Suppose one has two locally finite quasivarieties $\mathcal{V}$ and $\mathcal{W}$.

Further suppose that:

  1. $\mathcal{V}$ is a variety.

  2. The finite algebras $\mathcal{V}_f$ are dually equivalent to $\mathcal{W}_f$.

Does it follow that $\mathcal{W}$ is also a variety?

Background: This question fits into the area of `Natural duality'. I am hoping someone can provide a counterexample, or perhaps provide additional conditions.

Here is a way to manufacture such examples. Let $\mathcal{V}$ be a locally finite variety such that (i) $\mathcal{V}$-epis are surjective, (ii) $\mathcal{V}_f$ has an injective cogenerator $A$. Define the algebra $B$ to have the same carrier as $A$, its finitary operations between all $\mathcal{V}$-morphisms $A^n \to A$. Further define $\mathcal{W} := \mathcal{Q}(B) = \mathbb{S}\mathbb{P}(B)$ to be the quasivariety generated by $B$. Then $\mathcal{V}_f$ is dually equivalent to $\mathcal{W}_f$ via the contravariant hom-functors for $A$ and $B$, endowed with additional algebraic structure.

e.g. Let $\mathcal{V} = \mathsf{Vect}(\mathbb{F})$ for a finite field $\mathbb{F}$ and take $A = \mathbb{F}$. Then $\mathcal{W}$ is essentially $\mathcal{V}$ again, where every linear combination of variables is viewed as an operation.


Here is a counterexample.

Let $A = \langle \{0,1\}; d, s\rangle$ be the algebra on $\{0,1\}$ whose defining operations are the discriminator operation $d$ on $\{0,1\}$ and the "switch" ($s(0)=1$, $s(1)=0$). According to the NU Duality Theorem of natural duality theory, $A$ is dualizable, and the topological quasivariety dual to $\mathcal V:=SP(A)$ is $\mathcal W:=S_cP(\langle \{0,1\}; s, {\mathcal T}\rangle)$, where $\mathcal T$ is the discrete topology.

I am allowing the empty algebra, else the duality doesn't work out perfectly (empty is dual to singleton in both directions).

It follows from Jonsson's Lemma that $\mathcal V:=SP(A)$ is a variety, but it is easy to see that $\mathcal W_f$ is not the finite part of a variety. $\mathcal W_f$ is the finite part of the quasivariety of $\mathbb Z_2$-sets having no fixed points, while the variety generated by $\mathcal W_f$ contains all $\mathbb Z_2$-sets.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.