MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $S_4 = \left(\begin{array}{cc}0&-1 \\\ 1&0 \end{array}\right) \textrm{ and } S_6 = \left(\begin{array}{cc} 1&-1 \\\ 1&0\end{array}\right)$. Serre proves in his book on trees that $SL_2(\mathbb{Z}) \cong \mathbb{Z}/4 *_{\mathbb{Z}/2} \mathbb{Z}/6$, and $S_4$ and $S_6$ are the elements corresponding to the generators of $\mathbb Z/4$ and $\mathbb Z/6$ (I'm not sure if this is related to my question). Then let $a = S_4 S_6$ and $b = S_4 S_6^2$. I believe every element of $SL_2(\mathbb Z)$ can be written as $S_6^d w S_6^e$, where $w$ is a word in $a$ and $b$ but not $a^{-1}$ or $b^{-1}$.

I wrote a program (for other purposes) that seems to show that there aren't any relations between $a$ and $b$ that have length 15 or less and don't involve $a^{-1}$ or $b^{-1}$. I'm not certain that the program is right, but if it is, one might make a naive guess that these two elements generate a free group. This makes me suspicious.

1) Does $SL_2(\mathbb Z)$ contain a free group (of rank > 1)? If it does, is there an easy way to determine whether the subgroup generated by $a$ and $b$ is free?

2) A slightly less naive guess is that $a$ and $b$ generate a free monoid in $SL_2(\mathbb Z)$. Is there a reason why $SL_2(\mathbb Z)$ can't contain a free monoid, or an example showing that it does?

EDIT: Thanks for the quick replies. As Robin and Jack pointed out, $a$ and $b$ generate SL(2,Z), so clearly don't generate a free group. Also, there are free subgroups that are easy to write down. I'm still curious about #2, though.

share|cite|improve this question
You have to use an extra \ to get the matrices to display correctly. Similarly, to display { } use \\{ \\}. – Douglas Zare Mar 18 '10 at 7:34
Thanks for the tip, Doug. – Robin Chapman Mar 18 '10 at 7:43
First you should work in $PSL_2(\mathbb{Z})$ just to get rid of the annoying $\pm 1$. As you stated the result is that $PSL$ is the free product of the images of $S_4$ and $S_6$. This means that every element has a unique representation as a reduced word in them: i.e a word of the form $S_4^{a_1} S_6^{b_1} \dots S_4^{a_n} S_6^{b_n}$ where $a_i = 1$ and $b_i \in \{1,2\}$ and we can optionally omit the leading and/or trailing factors. From this it's easy to solve any word problem in your $a$ and $b$ (almost as easy as in a free group) – Victor Miller Mar 18 '10 at 22:04
up vote 11 down vote accepted

Certainly $\mathrm{SL}(2,\mathbb{Z})$ contains a free group. For instance $\Gamma(2)$, the subgroup of all matrices congruent to the identity modulo $2$, is free of rank $2$. The matrices $\left(\begin{array}{cc}1&2\\\ 0&1\end{array}\right)$ and $\left(\begin{array}{cc}1&0\\\ 2&1\end{array}\right)$ freely generate $\Gamma(2)$. This can be proved by considering the action on the upper half-plane or by careful examination of reduced words. There's a nice proof in chapter 18 of David Ullrich's book Complex Made Simple.

Your $a$ and $b$ don't generate a free group alas, since they generate all of $\mathrm{SL}(2,\mathbb{Z})$.

Re the edited question. Let's write $$T=\left(\begin{array}{cc}1&1\\\ 0&1\end{array}\right)\qquad \textrm{and}\qquad U=\left(\begin{array}{cc}1&0\\\ 1&1\end{array}\right).$$ As both Jack and I pointed out, $T^2$ and $U^2$ generate a free subgroup of rank $2$. Now it's an easy exercise to prove that $T$ and $U$ freely generate a free monoid of rank $2$ (because their entries are non-negative). On the other hand, they generate thw whole group $\mathrm{SL}(2,\mathbb{Z})$ which is certainly not free. Your matrices $a$ and $b$ are, if my calculations are right, $-U^{-1}$ and $-T^{-1}$. The matrix $S_4$ conjugates $T$ and $U$ into $U^{-1}$ and $T^{-1}$ so $U^{-1}$ and $T^{-1}$ freely generate a free monoid of rank $2$. The same must be true of $U^{-1}$ and $T^{-1}$, that is, of $a$ and $b$.

share|cite|improve this answer
Great, thanks for the nice answer. – Peter Samuelson Mar 18 '10 at 21:45

For a broader perspective on proving that groups of matrices and what not are free, I highly recommend reading chapter II.B of Pierre de la Harpe's book "Topics in Geometric Group Theory", which gives a beautiful account with numerous references of known facts, especially for subgroups of SL_2.

share|cite|improve this answer

BabAba = 1, where Bb=1 and Aa=1. SL(2,Z) does contain free subgroups, for instance on [1,2;0,1] and [1,0;2,1], so there is no reason it could not contain free subgroups, it just so happens that {a,b} does not generate one. Of course {a,b} generates SL(2,Z) since S4=aBa and S6=Ab.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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