2-Dimensional Birkhoff Variety Theorem

• The paper establishes a Cat-enriched analogue of classical Birkhoff’s theorem by characterizing equational sub-2-categories via quotient 2-monads.
• It shows that a full 2-subcategory in Alg(T) is equationally defined if and only if it is closed under 2-products, subalgebras, quotient algebras, and sifted colimits.
• The theorem leverages a (b.o. full, faithful) factorisation system, ensuring that equations are encoded through quotient 2-monad morphisms without altering the underlying operations.

Searching arXiv for the core paper and closely related work on 2-monads, Lawvere 2-theories, and 2-dimensional exactness. Search query: "(Dostál, 2015) two-dimensional Birkhoff theorem strongly finitary 2-monad Cat Lawvere 2-theories Bourke Garner" The 2-dimensional Birkhoff variety theorem is a Cat-enriched analogue of classical Birkhoff’s variety theorem in universal algebra. It characterizes when a full 2-subcategory of an algebraic 2-category $\Alg(T)$, for a strongly finitary 2-monad $T$ on $\Cat$, is equationally defined. The theorem states that such a subcategory is, up to equivalence, the 2-category of algebras for a quotient 2-monad $T'$ precisely when it is closed under 2-products, subalgebras, quotient algebras, and sifted colimits. In this setting, equations are encoded not by Set-based term identifications but by quotient 2-monad morphisms whose components are bijective on objects and full, so that the 2-dimensional structure of operations is constrained while the underlying operations are not replaced by new ones (Dostál, 2015).

1. The theorem and its formal statement

Fix the base 2-category $\V=\Cat$. Let $T$ be a strongly finitary 2-monad on $\Cat$, and let $\Alg(T)$ denote the 2-category of strict $T$-algebras and strict morphisms. The two-dimensional Birkhoff theorem asserts the equivalence of two conditions for a full 2-subcategory

$J:\mathcal{A}\hookrightarrow \Alg(T).$

The first condition is the existence of a strongly finitary 2-monad $T$0 on $T$1 and a quotient monad morphism

$T$2

such that each component $T$3 is bijective on objects and full, and such that the induced algebraic 2-functor

$T$4

is fully faithful with essential image equivalent to $T$5. The second condition is that $T$6 is closed in $T$7 under 2-products, subalgebras, quotient algebras, and sifted colimits. The theorem is formulated relative to the $T$8 factorisation system on $T$9, and this choice is decisive for the meaning of “equational” in dimension $\Cat$0 (Dostál, 2015).

Conceptually, the theorem transfers the classical HSP pattern to Cat-enriched algebra: varieties become 2-categories of algebras for strongly finitary 2-monads, and equationally defined subvarieties become full sub-2-categories arising from quotient 2-monads. A $\Cat$1-algebra is then a $\Cat$2-algebra satisfying the equations encoded by the quotient $\Cat$3.

2. Ambient 2-categorical framework

The ambient setting is enrichment over $\Cat$4, viewed as a symmetric monoidal closed category. Accordingly, the basic objects are 2-categories, 2-functors, and 2-natural transformations. In this framework, a strongly finitary 2-monad on $\Cat$5 is one whose underlying 2-functor preserves sifted colimits. The paper emphasizes that, unlike the situation in $\Cat$6, “finitary” and “strongly finitary” differ in $\Cat$7 (Dostál, 2015).

These strongly finitary 2-monads are precisely the one-sorted algebraic 2-categories, equivalently categories of algebras for Cat-enriched Lawvere 2-theories. More formally, the 2-category $\Cat$8 of strongly finitary 2-monads is locally finitely presentable in the 2-sense and is equivalent to the 2-category of Lawvere 2-theories: $\Cat$9

For each such $T'$0, the underlying 2-functor

$T'$1

creates all limits and sifted colimits. This is the structural basis for the closure conditions in the theorem. Closure under 2-products is meaningful because products in $T'$2 are computed pointwise on underlying categories, and closure under sifted colimits is forced by the strong finitarity hypothesis: the relevant algebraic structure is preserved exactly along those colimits.

A “variety” in the 2-dimensional sense is therefore a 2-category equivalent to $T'$3 for some strongly finitary 2-monad $T'$4. An equationally defined subcategory is a full 2-subcategory of $T'$5 equivalent to $T'$6 for a suitable quotient $T'$7.

3. Quotients, equations, and algebraic closure

The central technical issue is the correct notion of quotient. In the classical Set-based setting, quotients are controlled by the regular epi-mono factorisation system. In $T'$8, the theorem instead uses

$T'$9

This $\V=\Cat$0 factorisation system comes from a kernel-quotient system $\V=\Cat$1. The paper states that $\V=\Cat$2 is $\V=\Cat$3-exact and that $\V=\Cat$4 is also exact for this system, so the same factorisation structure lifts to strongly finitary 2-monads (Dostál, 2015).

A morphism $\V=\Cat$5 of strongly finitary 2-monads is therefore a quotient exactly when every component

$\V=\Cat$6

is b.o. full. Such quotients are computed as coequifiers, and, in the monadic situation relevant here, they may be taken to be reflexive coequifiers, hence sifted colimits.

The equational interpretation of these quotients is explicitly 2-dimensional. For each finite discrete category $\V=\Cat$7,

$\V=\Cat$8

is b.o. full. Objects of $\V=\Cat$9 and $T$0 represent $T$1-ary operations or terms. Bijectivity on objects means no new operations are introduced and terms are not collapsed to terms. Fullness acts on morphisms between such objects, which encode 2-dimensional structure, coherence constraints, or equations between derived operations. The intended reading is that $T$2 is obtained from $T$3 by imposing equations between morphisms, that is, between 2-cells, rather than by altering the stock of operations themselves.

The closure conditions in $T$4 match this algebraic reading. A subalgebra of $T$5 is an algebra $T$6 equipped with a faithful functor $T$7 making the defining square commute. A quotient algebra is an algebra $T$8 equipped with a b.o. full homomorphism $T$9 making the corresponding square commute. Because $\Cat$0 is strongly finitary, $\Cat$1 is again b.o. full. The paper also requires that the subcategory be closed under sifted colimits, including filtered colimits and reflexive coequalisers, and notes that $\Cat$2 is cowellpowered with respect to these quotients.

4. Proof architecture

The direction from equations to closure begins with a b.o. full quotient $\Cat$3. The induced algebraic 2-functor

$\Cat$4

acts by precomposition of the algebra structure,

$\Cat$5

and the b.o. full hypothesis implies that any $\Cat$6-algebra morphism or 2-cell between such images is automatically a $\Cat$7-algebra morphism or 2-cell. This yields full faithfulness. The same functor preserves limits and sifted colimits, since both algebraic 2-categories have them and these structures are created by the respective underlying 2-functors. Closure under subalgebras and quotient algebras is then established by lifting faithful embeddings and b.o. full quotients through the quotient monad situation, using exactness and the creation of coequifiers (Dostál, 2015).

The converse direction reconstructs the quotient monad from closure properties. Assume $\Cat$8 is full, replete, and closed under 2-products, subalgebras, quotient algebras, and sifted colimits. First, one shows that the inclusion $\Cat$9 has a left adjoint $\Alg(T)$0, initially at the ordinary categorical level, via Freyd’s Adjoint Functor Theorem and the solution-set argument enabled by cowellpoweredness with respect to b.o. full quotients. The paper then applies a 2-dimensional adjoint functor theorem: under 2-continuity, 2-completeness, and a suitable b.o. full “2-solution set” condition, the ordinary adjunction lifts to a 2-adjunction.

From the composite adjunctions

$\Alg(T)$1

one defines

$\Alg(T)$2

The units and counits of the two adjunctions induce the unit and multiplication of $\Alg(T)$3, making $\Alg(T)$4 a 2-monad. The paper states that $\Alg(T)$5 is strongly finitary because it is built from strongly finitary pieces and its underlying 2-functor preserves sifted colimits. Beck-style monadicity then gives an equivalence

$\Alg(T)$6

Finally, the induced monad morphism

$\Alg(T)$7

is shown to be pointwise b.o. full, since each unit component $\Alg(T)$8 is b.o. full and $\Alg(T)$9 preserves such quotients. One then identifies $T$0, up to isomorphism, with

$T$1

which completes the equivalence of the theorem.

5. Relation to classical Birkhoff and orthogonality

The 2-dimensional Birkhoff variety theorem is a direct generalization of classical Birkhoff. In the classical theorem, for a finitary Set-monad $T$2, a full subcategory of $T$3 is equational exactly when it is closed under homomorphic images, subalgebras, and products, with filtered colimits entering in many-sorted or algebraic-theory formulations. In the Cat-enriched version, the ambient variety is $T$4 for a strongly finitary 2-monad, homomorphic images are replaced by quotient algebras with underlying b.o. full maps, subalgebras are encoded by faithful morphisms, products become 2-products, and filtered colimits are subsumed by sifted colimits (Dostál, 2015).

This comparison is not merely formal. The passage from Set to $T$5 changes what counts as an equation. In the classical case, quotient monads can identify terms. Here, the b.o. full condition on $T$6 prevents the introduction of new operations and preserves objects of the operation categories, while allowing identifications among morphisms. The theorem therefore isolates a specifically 2-dimensional notion of equational consequence.

The paper also describes equational subcategories as reflective sub-2-categories and relates them to orthogonality. If $T$7 is a b.o. full quotient, then $T$8 is fully faithful and its essential image is reflective in $T$9. Conversely, equational subcategories are exactly classes of objects orthogonal to a set of quotient maps

$J:\mathcal{A}\hookrightarrow \Alg(T).$0

with finitely generated free domains. This places the theorem within the general interaction among monadicity, reflectivity, and orthogonality in enriched universal algebra.

6. Examples, scope, and limitations

A guiding example in the paper is monoidal structure. There is a strongly finitary 2-monad $J:\mathcal{A}\hookrightarrow \Alg(T).$1 such that $J:\mathcal{A}\hookrightarrow \Alg(T).$2, where objects are categories equipped with a nullary operation $J:\mathcal{A}\hookrightarrow \Alg(T).$3, a binary operation $J:\mathcal{A}\hookrightarrow \Alg(T).$4, and natural isomorphisms $J:\mathcal{A}\hookrightarrow \Alg(T).$5, but without the usual monoidal coherence axioms. There is another strongly finitary 2-monad $J:\mathcal{A}\hookrightarrow \Alg(T).$6 such that $J:\mathcal{A}\hookrightarrow \Alg(T).$7 (strict or weak monoidal categories, depending on the encoding). A monad morphism

$J:\mathcal{A}\hookrightarrow \Alg(T).$8

with b.o. full components then quotients by the coherence equations, and the induced fully faithful algebraic 2-functor identifies the coherent monoidal categories as an equational full sub-2-category of the larger algebraic 2-category (Dostál, 2015).

The theorem is also delimited by its dependence on the $J:\mathcal{A}\hookrightarrow \Alg(T).$9 factorisation system. The paper discusses two alternative systems, $T$00 and $T$01, and states that the theory deteriorates there. In particular, a monad morphism with components merely bijective on objects need not induce a fully faithful algebraic 2-functor, b.o. functors need not be epimorphic in $T$02, $T$03 is not cowellpowered with respect to b.o. quotients, and pointwise b.o. quotients may alter the 2-dimensional shape of the theory by introducing new 2-cells. For these reasons, the resulting subcategories are no longer “equational” in the same sense.

The paper closes with two open directions. One is the development of an explicit equational logic, sound and complete for the notion of equational consequence induced by b.o. full monad quotients in this 2-dimensional setting. The other is the extension of the framework beyond $T$04 and beyond strongly finitary 2-monads, especially toward finitary but not strongly finitary monads where preservation of sifted colimits fails.

The terminology should also be distinguished from other uses of “Birkhoff theorem.” In mathematical physics and differential geometry, the phrase refers to results about extra isometries of Einstein warped products and the special role of two-dimensional pseudo-Riemannian geometry, rather than to algebraic 2-categories or 2-monads (Schmidt, 2012). The 2-dimensional Birkhoff variety theorem belongs instead to 2-dimensional universal algebra and the theory of Cat-enriched algebraic structures.