stratified:

  • Left exact: exact on sequences of the form

\[ 0 \to A \to B \to C \] (injectivity at the left and kernel preservation).

  • Right exact: exact on sequences of the form

\[ A \to B \to C \to 0 \] (surjectivity at the right and cokernel preservation).

  • Exact: both properties hold, so all short exact sequences are preserved.

This distinction is central in applications, where certain operations naturally preserve only one side of exactness.

2 Homological-algebra viewpoint

2.1 Functorial behavior on long exact sequences

Exact functors preserve short exact sequences, and this has consequences for the behavior of derived constructions. Given a short exact sequence of objects or coefficients, homological algebra often produces long exact sequences of groups or modules. If a functor is exact, then applying it to the short exact data yields corresponding exactness results without requiring derived correction terms.

By contrast, when a functor is only left exact or only right exact, long exact sequences generally come with additional terms or require derived functors to restore exactness.

2.1.1 Derived functors and loss of exactness

A left exact functor may fail to be right exact, and the “defect” is measured by its right derived functors. Dually, a right exact functor’s failure to be left exact is measured by its left derived functors.

In practice, if \(F\) is not exact, one expects sequences that are exact before applying \(F\) to become non-exact after applying it. Derived functors correct this by producing a new functorial apparatus whose output fits into exact sequences again.

2.2 Relation to projective and injective objects

Abelian categories come equipped with notions of projective and injective objects, and these determine when certain functors preserve exactness without derivation.

2.2.1 Preserving projectives

A common sufficient condition is: if a right exact functor sends projective objects to projective objects, then it behaves more predictably on resolutions built from projectives. In such circumstances, the functor can be computed by applying it to projective resolutions, and the resulting derived functors simplify.

However, preserving projectives alone does not guarantee exactness; it primarily ensures good behavior on specific classes of objects used in homological computations.

2.2.2 Preserving injectives

Dually, a left exact functor that takes injectives to injectives facilitates computation via injective resolutions. Again, injective preservation supports the derived framework, but it is not synonymous with two-sided exactness.

2.3 Ext and Tor interpretations (module-theoretic examples)

In module categories, exactness defects are encoded by classical invariants.

2.3.1 Right exactness via Tor

For modules over a ring \(R\), the tensor product functor \(-\otimes_R M\) is right exact in its first variable. Its left derived functors are the Tor groups: \[ \operatorname{Tor}_i^R(-,M). \] When \(\operatorname{Tor}_1^R(-,M)=0\) for relevant inputs, one obtains stronger forms of exactness. In particular, vanishing of \(\operatorname{Tor}_1\) is closely tied to the tensor functor turning certain short exact sequences into exact sequences on the left.

2.3.2 Left exactness via Ext

Similarly, \(\operatorname{Hom}_R(M,-)\) is left exact in its second variable. Its right derived functors are the Ext groups: \[ \operatorname{Ext}_R^i(M,-). \] Vanishing of \(\operatorname{Ext}^1\) in a range of cases indicates that the relevant extensions split in a way compatible with exactness properties of the Hom functor.

These interpretations provide concrete diagnostics: exactness is not only a categorical property but also something detectable by computable invariants.

3 Standard examples and constructions

3.1 Tensoring with a module (and conditions for exactness)

Given a fixed right \(R\)-module \(M\), the functor \(-\otimes_R M\) is right exact. It is exact precisely when the derived obstruction disappears, which can be characterized using Tor:

  • If \(\operatorname{Tor}_1^R(-,M)=0\) for all inputs, then \(-\otimes_R M\) is exact.
  • A common sufficient condition is that \(M\) is flat; then tensoring preserves short exact sequences.

In many applied contexts—especially those involving linear algebra and base change—flatness is the mechanism ensuring “no information loss” under tensor-based transformations.

3.2 Hom-functors (covariant/contravariant exactness)

For a fixed module \(M\), the Hom functor \(\operatorname{Hom}_R(M,-)\) is left exact, while \(\operatorname{Hom}_R(-,M)\) is left exact in the second argument but behaves differently in the first because of contravariance. In abelian categories, exactness statements must respect how morphism directions are reversed by contravariant functors.

In practice, the exactness properties of Hom functors are also controlled by Ext:

  • Failure of right exactness corresponds to nonzero \(\operatorname{Ext}^1\).
  • Exactness on short exact sequences can be tested by how Ext behaves with the chosen module \(M\).

Subobject- and quotient-type constructions can be subtle. While inclusion and projection maps induce functors between slice categories or categories of subobjects, preserving exactness depends on how these functors interact with kernels and cokernels.

In general, functors that “forget structure” or change the ambient context may destroy exactness unless additional conditions are imposed (such as exactness of restriction/extension procedures in the relevant abelian setup).

3.4 Change-of-rings and base change functors

Given a ring homomorphism \(R\to S\), one often studies how module categories over \(R\) and \(S\) relate. Base change frequently uses tensor products, and thus inherits right exactness from tensoring. Exactness of the associated functors is influenced by how \(S\) behaves as an \(R\)-module:

  • If \(S\) is flat over \(R\), then tensoring with \(S\) is exact, producing a well-behaved transformation between module categories.
  • When flatness fails, exact sequences may become non-exact after change of rings, and derived functors (Tor/Ext) are used to control the discrepancy.

4 Characterizations and test methods

4.1 Checking exactness on generating short exact sequences

In many abelian categories, not every short exact sequence needs to be checked explicitly to determine exactness. When there are generating families of objects or morphisms, exactness can be verified on a strategically chosen set of short exact sequences.

For instance, if the category is module-based, one can test exactness on sequences built from generators and relations. The guiding idea is that functors are determined by their action on enough morphisms, and exactness can be reduced to a smaller verification task.

4.2 Exactness via diagram chasing

Exactness preservation often admits proofs by diagram chasing: one takes the image under \(F\) of a short exact sequence and then shows that kernels match, images align, and cokernels match by using universal properties.

Diagram arguments typically proceed by comparing:

  • \(\ker(Fg)\) with \(F(\ker g)\),
  • \(\operatorname{coker}(Ff)\) with \(F(\operatorname{coker} f)\),

and by using the characterization of exactness in abelian categories via kernels and cokernels.

This method is systematic and avoids reliance on heavy machinery, though it can become intricate in more complex categories.

4.3 Preservation of monomorphisms and epimorphisms

A left exact functor preserves monomorphisms because kernels detect injectivity in abelian settings. Likewise, a right exact functor preserves epimorphisms.

However, exactness requires more than preserving monos or epis. A functor might send injective maps to injective maps (or surjective maps to surjective maps) while still failing to preserve kernel–cokernel identifications in the middle. Thus monomorphism/epimorphism preservation is a necessary ingredient but not by itself sufficient for full exactness.

4.4 Faithfulness vs exactness (what each implies)

Faithfulness of a functor means it reflects distinct morphisms: it does not collapse different arrows into one. Exactness refers instead to preserving structural exactness patterns in sequences.

These notions are logically independent:

  • A functor can be faithful but non-exact.
  • A functor can be exact without being faithful.

In applied work, confusing these can lead to incorrect assumptions: faithful functorial behavior ensures information about morphisms, whereas exactness ensures compatibility with homological structure.

5 Exact functors in applied settings

5.1 Module categories as linear-algebraic models

Module categories model many “linear” constructions. When a transformation between module categories is exact, it respects the algebraic analog of “subspace/kernel” and “quotient/cokernel,” ensuring that rank-like and dimension-like invariants behave predictably.

This is why exact functors appear naturally in representations, in computations of invariants, and in any setting where exact sequences encode constraints and compatibility conditions.

5.2 Functorial pipelines in representation theory

Representation-theoretic operations often correspond to functors between categories of modules with additional structure. Exact functors help ensure that:

  • subrepresentations and quotient representations correspond appropriately,
  • extension data is handled correctly when exactness is preserved,
  • derived corrections are unnecessary when exactness holds.

Such behavior streamlines pipelines that translate problems from one representation category to another while maintaining the integrity of short exact decompositions.

5.3 Sheaves, presheaves, and exactness conditions

In sheaf theory, one frequently works with left exactness properties (for example, sections with respect to certain constructions) and then asks when they become exact. Exactness may fail globally even if it holds locally, due to topological and geometric constraints.

Many sheaf-theoretic results are framed by identifying which functors are exact, which are only left or right exact, and which require derived functors to recover global exactness. The categorical perspective clarifies where obstructions originate.

5.4 Stability of invariants under transformations

An exact functor preserves not only short exact sequences but also many invariants derived from homological algebra. When exactness holds, quantities computed from resolutions or from kernel/cokernel constructions often remain stable under the transformation.

This stability is valuable in applied settings where one wants to compare models: if two descriptions are connected by an exact functor, then the homological “signal” encoded by exact sequences is transported without distortion.

6 Functor composition and stability properties

6.1 Composition of exact functors

Exactness is stable under composition. If \(F:\mathcal{A}\to\mathcal{B}\) and \(G:\mathcal{B}\to\mathcal{C}\) are exact, then the composite \(G\circ F:\mathcal{A}\to\mathcal{C}\) is exact.

This follows because applying \(F\) preserves short exact sequences, and then applying \(G\) preserves short exact sequences as well. The combined effect is preservation across the full pipeline.

6.2 Natural transformations and exactness

Natural transformations do not automatically preserve exactness properties; rather, they provide a way to compare functors. If two exact functors are connected by a natural transformation, one can sometimes transfer structural consequences between them, but the transformation itself need not be an exactness-preserving object.

In practice, naturality ensures compatibility of morphism assignments across the category, while exactness is a property checked on sequences. Both aspects often appear together in proofs and constructions.

6.3 Restriction to subcategories and quotient categories

When passing to full subcategories or quotient constructions, exactness can persist or fail depending on whether the new subcategory remains abelian (or inherits an exact structure) and whether kernels/cokernels computed in the ambient category agree with those in the subcategory.

A key technical issue is whether the functor restricts to the subcategory and whether it preserves the relevant exact structure. Without these alignment conditions, exactness can be lost even if the original functor was exact.

6.4 Exactness under equivalence of categories

An equivalence of categories between abelian categories preserves exactness in a strong sense. If \(E:\mathcal{A}\to\mathcal{B}\) is an equivalence and \(\mathcal{A},\mathcal{B}\) are abelian, then exact sequences correspond under \(E\), and exactness of functors transported along the equivalence can be characterized via conjugation with the equivalence and its quasi-inverse.

Thus, exactness behaves like a categorical invariant: it is not an artifact of a particular presentation.

7 Limitations and common pitfalls

7.1 Non-exact functors that are nevertheless useful

Many constructions used in computation are not exact, yet they remain indispensable. Right exact functors (or left exact ones) can still provide correct results in certain degrees, and derived functors quantify what goes wrong.

In applied contexts, it is common to accept non-exactness and then work with correction terms, because this yields computationally manageable frameworks (e.g., using Tor or Ext rather than trying to enforce full exactness).

7.2 Confusing left/right exactness

A frequent source of confusion is assuming that preserving kernels implies full exactness, or that preserving cokernels implies the other side automatically. In general, neither implication holds without additional hypotheses.

The practical remedy is to identify which side is preserved by the construction at hand and then determine whether the missing side can be repaired via conditions like flatness/projectivity or injectivity.

7.3 Exactness failing in non-abelian contexts

Outside abelian categories, notions of exact sequences can be more delicate. In non-abelian settings, kernels and cokernels may not interact with monomorphisms and epimorphisms in the same way, and “exactness of sequences” may not be well-behaved or may require additional categorical structure (such as exact categories, protomodularity, or other frameworks).

Thus, the abelian definition and criteria should not be used blindly in broader contexts.

7.4 Conditions that guarantee exactness

Exactness can be guaranteed by various structural conditions, depending on the functor:

  • preservation of kernels and cokernels,
  • flatness conditions for tensor-type functors,
  • injectivity or projectivity conditions when Hom-type functors are involved,
  • compatibility of functor construction with the abelian exact structure.

When such hypotheses are satisfied, the functor’s behavior on short exact sequences becomes predictable, enabling reliable transport of homological information.

8 Notation, terminology, and references

8.1 Standard notation for exactness

Exactness is commonly denoted by phrases such as “\(F\) is exact,” “\(F\) is left exact,” or “\(F\) is right exact.” Exact sequences are written with \(0\) at both ends and arrows indicating the morphisms whose images and kernels match appropriately.

In module contexts, Tor and Ext are written with subscripts/superscripts indicating homological degree, such as \(\operatorname{Tor}_i^R(-,M)\) and \(\operatorname{Ext}_R^i(M,-)\).

8.2 Equivalent definitions across literature

Different texts may phrase definitions using:

  • preservation of kernels/cokernels,
  • preservation of short exact sequences,
  • exactness of the induced morphisms on certain functor categories,
  • exactness in terms of vanishing of derived functors.

These formulations coincide in abelian categories, but the equivalence may rely on standard assumptions about existence and properties of kernels and cokernels.

A reader typically benefits from familiarity with:

  • basic category theory (functors, natural transformations),
  • abelian categories and their exact sequences,
  • kernels/cokernels and their universal properties,
  • introductory homological algebra concepts like projective/injective objects.

With these, exactness criteria can be interpreted both categorically and homologically.

8.4 Further reading and classic sources

Classic references typically include foundational texts in category theory and homological algebra, where exact functors are introduced alongside abelian categories, derived functors, and module-theoretic examples. Standard references also discuss how exactness interacts with resolutions, Ext, and Tor, providing both conceptual and computational perspectives.