Naturally dualizable algebras omitting types 1 and 5 have a cube term (Q271696): Difference between revisions

The paper shows that if a finite algebra omits the tame congruence theory types 1 and 5 (or, equivalently, the variety generated by the algebra satisfies a nontrivial congruence identity, see [\textit{D. Hobby} and \textit{R. McKenzie}, The structure of finite algebra. Providence, RI: American Mathematical Society (1988; Zbl 0721.08001)]) and the algebra is dualizable, then it must have a cube term. Natural dualities are a general framework that generalizes celebrated dualities of Stone and Priestley [\textit{D. M. Clark} and \textit{B. A. Davey}, Natural dualities for the working algebraist. Cambridge: Cambridge University Press (1998; Zbl 0910.08001)]. A cube term is a weaker version of the near unanimity (NU) operation. The role of cube terms in congruence modular varieties is similar to the role that NU operations play in congruence distributive varieties. It is known [\textit{B. A. Davey} et al., Algebra Univers. 33, No. 3, 428--439 (1995; Zbl 0824.08007)] that if an algebra generates a congruence distributive variety and is dualizable, then it has an NU operation. The paper improves this result as omitting types 1 and 5 is a weaker assumption than generating a congruence distributive variety. (While cube terms are in general weaker than NU operations, if the algebra in question is finite and generates a congruence distributive variety, then having a cube term implies having an NU operation. One way to see this is to combine Theorem 1.2., Proposition 2.2, and Corollary 2.6 of [\textit{L. Barto} and \textit{A. Kazda}, Int. J. Algebra Comput. 26, No. 5, 1033--1060 (2016; Zbl 1352.08006)]). The proof itself is short and self-contained. The method of the proof is to assume that the algebra \(\mathbb A\) has a cube term blocker (an obstruction to having a cube term) and to construct a subuniverse \(\mathbb B\leq A^Z\) and a subset \(B_0\subset B\) that satisfy the assumptions of a theorem by \textit{B. A. Davey} et al. [Discrete Math. 214, No. 1--3, 145--172 (2000; Zbl 0945.08001)] giving a sufficient condition for inherent non-dualizability. In this way, the author actually shows that if an algebra omits types 1 and 5 and fails to have a cube term, then it must be inherently non-dualizable, a stronger result.