No label defined (Q4029634)

[For the entire collection see Zbl 0745.00017.] A quasigroup can be thought of either as an algebra of type 2, with an operation \(*\), such that, in the equation \(a*b=c\), any two elements uniquely determine the third, or as an algebra of type 2, 2, 2 with operations \(*, /, \backslash\), where, if \(a*b=c\) then \(a=c/b\) and \(b=a\backslash c\). When a single quasigroup is being considered, the distinction between the two viewpoints is not so important, but it becomes vital when varieties or quasivarieties enter the picture; since the variety generated by a quasigroup, considered as a groupoid, may contain algebras which are not quasigroups. It is well known that the variety generated by a finite quasigroup (qua groupoid) is in fact a variety of quasigroups; indeed, it is sufficient that the two generator free groupoid in the variety generated by a quasigroup \({\mathbf Q}\) be finite. This is proved by showing that the operations / and \(\backslash\) can be expressed as terms in \(*\). This leads to two questions: (1) What quasigroups have the property that every essentially binary operation is a quasigroup operation? (2) Are there quasigroups such that every function is a term function? The former question is illustrated by examples of quasigroups associated with certain decompositions of the complete graph into edge-disjoint cycles, and the second by consideration of Sheffer-quasigroups, which are quasigroups \(\langle Q;\bullet\rangle\) in which the clone generated by \(\bullet\) is the clone of all functions on \(Q\). (Such quasigroups are shown to exist for all finite orders greater than 3.) The paper also considers a number of other interesting topics, including quasivarieties of quasigroups and spectra of varieties of quasigroups.