Proof
No two groups from the table are isomorphic. This is easily verified by use of the following remarks: a commutative group is not isomorphic with a non-commutative group; two isomorphic groups have the same number of elements of a given order.
We now determine the isomorphism types of the groups of order
G
of order 2.
But then g is a generator of G.
G
of order 3.
But then g is a generator of G.
G
of order 5.
But then g is a generator of G.
G
of order 7.
But then g is a generator of G.
G
of order 11.
But then g is a generator of G.
G
of order 13.
But then g is a generator of G.
It is not hard to verify that the map f:
Z/2Z × Z/2Z -> G with
(i,j) ->
aibj is an
isomorphism.
Let G be of order 6. By a previous lemma, there must be an element, say a of order 2 and an element, say b, of order 3.
Consider now the permutation representation G ->
Sym(G/<a>) of a previous
theorem. If this map is an isomorphism, then G is
isomorphic with S3. Otherwise, the kernel is equal
to <a>. But then <a> is a normal subgroup of
G and we must have bab-1 = a, that is
ab = ba. The element ab must be a generator of
the group: ab
e,
(ab)2 = a2b2 =
b2
e,
(ab)3 = a3 b3 =
a
e. Hence the
group is isomorphic with Z/6Z.
First, suppose all nontrivial elements
have order 2. Then G is commutative according to
the previous lemma. Choose three distinct elements a, b, c of
order 2 in G so that c
ab. Now K = {e,a,b,ab} is a
subgroup of order 4 and H = {e,c} a subgroup of order
2. It is easily verified that
the map Z×Z×Z -> G,
(i,j,k) -> aibjck
is an isomorphism.
Suppose now that G
has an element b of order 4. Choose, if possible, an element
a of order 2 in G\<b>. Note that G = <a,b>. As
<b> has index 2 in G, it is a normal subgroup of
G. In particular, aba-1
<b>. The element
aba-1 has order 4, so we can distinguish two cases:
aba-1 = b or aba-1 =
b-1.
If aba-1 = b,
then G is commutative. But then Z/2Z×Z/4Z -> G,
(i,j) -> aibj is an isomorphism.
From now on, suppose aba-1 = b-1. Consider the permutation representation L : G -> Sym(G/<a>) given in a previous theorem. The 4 cosets of <a> in G are <a>, b<a>, b2<a>, and b3<a>. Numbering these cosets 1, 2, 3, 4, respectively, we find Lb = (1,2,3,4) and La = (2,4). Hence, G is isomorphic to H = <(2,4),(1,2,3,4)>. In turn, this group is isomorphic with D4, the dihedral group of order 4.
Remains the case where each a
G\<b> is of order 4.
Choose such an a. Then a2 = b2,
the only element of order 2 in G. It follows that
ab
ba. For,
otherwise (ab)2 =
a2b2 = e and we have found
a second element of order 2. We conclude that aba-1
= b-1. The map f : G -> Q
with f(a) = i and f(b) = j
now is an isomorphism between G and
the quaternion group Q.
By a previous theorem, its kernel is contained in <a>. On the other hand it cannot be trivial (i.e., equal to {e}), for otherwise, the image of G under L would be a subgroup of Sym(G/<a>) of order 9, and so, by Lagrange's theorem, 9 would divide the order of Sym(G/<a>), which is 6.
Hence,
the kernel of L is <a>.
Consequently, by a previous theorem,
<a> is a normal subgroup of G.
In particular, the conjugacy class C of a is
contained in {e, a, a2}.
Clearly, e cannot be conjugate to a.
Therefore, C has at most 2 elements.
But, by the orbit theorem,
the number of elements of C is a divisor of 9, so
C = {a}. For each element b
G, we have bab-1 = a,
that is, ab = ba. In other words, a lies in the center of G.
Choose now b
G\<a>.
Since a was chosen arbitrarily amongst the nontrivial elements of G,
also b has
order 3 and lies in Z(G).
It follows that each element of
G is of the form aibj.
Letting this element correspond to (i,j)
Z/3Z×Z/3Z,
we find that G is isomorphic with Z/3Z×Z/3Z.
{1,2,3,4}.
Rewriting every occurrence of ba by akb,
we see that
We show that k = 1, 4. For, a = b2ab-2 = b(bab-1)b-1 = b(ak)b-1 = (bab-1)k = (ak)k = ak2, so k2 = 1 mod 5, whence k = 1, 4.
If k = 1, then a and b commute, and Z/5Z×Z/2Z -> G, (i,j) -> aibj, is readily seen to be an isomorphism.
If k = 4, the map
G -> D10,
determined by
a -> (1,2,3,4,5),
b -> (1,5)(2,4)
is readily seen to be an isomorphism.