Elementary equivalence of Cartesian powers of the same group. (Q616723)

The authors consider groups \(G\) satisfying: (i) There exists a finite subset \(P\) of \(G\) such that the centralizer of all conjugates of \(P\) is exactly \(Z(G)\). (ii) For any two normal subgroups \(H\) and \(K\) containing \(Z(G)\), if \(H\cup K\) generates \(G\) then one of \(H\) or \(K\) is equal to \(G\). (iii) \(G/Z(G)\) is not \(\aleph_0\)-categorical. They prove that for any group \(G\) with these three properties the Cartesian products \(G^I\) and \(G^J\), with \(I\) countably infinite and \(J\) uncountable, are not elementarily equivalent in the \(L_{\omega_1\omega}\) logic. They mention many examples of nonabelian groups (free, solvable, free nilpotent, virtually Abelian, \dots) for which the preceding theorem applies. They also sketch a construction ``by hand'' of a torsion-free Abelian group with the above property. On the other hand, the authors state that for any Abelian torsion group \(G\) and for any infinite \(I\) and \(J\), the Cartesian powers \(G^I\) and \(G^J\) are elementarily equivalent in the \(L_{\infty\omega}\) logic. They introduce the key ingredient, which allows for the proof arguments similar to those in Ulm's Theorem. This new and simple property has a Cantor-Bernstein flavour, and they expect that it may have other applications in other generalizations of Ulm's Theorem.