# Bounds on number of conjugacy classes in terms of number of elements of a group ?

What are bounds on number of conjugacy classes in terms of number of elements of a group ?

(I allowed myself to edit the question in spirit of remarkable answers given to it by Gerry Myerson and Geoff Robinson. Below is original text of the question. (Alexander Chervov) ).

It's about the first step to find an upper bound to the order of a finite group with h conjugacy classes (right or left) that depends only on h. (h a natural non nul integer).

I have some doubts about the rigor of my proof that I am sharing with you so that you can help me find a likely error or an omited step.

I have attached the scan of my proof to this post.

Many thanks

• freeimagehosting.net/image.php?fdc742830a.jpg links to the scan (not anymore -- Todd Trimble) – AlGoRiS Mar 17 '11 at 23:41
• Is this a supervision problem? Certainly the calculation reminds me of something I've seen, but perhaps I have misread. – Yemon Choi Mar 17 '11 at 23:49
• I also have misgivings about using MO to "check if my argument is correct" – Yemon Choi Mar 17 '11 at 23:49
• Fair point, but wouldn't math.stackexchange.com be a more natural home for this? – Yemon Choi Mar 18 '11 at 0:02
• springerlink.com/content/n737pq17655582qv 1970 The number of conjugacy classes in a finite group Patrick X. Gallagher compares group and subgroups cc – Alexander Chervov Sep 13 '12 at 10:07

There is a theorem of E. Landau which proves that if you fix a positive integer h, there are only finitely many finite groups with h conjugacy classes. This proof is more number theory than group theory, in fact. More recently, one person who has worked more extensively on this question using more group theory is L. Pyber.

• These are useful observations. But references to the existing literature would also be quite helpful. People often start thinking about problems without knowing precisely what has already been done, which is fine for their education but unfortunately is usually not efficient. – Jim Humphreys Apr 17 '11 at 22:10
• The reference kindly given below gives a reference to Landau's paper in its own bibliography. – Geoff Robinson Apr 17 '11 at 23:59

See MR1182481 (93i:20028) Pyber, L., Finite groups have many conjugacy classes, J. London Math. Soc. (2) 46 (1992), no. 2, 239–249. From the review by I. Ya. Subbotin:

Let $k(G)$ denote the number of conjugacy classes of a finite group $G$. R. Brauer observed that for every group $G$ of order $n$ we have $k(G)\ge\log\log n$, and proposed the problem of finding substantially better bounds [R. Brauer, in Lectures on modern mathematics, Vol. 1, 133--175, Wiley, New York, 1963; MR0178056 (31 #2314)]. The author proves that every group of order $n$ contains at least $\epsilon\log n/(\log\log n)^8$ conjugacy classes for some fixed $\epsilon$. This essentially settles the problem of Brauer.

• Why does it settle the problem? – user6976 Dec 13 '19 at 1:38
• @Mark, Brauer "proposed the problem of finding substantially better bounds". Pyber found a substantially better bound, thus settling Brauer's problem. I'm not claiming Pyber found the best possible bound, just that Pyber did what Brauer proposed should be done. – Gerry Myerson Dec 13 '19 at 1:46

Zbl 1227.20014. of A. Jaikin-Zapirain paper says that there is conjecture by Ya. Berkovich and V. Zhmud that: number of conj classes > number of prime factos of G (achieved for M_22 and PSL(3,4) )

Quote from Andrei Jaikin-Zapirain paper (Adv. Math. 227, No. 3, 1129-1143 (2011). )

Conjecture. There exists a constant C> 0 such that any finite group G of order n satisfies k(G) > C log2( n).

Main theorem of this paper is the following:

In this paper we establish the first super-logarithmic lower bound for the number of conjugacy classes of a finite nilpotent group.

Theorem 1.1. There exists a (explicitly computable) constant C > 0 such that every finite nilpotent group G of order n 8 satisfies

k(G) > C log2 (n) ((log2 log2 n) / ( log2 log2 log2 n) )

Introduction to the paper contains discussion of some history of the subject is quite readable.

On p-groups having the minimal number of conjugacy classes of maximal size A. Jaikin-Zapirain, M. F. Newman and E. A. O’Brien

A long-standing question is the following: do there exist p-groups of odd order having precisely p-1 conjugacy classes of the largest possible size? We exhibit a 3-group with this property.

http://springerlink.com/content/n737pq17655582qv 1970 The number of conjugacy classes in a finite group. Patrick X. Gallagher.

This paper contains results comparing number of conjugacy classes in a group and in its subgroup.

http://arxiv.org/abs/1102.4107

Multiplicities of conjugacy class sizes of finite groups

Hung Ngoc Nguyen

It has been proved recently by Moreto and Craven that the order of a finite group is bounded in terms of the largest multiplicity of its irreducible character degrees. A conjugacy class version of this result was proved for solvable groups by Zaikin-Zapirain. In this note, we prove that if $G$ is a finite simple group then the order of $G$, denoted by $|G|$, is bounded in terms of the largest multiplicity of its conjugacy class sizes and that if the largest multiplicity of conjugacy class sizes of any quotient of a finite group $G$ is $m$, then $|G|$ is bounded in terms of $m$.

In the meantime, Barbara Baumeister, Attila Maróti and Hung P. Tong-Viet have obtained a better lower bound on the number of conjugacy classes of a group of given order. -- Namely, in 2015 they have proved that for every $$\epsilon > 0$$ there exists a $$\delta > 0$$ such that every group of order $$n \geq 3$$ has at least $$\frac{\delta \log_2(n)}{\log_2(\log_2(n))^{3+\epsilon}}$$ conjugacy classes. -- See here. In this paper, the authors further cite a conjecture by Edward A. Bertram which asserts that the number of conjugacy classes of a finite group is bounded below by the logarithm for base $$3$$ of its order. This is basically the best one can expect to be true, as the Mathieu group $${\rm M}_{22}$$ has order $$443520 > 3^{11}$$ and only $$12$$ conjugacy classes. The authors also show that the conjecture holds for groups with trivial solvable radical.