6

How to find (describe) all groups which have 3 conjugacy classes?

Thanks in advance!

michael
  • 101
  • 1
    $|Z(G)| \leq 3$. So there is not any p-Group with $p \not= 3$. I'll try to obtain more. – Ivan Di Liberti Dec 26 '12 at 14:15
  • It's good to post the work you've done up until now, whenever you post a question. – rschwieb Dec 26 '12 at 14:33
  • 2
    I would guess that this is very difficult (probably impossible) for infinite groups. There are many infinite groups with just two conjugacy classes. Perhaps the exercise was just about finite groups? – Derek Holt Dec 26 '12 at 17:34
  • http://math.stackexchange.com/questions/52350/finite-groups-with-exactly-n-conjugacy-classes-n-2-3?rq=1 – Ivan Di Liberti Dec 26 '12 at 21:03
  • 1
    Duplicate: https://math.stackexchange.com/questions/2664300/for-general-n-in-bbb-n-how-to-determine-all-groups-both-finite-and-infini – mr_e_man Sep 20 '23 at 19:54

1 Answers1

2

Well, the unit element is always a conjugacy class by itself, just as any element is in any abelian group, so you need other two classes...for example, the (cyclic) abelian group of order $\,3\,$, but also the permutation group $\,S_3\,$, which only has transpositions and $\,3-$cycles.

It could be now a nice exercise to show the above are the only finite groups, up to isomorphism, with three conjugacy classes.

About infinite groups I don't know: constructions like HNN show there can be very funny groups all the non-unit elements of which are conjugated, so this may require lots of care

DonAntonio
  • 214,715
  • Are you sure that there is not any other? – Ivan Di Liberti Dec 26 '12 at 14:38
  • 1
    In fact I remembered once having done this, yet I wasn't completely sure. The following paper though supports this: http://link.springer.com/article/10.1007%2FBF02764723 – DonAntonio Dec 26 '12 at 18:30
  • I'm trying to understand why. I just discovered that if we call $C_1, C_2, C_3$ the conjugacy classes than $|G|= \frac{|C_2||C_3|}{|C_2||C_3|-(|C_2|+|C_3|)}$, calling $C_1$ the class of $e$. This seems to be possible just for 3 and 6 (Wolfram computing). – Ivan Di Liberti Dec 26 '12 at 19:37
  • 2
    I'd rather write the above as $$|G|=1+\frac{|G|}{|C_2|}+\frac{|G|}{|C_3|}\Longleftrightarrow 1=\frac{1}{|G|}+\frac{1}{|C_2|}+\frac{1}{|C_3|}$$ Observe that we can assume $,|G|>3,$ and non abelian, so it can't be that $,|C_2|\geq 4,$ and etc. Try to finish this. – DonAntonio Dec 27 '12 at 00:40