Core Universal Algebra

The Subalgebra Lattice as an Algebraic Lattice

Sub(A) is algebraic for every algebra, and every algebraic lattice is Sub(A) for some algebra. The characterisation is exact, which is both satisfying and a dead end.

Engineering · Mathematics3 min readKV-MATH-0211

Learning objectives

Show Sub(A) is a complete lattice and compute its meets and joins.
Identify the compact elements of Sub(A).
State the Birkhoff–Frink theorem in both directions.
Explain why an exact characterisation ends the general question.
Contrast the arbitrary and finite representation problems.

01Sub(A) as a complete lattice

Operations in Sub(A)
Operation	Formula	Cost
Meet	B ∧ C = B ∩ C	Trivial — intersection is already closed.
Join	B ∨ C = Sg(B ∪ C)	Requires the generation construction.
Arbitrary meet	⋂ of the family	Trivial.
Arbitrary join	Sg of the union	Requires generation.
Bottom	Sg(∅)	∅ when there are no constants.
Top	A	Always present.

Completeness follows from the one-sided criterion: arbitrary intersections exist, so arbitrary joins do too. No separate verification is needed, and this is the standard economy in every closure-system argument.

02Compactness and algebraicity

The compact elements of Sub(A) are exactly the finitely generated subuniverses. Suppose B = Sg(Y) with Y finite and B ≤ ⋁ᵢ Cᵢ. Each element of Y lies in the join, hence in Sg of the union, hence by finitariness in Sg of finitely many of the Cᵢ. Taking the union over the finitely many elements of Y gives a finite subfamily whose join contains B.

Key resultAlgebraicity is finitariness in lattice clothing

Every element of Sub(A) is the join of the finitely generated subuniverses below it, because every subuniverse is the union of the subuniverses generated by its finite subsets. So Sub(A) is algebraic for every algebra A, with no hypotheses whatsoever on A.

03The Birkhoff–Frink representation theorem

ProcedureConstructing an algebra from an algebraic lattice

in: algebraic lattice L → out: algebra A with Sub(A) ≅ L

input: algebraic lattice L
let A := the set of compact elements of L
for each finite subset {c₁,…,cₙ} of A and each compact d ≤ c₁ ∨ ⋯ ∨ cₙ:
add an n-ary operation f with f(c₁,…,cₙ) = d
(extend arbitrarily elsewhere, respecting closure)
the subuniverses of the resulting algebra correspond to the elements of L
output: algebra A with Sub(A) ≅ L

Correctness: the ideal of compact elements below a given element is closed under the constructed operations, and conversely. Caveat: the type is typically enormous — one operation per compact join relation — so the representing algebra is of no computational use.

Combined with the forward direction, this gives an exact characterisation: a lattice is isomorphic to Sub(A) for some algebra A if and only if it is algebraic. Nothing more and nothing less.

04What an exact characterisation costs

An exact characterisation is the strongest possible answer to a representation question, and it terminates the enquiry. Since every algebraic lattice occurs, knowing that a lattice is Sub(A) for some A conveys no information beyond algebraicity.

Arbitrary algebras

Settled

Sub(A) ranges over exactly the algebraic lattices. Con A likewise, by Grätzer–Schmidt. No further structure theory is possible at this level of generality.

Finite algebras

Open in the congruence case

Whether every finite lattice is Con A for a finite algebra A is a substantially harder question that the general theorem does not touch, and it drove much later work.

The methodological lesson generalises. When a representation theorem is exact, progress requires changing the question — restricting to finite algebras, to a fixed variety, or to a fixed type. That is precisely what the later development of the subject did.

05Special shapes and what they signal

Modular

Vector spaces

The subspace lattice of a vector space is modular, and complemented besides. Very few algebras have modular subuniverse lattices.

Distributive

Unary algebras

Algebras with only unary operations have distributive subuniverse lattices, since generation reduces to orbit closure.

Arbitrary

The general case

For a general algebra Sub(A) can be any algebraic lattice, so no shape is excluded and no shape is informative on its own.

Frequently asked

Is Sub(A) ever finite for an infinite algebra?

Yes. An infinite algebra with a single unary operation acting as a cyclic shift on the integers has very few subuniverses. More strikingly, an algebra can be infinite and simple in the subalgebra sense, with only ∅ and A as subuniverses.

Does Sub(A) determine A?

Not remotely. Wildly different algebras share the same subuniverse lattice, and the Birkhoff–Frink construction shows the representing algebra is far from unique. Recovering an algebra from a lattice invariant is not a realistic goal.

Why is Con A studied more than Sub(A)?

Because congruences govern quotients, and quotient behaviour is what distinguishes varieties. The Mal'cev conditions, the commutator, the discriminator theory and the decidability results are all statements about Con A. Sub(A) is used chiefly as a source of examples and as the cleanest illustration of algebraicity.

Sources and further reading

S. Burris and H. P. Sankappanavar, A Course in Universal Algebra, Millennium Edition (a corrected re-typesetting of Springer GTM 78, 1981).
G. Grätzer, Universal Algebra, 2nd edition, Springer.
R. McKenzie, G. McNulty and W. Taylor, Algebras, Lattices, Varieties, Volume I.

Original KEVOS® explanatory article. Written from the topic map of the cited works; no text is reproduced from them.