On subnormal series with factors of finite rank: The join problem (Q2366582)

This paper deals primarily with the following situation: \(G\) is a group, not necessarily finite; \(H\) and \(K\) are subnormal subgroups of \(G\) (write \(H\) sn \(G\)) and \(G=\langle H,K\rangle\); \(A\) and \(B\) are subnormal subgroups of \(H\) and \(K\) respectively, with subnormal series from \(A\) to \(H\) and \(B\) to \(K\) with factors of finite rank (write \(A\) sn (fr) \(H\)); \(J=\langle A, B\rangle\). This is a relaxation of the traditional ``join problem'' setup, where \(| H:A|\) and \(| K:B|\) are finite, and the question is whether \(J\) sn \(G\). Theorem A: If in the situation above we have \(G=HK\) then \(J\) sn (fr) \(G\). Now introduce two finiteness conditions: (1) \([H,K]\) has finite rank modulo every term of the lower central series of \(G\); (2) \(H/H'\otimes K/K'\) is a direct sum of a group of finite rank and a periodic divisible group. Assume the situation above. Theorem B: If \(A\) and \(B\) are normal in \(H\) and \(K\) respectively, and either (1) or (2) holds, then \(J\) sn \(G\). Theorem C: If \(J\) sn \(G\) and (1) holds then \(J\) sn (fr) \(G\) (but an example shows this fails if (1) is replaced by (2)). Theorem D: If \(G\) is soluble and (1) or (2) holds then \(J\) sn \(G\) (and it is conjectured that solubility is unnecessary). The paper, which is economically written and well set out, ends with an extension of theorems \(C\) and \(D\) to the situation \(G=\langle H_ 1,\dots, H_ n\rangle\), \(J=\langle A_ 1,\dots, A_ n\rangle\) etc., under the condition that \(G'/[G',G]\) should have finite rank.