Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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 $G=\langle X \mid R \rangle$ be a presentation where $X$ and $R$ are finite and $R$ contains words of length at most 3. Can we deduce that $G$ is either virtually abelian or virtually free?

share|cite|improve this question
Do you have any reason to believe that this is true? I can't see one, and a quick search turned up no references to this or similar results. In fact, I'm given to think that $X = \{a,b\}, R = \{ab\}$ might be a counterexample, as this definitely isn't virtually free and I can't think of a subgroup of finite index of the free group on two elements which contains $ab$. – Alex Becker Dec 25 '11 at 6:10
Isn't your example an abelian group? – Mustafa Gokhan Benli Dec 25 '11 at 6:14
It would be even better to use $X = \{a,b,c\}$ I think. But I'm pretty sure group in my earlier comment is not abelian. It would be true if $R = \{ab,ba\}$, but in my example $ab = 1$ while $ba\neq 1$ as $ba$ is not in the subgroup generated by $ab$. The following computations show this, and are performed in $F_2$ rather than $<X,R>$: $w(ab)^nw^{-1} = ba\implies [w,(ab)^n] = w(ab)^nw^{-1}(ab)^{-n} = ba(ab)^{-n}$ and this can only be true for $n=1$ as otherwise the image of $[w,(ab)^n]$ in the free abelian group would be nontrivial, but then $[w,ab] = 1$ so $wabw^{-1} = ab\neq ba$. – Alex Becker Dec 25 '11 at 6:30
Dear Alex, if in a group the product $ab$ of two elements $a$ and $b$ is equal to $1$, then the product $ba$ is also equal to $1$. (The important thing is, is $ba$ in the normal subgroup generated by $ab$?) – Mariano Suárez-Alvarez Dec 25 '11 at 6:32
This has nothing to do, really, with that :D If $ab=1$, then $a=b^{-1}$, and obviously $a$ and $b$ commute! – Mariano Suárez-Alvarez Dec 25 '11 at 6:39
up vote 4 down vote accepted

If $r=x_1\cdots x_n$ is a relation of length $n>3$ in a presentation of a group, introducing a new generator $v$, and removing $r$ from the relations and adding in its place the two relations $x_1x_2u$ and $u^{-1}x_3\cdots x_n$ one gets a new presentation for the same group, which is closer to being «with relators of length at most $3$».

If we start with a finitely presented group, iterating this will give us a presentation in which all relations are of length at most $3$.

It follows that every finitely presented group has a presentation with relations of length at most $3$. (Of course, the same idea will also work for some infinite presentations...)

share|cite|improve this answer

Any group has a presentation in which all relations have length 3. Take the group elements as generators, and all entries $ab=c$ of its multiplication table as relations. Proving this really is a presentation is often given as an early exercise in the theory of group presentations.

share|cite|improve this answer
Is there an oh-duh way of seeing from this construction that if you start with a finite presentation you can produce a cubic finite presentation? – Mariano Suárez-Alvarez Dec 28 '11 at 4:44
I think the method you described in your earlier post is the easiest way of doing this. – Derek Holt Dec 28 '11 at 12:10

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.