I am interested in the growth rate of this type of group: $G=\mathbb{Z}^2\rtimes_{\sigma} \mathbb{Z}$, where $\sigma(a)=\begin{pmatrix}x&y\\z&w\end{pmatrix}\in SL_2(\mathbb{Z})$, where $a$ is the generator on the right copy of $\mathbb{Z}$ and the action is just by matrix multiplication.

Here are two examples:

For $\sigma(a)=\begin{pmatrix}1&1\\0&1\end{pmatrix}$, this gives us the discrete Heisenberg group $H_3$, which is nilpotent, and hence by Gromov's theorem, it has polynomial growth rate(see here).

When $\sigma(a)=\begin{pmatrix}2&1\\1&1\end{pmatrix}$, it was mentioned here this group has exponential growth rate.

So my first question is:

1, Could anyone give me a reference to show the link between whether the above group $G$ has polynomial growth rate or not and the property, say eignvalue, of the matrix $\sigma(a)$?

Note that the above $G$ is a polycyclic-by-finite group, my question is:

2, Could anyone give me a polycyclic-by-finite group not of the type of $G$ with exponential growth rate?

  • $\begingroup$ The theorem that $H_3$ and other (virtually) nilpotent groups have polynomial growth is a theorem of Milnor, with the exact degree of polynomial growth computed by Bass. Gromov's theorem is the converse: every group of polynomial growth is virtually nilpotent. $\endgroup$
    – Lee Mosher
    Commented Aug 11, 2013 at 16:02
  • $\begingroup$ @Lee, in your answer, you mentioned Milnor's paper, I checked it, but it is still not clear to me how to relate the nilpotentness of $G$ to the property of $\sigma(a)$, could you give more hints? $\endgroup$
    – Jiang
    Commented Aug 11, 2013 at 22:14
  • $\begingroup$ @Lee, especially, is the lemma 1 in Milnor's paper useful in our situation? $\endgroup$
    – Jiang
    Commented Aug 11, 2013 at 22:17
  • $\begingroup$ @Jiang: For your group to be (virtually) nilpotent, $\sigma(a)$ must fix a point (otherwise the center of $G$ would be trivial). You can also check this is a sufficient condition (quotient by the fixed subgroup, and check it is (virtually) abelian). $\endgroup$
    – Steve D
    Commented Aug 13, 2013 at 2:47

1 Answer 1


To answer your 1st question see the paper of Milnor, "Growth of finitely generated solvable groups."

To answer your 2nd question, simply generalize your second example to higher dimensions, e.g. take $\mathbb{Z}^3\rtimes_{\sigma} \mathbb{Z}$ where $\sigma \in SL_3(\mathbb{Z})$ has an eigenvalue not on the unit circle.

  • $\begingroup$ thanks! For the type of $G$, I mean the type of $\mathbb{Z}^d\rtimes \mathbb{Z}$, maybe I should state it clearly next time. $\endgroup$
    – Jiang
    Commented Aug 11, 2013 at 16:14
  • $\begingroup$ Would one of type ${\mathbb Z}^d \lhd {\mathbb Z}^2$ suit? $\endgroup$
    – Derek Holt
    Commented Aug 11, 2013 at 17:36
  • $\begingroup$ @DerekHolt, what does $\vartriangleleft$ mean? $\endgroup$
    – Jiang
    Commented Aug 11, 2013 at 19:28
  • $\begingroup$ I typed the wrong symbol. I meant ${\mathbb Z}^d \rtimes {\mathbb Z}^k$ with $k>1$. You can construct examples like that from algebraic number fields $K$, where you let the torsion-free part of the group of units of $K$ act on the additive group of the integers of $K$, and the action is given by multiplication in $K$. $\endgroup$
    – Derek Holt
    Commented Aug 11, 2013 at 20:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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