generalized quaternion group
The groups given by the presentation
are the generalized quaternion groups. Generally one insists that as the properties of generalized quaternions become more uniform at this stage.However if then one observes so .Dihedral group properties are strongly related to generalized quaterniongroup properties because of their highly related presentations. We will see this in many of our results.
Proposition 1.
- 1.
.
- 2.
is abelian
if and only if .
- 3.
Every element can be written uniquely as where and .
- 4.
.
- 5.
.
Proof.
Given the relation (rather treating it as ) thenas with dihedral groups
we canshuffle words in to group all the at the beginning and the at the end. So every word takes the form . As and we have and . However we have anadded relation that so we can write andalso so we restrict to . This gives us elementsof this form which makes the order of at most .
As it follows . So is central. If we quotient by then we havethe presentation
which we recognize as the presentation of the dihedral group. Thus. This prove the order of is exactly . Moreover, given we have
So we have . So . Then forces .This means .∎
1 Examples
As mentioned, if then . If then we have theusual quaternion group . Because of the genesis of quaternions
, this group is often denoted with relations as follows:
These relations are responsible for many useful results such as defining cross products for three-dimensional manipulations, and are also responsible for themost common example of a division ring. As a group, is a curious specimen of a -group in that it has only normal subgroups yet is non-abelian
, it has a unique minimal
subgroup
and cannot be represented faithfully except by a regular representation
– thus requiring degree 8. [To see this note that the unique minmal subgroup is necessarily normal, thus if a proper subgroup
is the stabilizer
of an action, then the minimal normal subgroup is in the kernel so the representation
is not faithful.]
A common work around is to use matrices over but to treat these as matrices over .
A worthwhile additional example is . For thisproduces a group order 12 which is often overlooked.
2 Subgroup structure
Proposition 2.
is Hamiltonian – meaning all a non-abelian group
whose subgroups are normal – if and onlyif .
Proof.
As , then if is Hamiltonian then we require to be as well. However when we know has non-normal subgroups,for example . So we require . If then is cyclic and so trivially Hamiltonian. When we have the usual quaterniongroup of order 8 which is Hamiltonian by direct inspection: the conjugacy classesare , , , and , more commonlydescribed by , , , and . In any case,all subgroups are normal.∎
By way of converse it can be shown that the only finite Hamiltonian groups are where is abelian without an element of order 4.One sees already in that the subgroup is conjugate to the distinct subgroup and so such groups are not Hamiltonian.
Proposition 3.
- 1.
for and for all .
- 2.
Every subgroup of is either cyclic or a generalizedquaternion.
- 3.
The normal subgroups of are either subgroups of or and it is maximal subgroups (of index 2) ofwhich there are 2 acyclic ones.
Proof.
The order of elements of follows from standard cyclic group theory.Now for we simply compute: .So .
Now let be a subgroup of . If then isa subgroup of . We know the subgroups of are either cyclic ordihedral. If is cyclic then is cyclic (indeed it is a subgroupof or ). So assume that isdihedral. Then we have a dihedral presentation for . Now pullbackthis presentation to and we find is quaternion.
Finally, if does not contain then does not contain an elementof the form , so and so it is cyclic.
For the normal subgroup structure, from the relation wesee is normal. Thus all subgroups of are normal as is a normal cyclic subgroup. Next suppose is a normal subgroup not contained in . Then containssome , and so contains . Thus is a normalsubgroup of . We know this forces to be contained in, a contradiction
on our assumptions
on , or and is a maximal subgroup (of index 2).∎
Proposition 4.
has a unique minimal subgroup if and only if .
Proof.
If and then has order and so the subgroup is of order , so it is minimal. As the centeris also a minimal subgroup of order 2, then we do not have a unique minimal subgroupin these conditions. Thus .
Now suppose then is a -group so the minimal subgroups must allbe of order 2. So we locate the elements of order 2. We have shown forany , and furthermore that . The only other minimal subgroupswill be generated by for some , and as there is a unique minimalsubgroup.∎
It can also be shown that any finite group with a unique minimal subgroup is eithercyclic of prime power order, or for some .We note that these groups have only regular
faithful representations
.