The third isomorphism theorem states that we can relate an isomorphic relation between two normal subgroups of a group $G$. My question is can we infer anything about the two groups structures itself given that the factor/quotient groups in the derived series of two solvable groups are isomorphic to one another? In other words, if we have two solvable groups $G=\langle g_1,g_2,\ldots, g_n \rangle$ and $H=\langle h_1,h_2,\ldots h_p \rangle$, given that $G^{j}/G^{j+1}$ is isomorphic to $H^{j}/H^{j+1}$, what can be said about $G$ and $H$? (I don't think they are isomorphic but anything less strong be said?)

  • 3
    $\begingroup$ I think this question would be much clearer if you used letters to refer to the various groups and subgroups you're thinking about. $\endgroup$ Mar 2, 2013 at 21:53
  • $\begingroup$ Hope my reformulating the question helps :) $\endgroup$
    – Annonymous
    Mar 2, 2013 at 22:14
  • 1
    $\begingroup$ That is helpful :) If you can think of a more descriptive title, I think that'd also be quite useful. $\endgroup$ Mar 2, 2013 at 22:30
  • 5
    $\begingroup$ They are certainly not (necessarily) isomorphic; they certainly do have the same order; I'm not sure much more can be said. For example, there are zillions of groups of order $2^n$, even for modest values of $n$, and they all have isomorphic factor groups. $\endgroup$ Mar 2, 2013 at 22:52
  • $\begingroup$ Per chance we shouuld apply group cohomology to compute these groups? $\endgroup$
    – awllower
    Mar 4, 2013 at 10:09

2 Answers 2


There is a related concept that is very useful. Given a residually nilpotent group $G$, one can consider the set $$ L(G) = \bigoplus_{i}^{\infty} G_{i}/G_{i+1} $$ where the $G_{i}$ are the terms of the lower central series, defined by $G_{1} = G$, and $G_{i} = [G_{i-1}, G]$ for $i > 1$. Then one can show that $L(G)$ can be given the structure of a Lie ring, where the Lie bracket comes from the group commutator. Several problems on $G$ can be dealt with more efficiently by looking at $L(G)$.

However many non-isomorphic groups $G$ may have isomorphic $L(G)$, so of course not all properties of $G$ can be read off $L(G)$. This is the case for instance for $p$-groups of maximal class $G$ of a fixed order $p^{n}$, with $n > p+1$, that share the same $L(G)$.

  • $\begingroup$ @Annonymous, thanks a lot $\endgroup$ Mar 12, 2013 at 16:03

If finite groups $G, H$ are metacyclic (i.e. $G'$ and $G/G'$ are cyclic) then your demand implies that their Sylow subroups are isomorphic (since in this case Sylow subroups are cyclic); see M.Hall, The Theory of Groups, Theorem 9.4.3.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.