Introduction
Perhaps one of the most interesting things in group theory is to consider whether two given groups are in fact the same. Even if different symbols are used to denote the elements and operations of the given groups, how can we know if the groups in fact have identical structure? This is precisely the function of group isomorphisms. If there exists an isomorphism between two groups, then we say that the two groups are isomorphic and this means that despite any notational differences, the two perhaps seemingly different groups are in actuality the same. The word isomorphism is derived from the Ancient Greek: ἴσος/isos meaning “equal”, and μορφή/morphe meaning “form” or “shape”.

For two groups G, and H, we say that \alpha: G \rightarrow H is a group isomorphism if \alpha(g_1g_2)=\alpha(g_1)\alpha(g_2), \forall g_1,g_2 \in G and \alpha is bijective. In other words, \alpha is a bijective homomorphism.

Theorem: On any collection of groups, isomorphism is an equivalence relation

Proof: By considering the function \alpha: G \rightarrow given by \alpha(g)=g, which is an isomorphism, it’s clear that reflexivity is satisfied. Now for symmetry. Suppose that \alpha: G \rightarrow H is an isomorphism. There exists a function \beta: G \rightarrow H given by\beta(g)=h where \alpha(g)=h. It is clear that \beta is bijective so now all we have to do is verify that it is a homomorphism. Take any h_1,h_2 \in H and suppose that \beta(h_i)=g_i. Then \alpha(g_1g_2)=\alpha(g_1)\alpha(g_2)=h_1h_2; thus \beta(h_1h_2)=g_1g_2=\beta(h_1)\beta(h_2), as desired. Now we’ll verify transitivity. Suppose that \alpha: G \rightarrow H and \beta: H \rightarrow  K are isomorphisms. Let \gamma=\beta \circ \alpha. It’s rather clear that \gamma is a bijection (if this isn’t clear to you then take the time to convince yourself of it) so all that remains is to check that it’s a homomorphism. Take any g_1,g_2 \in G. Then \gamma(g_1g_2)=\beta(\alpha(g_1)\alpha(g_2))=\beta(\alpha(g_1))\beta(\alpha(g_2))=\gamma(g_1)\gamma(g_2) as desired. \quad \blacksquare

Hence it makes sense to say that two groups are isomorphic; there is no ambiguity apropos of whether G is isomorphic to H or H is isomorphic to G. When two groups G and H are isomorphic, we write G \cong H

Examples
Time for some examples. Let’s being by showing that \mathbb{Z}_3 \times \mathbb{Z}_7 and \mathbb{Z}_{21} are isomorphic. We define \alpha: \mathbb{Z}_{21} \rightarrow \mathbb{Z}_3 \times \mathbb{Z}_7 via \alpha(a)=(a,a). We first must verify that this is well-defined. If a,b \in \mathbb{Z}_{21}, then 21|(a-b) and thus 3|(a-b) and 7|(a-b). Therefore, (a,a)=(b,b) in \mathbb{Z}_3 \times \mathbb{Z}_7 and hence the operation is indeed well-defined. Now let’s verify that it’s a homomorphism. If a,b \in \mathbb{Z}_{21} then \alpha(a+b)=(a+b,a+b)=(a,a)+(b,b)=\alpha(a)+\alpha(b). Now we’ll check that the function is injective. If a \in ker(\alpha), then (a,a)=(0,0). That is 3|a and 7|a. Therefore, 21|a and so a=0 in \mathbb{Z}_{21}. Thus ker(\alpha)=\{0\}. As shown in my group homomorphism post, \alpha is injective if and only if ker(\alpha)=\{0\}, and thus \alpha must be injective. It’s rather clear to see that \alpha is surjective as the 21 elements of \mathbb{Z}_{21} map to 21 different elements of \mathbb{Z}_3 \times \mathbb{Z}_7, but \mathbb{Z}_3 \times \mathbb{Z}_7 only has 21 elements, and so \alpha must be surjective.

Note that it is however not generally true that \mathbb{Z}_m \times \mathbb{Z}_n is isomorphic to \mathbb{Z}_{mn}. For example, \mathbb{Z}_4 \times \mathbb{Z}_6 is not isomorphic to \mathbb{Z}_{24} as 1 has order 24 in \mathbb{Z}_{24} however since 12(a,b)=(0,0) for every (a,b) in \mathbb{Z}_4 \times \mathbb{Z}_6, every element must have an order dividing 12, and so the two groups can’t be isomorphic.

Here’s another example. The set G=\{1,-1,i,-i\} is a group under multiplication that is in fact isomorphic to the additive group \mathbb{Z}_4. Since we’re dealing with groups with only a few elements, in this case it’s plausible to simply define what each individual element gets mapped to in the isomorphism. We define \alpha:\mathbb{Z}_4 \rightarrow G via \alpha(0)=1, \alpha(1)=i, \alpha(2)=-1 and \alpha(3)=-i. The function is clearly bijective and it is perhaps easiest to verify that it respects the group operations by comparing group tables.

A little inspection reveals that these groups both have group tables of the following form:

General Results
Now I’ll present some more general results, many of which are elegantly surprising.

Theorem: Let G=\langle a \rangle be a cyclic group. If a has infinite order, then G \cong \mathbb{Z} and if a has order n<\infty, then G \cong \mathbb{Z}_n

Proof: Let G be infinite cyclic. Define \alpha: \mathbb{Z} \rightarrow G via \alpha(i)=a^{i}. We’ll show that \alpha is an isomorphism. If i,j \in \mathbb{Z}, then \alpha(i+j)=a^{i+j}=a^{i}a^{j}=\alpha(i)\alpha(j), as required. If i \in ker(\alpha), then \alpha^{i}=e=\alpha^{0}. Hence ker(\alpha)=\{0\} and thus \alpha is injective. Furthermore, if i \in \mathbb{Z}, then \alpha(i) \in \alpha(\mathbb{Z}), and so \alpha is surjective too. Therefore \alpha is an isomorphism.
Now suppose that |a|=n<\infty. Define \alpha: \mathbb{Z}_n \rightarrow G via \alpha(i)=a^{i}. In this case, we must verify that \alpha is well-defined. But if i=j in \mathbb{Z}_n then n|(i-j) and hence a^{i}=a^{j}. The fact that \alpha is a surjective homomorphism follows as above. If i \in ker(\alpha) then a^{i}=e and thus n|i. Hence, i=0 in \mathbb{Z}_n and so \alpha is injective. Consequently, \alpha is an isomorphism as desired. \quad \blacksquare

An elegant corollary is that if a group G has prime order p, then G \cong \mathbb{Z}_p. This follows since any group of prime order is in fact cyclic, as shown in my post on Lagrange’s theorem. Take a moment to ruminate on this. Given any group of prime order p , we know that its structure is as simple as addition over the integers modulo p. No matter how peculiar the group operation may seem and no matter how unfamiliar the notation for the elements might be, when dealing with a group of prime order p, we are actually dealing with a group that we are in fact deeply familiar with.

Lemma: Let G be a group having distinct commuting elements a and b of order 2. Then G has a subgroup isomorphic to \mathbb{Z}_2 \times \mathbb{Z}_2.

Proof: Given the conditions upon a and b, we can see that H=\{e,a,b,ab\} is a subgroup (it contains the identity and it’s easy to verify closure). Also, the four elements are distinct. (It’s clear that e, a and b are dinstinct. If ab=e=bb then a=b. If ab=a=ae then b=e. If ab=b=eb then a=e. All of these configurations are impossible). Our claim is that H \cong \mathbb{Z}_2 \times \mathbb{Z}_2. Let \alpha: H \rightarrow  \mathbb{Z}_2 \times \mathbb{Z}_2 be given by \alpha(e)=(0,0), \alpha(a)=(1,0), \alpha(b)=(0,1) and \alpha(ab)=(1,1). It’s clear that this function is bijective and running through all possible pairs of group elements, we see that it is also a homomorphism. \quad \blacksquare

Corollary: Every group G of order 4 is isomorphic to either \mathbb{Z}_4 or \mathbb{Z}_2 \times \mathbb{Z}_2

Proof: In a group of order 4, every nonidentity element has order 2 or 4. If there’s an element of order 4 then G is cyclic, and by the previous theorem, is isomorphic to \mathbb{Z}_4. Otherwise, every nonidentity element has order 2. A group in which every element has order 1 or 2 is in fact always abelian (the proof is left as an exercise for the reader) and hence by what we showed in the previous lemma, G has a subgroup isomorphic to \mathbb{Z}_2 \times \mathbb{Z}_2. Given the order of the group, the proof is complete. \quad \blacksquare

Standard Results
Now for some results that are less elegant but are still undoubtedly fundamental in aiding the consolidation of our understanding of isomorphisms.

Proposition: Let G and H be isomorphic groups. Then:

1. G is abelian if and only if H is abelian
2. G is cyclic if and only if H is cyclic
3. For any positive integer n, G and H have the same number of elements of order n
4. For any positive integer n, G and H have the same number of subgroups of order n

Proof: Let \alpha: G \rightarrow H be an isomorphism.

(1) Suppose that G is abelian. Then \forall g_1,g_2 \in G, \alpha(g_1g_2)=\alpha(g_1)\alpha(g_2)=\alpha(g_1g_2)=\alpha(g_2g_1)=\alpha(g_2)\alpha(g_1). And since \alpha(G)=H, this means that H is abelian too. But the same can be said for \alpha^{-1}: H \rightarrow G.

(2) Suppose that G= \langle a \rangle. Then for any \alpha(g) \in \alpha(G), we have that g=a^{m} for some m \in \mathbb{Z}. Then \alpha(g)=\alpha(a^{m})=(\alpha(a))^{m}. Thus H=\alpha(G)=\langle \alpha(a) \rangle. But the same can be said for \alpha^{-1}: H \rightarrow G

(3) Take g \in G of order n. As shown in my group homomorphism post, |\alpha(g) divides |g|. But by the same argument, |g|=|\alpha^{-1}(\alpha(g))| divides |\alpha(g)|. Thus |g|=|\alpha(g)|. That is, the elements of order n in G are in one-to-one correspondence with the elements of order n in H

(4) Let L be a subgroup of G of order n. Then \alpha(L) is a subgroup of H and is isomorphic to L. Hence, it has the same order. If M is some other subgroup of G, then since \alpha is injective, \alpha(M) is a different group. Thus H has at least as many subgroups of order n as G does. But applying \alpha^{-1}, we see that G has at least as many subgroups of order n as H does. The result follows. \quad \blacksquare

Trending