Some time ago, I was discussing the Groups, Rings and Modules course with a friend, and this inspired me to conjecture and investigate a result in group theory.
The text is accessible to any reader with a sound grasp of standard group theory results (e.g. group actions, classification of finite abelian groups). The proofs are explained in particular detail, for the purpose of a not too taxing, enjoyable read!
Introduction
Groups encapsulate the idea of symmetry, and they hence arise in a plethora of mathematical topics.
Indeed, Cayley’s Theorem (see below) shows that every finite group is isomorphic to a subgroup of a symmetric group – that is to say, given a group of order , we can think of as a set of symmetries on objects.
But now a natural question to ask: do we really need all objects? Indeed, , the dihedral group of elements is in fact isomorphic to – it is the set of symmetries of just three objects, let alone six. But is not a subgroup of , thus it can only be embedded in . Which groups share this property – namely where we need all of our objects?
We define such a group as a Bountiful Group, and explore their properties. In fact, our final result shows that it is possible to classify them all.
Notation and Preliminaries
Cayley’s Theorem. Let be a group of order . Then is isomorphic to a subgroup of .
Cauchy’s Theorem. If divides the order of a (finite) group say, then has an element of order .
Notation:
- denotes the symmetric group of elements. A finite -group is a group of order a power of .
- The order of a group is the number of elements in it. The order of is the smallest positive integer such that (if it exists). denotes the symmetric group of elements. A finite -group is a group of order a power of .
- denotes the set of left cosets of a subgroup in a group .
Classification
Definition. A Bountiful Group is a (finite) group of order that is not isomorphic to a subgroup of for all .
In fact, it is enough that it is not isomorphic to a subgroup of since , but we choose the above definition as it reflects the motivation of symmetry given in the introduction.
We first note a useful result, and a standard result.
Claim 1. is isomorphic to a subgroup of , where are positive integers.
Proof. Consider objects, and an element of . Let act on the first objects and act on the other objects. This corresponds to a unique element of , giving an injective homomorphism say. Thus by the first isomorphism theorem.
Claim 2. If all elements of a group have order two, then is abelian.
Proof.
We now find a restriction on possible orders of bountiful groups.
Lemma. Let be a bountiful Group. Then has order a prime power.
Proof. Suppose not. Let be a bountiful group of order . Then there exist two distinct primes that both divide . By Cauchy’s Theorem, there exist two elements with orders respectively.
Let denote the cyclic group (of order ) generated by , and similarly for .
The left regular action of on induces a homomorphism . We get similarly.
Now consider the combined left regular action of on via:
.
This induces a homomorphism as follows:
Crucially, this is stronger than the natural homomorphism to .
Consider the kernel of . Suppose . Then by construction, i.e. . But due to the coprime orders of and . Hence the kernel is trivial. Thus the homomorphism is injective.
By the first isomorphism theorem, is isomorphic to a subgroup of . But by Claim 1, .
Since are distinct primes, so (smallest pair is ). , so is not a bountiful group, contradiction.
We can now prove the main result.
Theorem. Group is a bountiful group if and only if it is isomorphic to or a -group with a unique subgroup of order .
Proof.
Forward Direction.
Let be a bountiful group. The lemma from above gives that is a -group for some prime, order .
Case odd: Suppose for contradiction that does not have a unique subgroup of order . By Cauchy’s Theorem, there exists at least one subgroup of order . Thus we can find two non-equal subgroups of order in . Then , since if and shared a non-identity element, then this element would generate both groups, giving .
We now recycle the argument used in the lemma above using and . The kernel is trivial by construction of . Thus , but since odd, so is not a bountiful group, contradiction.
Case : The argument above fails, for the inequality at the end is not strict. So we need something slightly different.
If all the elements in have order 2, then is abelian by Claim 2 above. Thus by the classification of abelian groups, . But by extending the idea of Claim 1, .
If , then , so is not a bountiful group, contradiction. Else either , and indeed has a unique subgroup of order 2, or , yielding .
Backward Direction.
is not isomorphic to a subgroup of by orders (4 does not divide 6), so is a bountiful group.
Now let be a group of order , with a unique subgroup of order , say.
Claim. Every non-trivial subgroup of contains .
Proof. Let be a non-trivial subgroup of . Then divides , so by Cauchy’s Theorem contains an element of order . But then this element generates a subgroup of order in , which is also a subgroup of order in . So this subgroup is , hence .
Now suppose is isomorphic to a subgroup of for contradiction. That is to say, there is an injective homomorphism . This then induces an action of on a set of objects, say.
By the Orbit-Stabiliser Theorem, given , . Orbits partition , so . Hence .
Now is a subgroup of and is non-trivial from above. So . In other words, each element of fixes all , so . But this contradicts that is injective. So is a bountiful group.
It is natural to ask whether we can go further, to find all such groups. By perusing resources such as group order statistics (see Element structure of groups of order 32 for example), we are led to investigate the generalised quaternion groups . One can consult Keith Conrad’s excellent Generalised Quaternions article for a definition. In particular, we have a remarkable theorem:
Theorem. If a finite -group has a unique subgroup of order , then it is cyclic or generalised quaternion.
Proof. See Keith Conrad’s Generalised Quaternions article, Theorem 4.9.
Thus our final result follows:
Classification Theorem for Bountiful Groups: Let be a Bountiful Group. Then is isomorphic to one of:
- for some prime power , or
- the generalised quaternion groups .
Proof. Follows directly from the two theorems above.
Acknowledgements
Many thanks to Professor Imre Leader of Trinity College, Cambridge for his support and advice. Thanks also to two of my good friends Yiannis, for his initial investigations with me, and Weida, with whom a discussion led me to consider this idea.
And to the quaternion group , for being my canonical example to think about!