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

First of all, I have read this similar question and am satisfied that the answers there prove the result I am interested in. That being said, I'm more interested in this particular approach than in the actual result, so I think it's reasonable to post this as a new question, even if the approach covered there can be considered a "better" answer to the underlying question.

Fix the notation $$F_k := \textrm{ the free group on } k \textrm{ generators}.$$ Let $n,m \in \mathbb{Z}^+$ and furthermore let $n > 1$. I claim that there exists a subgroup $G$ of $F_n$ with $$G \cong F_m.$$ To "prove" this statement (actually, I believe the argument is only a heuristic):

First, the statement is trivial if $m < n$, so assume $m > n$. For the cases $n = 2, 3$, to demonstrate that $F_n$ contains an isomorphic copy of a "larger" free group, write $$F_2 = \langle a,b \rangle; \ \ \ G = \langle aa,bb,ab \rangle \subset F_2; \ \ \ G \cong F_3 $$ and $$F_3 = \langle a,b,c \rangle; \ \ \ G = \langle ab,ac,bc,aa \rangle \subset F_3; \ \ \ G \cong F_4. $$

Finally, to demonstrate that when $n \geq 4$, $F_n$ contains an isomorphic copy of a "larger" free group, write $$F_n = \langle g_1, \dots, g_n \rangle.$$

We use the observation that there are no relations between the elements of the set $$B := \{g_i g_j : 1 \leq i < j \leq n \}.$$ This set has cardinality ${n \choose 2} > n$, so the group $$G = \langle B \rangle \subset F_n$$ is isomorphic to $F_{n\choose 2}$.

Now, I'm not sure if the observation in the last few lines is actually true, and I would really love a counterexample if one exists or a hint of how I might begin a proof of it. But, supposing it holds true, then this line of reasoning apparently tells us how to write a proof of the original statement for any particular $n >1$, $m >n$. What I'm not sure of is how to write a rigorous argument that encapsulates every case without making use of the statement "and now if we wanted to we could..." or some such.

share|cite|improve this question
do you know how to prove that $F_3$ embeds in $F_2$? – janmarqz Jan 9 '14 at 4:51
There is an algorithmic process known as Nielsen reduction that can be used to decide whether any given subset of a free group freely generates a subgroup. – Derek Holt Jan 9 '14 at 8:59
Stallings foldings also work, which is a more visual method than Nielsen reduction. – user1729 Jan 9 '14 at 17:00
up vote 4 down vote accepted

In the case $n=4$ let the generators of $F_n = \langle g_1, \dots, g_n \rangle$ be for simplicity denoted $F_4 = \langle a,b,c,d \rangle$ so that your set $B$ contains all six two-letter products with the letters in ascending alphabetic order. Then from $ad,bd,bc$ we can express $ac$ as $$(ad)(bd)^{-1}(bc)=add^{-1}b^{-1}bc=ac.$$ So it seems the products in $B$ may have relations between them.

share|cite|improve this answer
Thanks--I don't know how I managed to miss this. It was rather late last night. After a little thought I believe I think I can extend your counterexample for this case to at least one counterexample for each case $n\geq 4$, so I'm satisfied that a different approach to proving this result is needed. – Nick Jan 9 '14 at 12:11
Any sequence of four generators $a,b,c,d$ among the initial list of generators which are in increasing lexicographical order, which is used to set up the set $B,$ will give this counterexample. IMO it wouold be interesting if one could get all the relations satisfied by the sets $B$ this way. That would give a description of a class of finitely generated groups, each probably infinite. – coffeemath Jan 9 '14 at 18:27

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.