Commutative idempotent groupoids and the constraint satisfaction problem

Abstract

A restatement of the Algebraic Dichotomy Conjecture, due to Maroti and McKenzie, postulates that if a finite algebra A possesses a weak near-unanimity term, then the corresponding constraint satisfaction problem is tractable. A binary operation is weak near-unanimity if and only if it is both commutative and idempotent. Thus if the dichotomy conjecture is true, any finite commutative, idempotent groupoid (CI groupoid) will be tractable. It is known that every semilattice (i.e., an associative CI groupoid) is tractable. A groupoid identity is of Bol-Moufang type if the same three variables appear on either side, one of the variables is repeated, the remaining two variables appear once, and the variables appear in the same order on either side (for example, x(x(yz)) ≈ (x(xy))z). These identities can be thought of as generalizations of associativity. We show that there are exactly 8 varieties of CI groupoids defined by a single additional identity of Bol-Moufang type, derive some of their important structural properties, and use that structure theory to show that 7 of the varieties are tractable. We also characterize the finite members of the variety of CI groupoids satisfying the self-distributive law x(yz) ≈ (xy)(xz), and show that they are tractable.

Semilattice sums of algebras and Mal’tsev products of varieties

“…Then for a subvariety V t of T t , let V t be defined by Γ t and (P1) -(P4). As observed by C. Bergman and D. Failing [3]…”

“…However it is possible for the regularization of a variety to be distinct from its pseudo-regularization. The first example was discovered in [3], and investigated in connection with the constraint satisfaction problem. SQ ⊂ SQ ⊂ SQ.…”

Logics of left variable inclusion and Płonka sums of matrices

“…Since · is a partition function for , we have x x · y. Together with the fact that A, F ∈ Mod( ) and a ∈ F, this implies that a · A b ∈ F. Observe that a · A b ∈ A j by (2) in Theorem 7 and, therefore, that a · A b ∈ F j . Hence, by (3), we have that f ij (a) = a · A b ∈ F j .…”

“…Moreover, every A i is the universe of a subalgebra A i of A. (2) The relation ≤ on I given by the rule…”

Cube term blockers without finiteness

“…Namely, it is of interest to understand whether the algebras in the pure cyclic term varieties have tractable CSP's. The d-ary pure cyclic term variety C d is defined with one d-ary operation satisfying Bergman and David Failing showed in [3] that if V is a subvariety of C 2 that is the join of a congruence permutable variety and the variety of semilattices, then the finite algebras in V have tractable associated CSP's. So, Bergman was really asking whether this theorem applied to every subvariety of C 2 that is a join of the variety of semilattices and a disjoint subvariety.…”