Almost locally free groups and the genus question

Abstract

Sacerdote [Sa] has shown that the non-Abelian free groups satisfy precisely the same universal-existential sentences Th(F$_2$)$\cap \forall \exists $ in a first-order language L$_o$ appropriate for group theory. It is shown that in every model of Th(F$_2$)$\cap \forall \exists $ the maximal Abelian subgroups are elementarily equivalent to locally cyclic groups (necessarily nontrivial and torsion free). Two classes of groups are interpolated between the non-Abelian locally free groups and Remeslennikov's $\exists $-free groups. These classes are the \textbf{almost locally free groups} and the \textbf{quasi-locally free groups}. In particular, the almost locally free% \textbf{\ }groups are the models of Th(F$_2$)$\cap \forall \exists $ while the quasi-locally free groups are the $\exists $-free groups with maximal Abelian subgroups elemenatarily equivalent to locally cyclic groups (necessarily nontrivial and torsion free). Two principal open questions at opposite ends of a spectrum are: (1.) Is every finitely generated almost locally free group free? (2.) Is every quasi-locally free group almost locally free? Examples abound of finitely generated quasi-locally free groups containing nontrivial torsion in their Abelianizations. The question of whether or not almost locally free groups have torsion free Abelianization is related to a bound in a free group on the number of factors needed to express certain elements of the derived group as a product of commutators.