Classification of Finite Abelian Groups

Math 317 C1 John Sullivan Spring 2003 

Classification of Finite Abelian Groups 

(Notes based on an article by Navarro in the Amer. Math. Monthly, February 2003.) 

The fundamental theorem of finite abelian groups expresses any such group as a product 

of cyclic groups: 

Theorem. Suppose G is a finite abelian group. Then G is (in a unique way) a direct product 

of cyclic groups of order p k with p prime. 

Our first step will be a special case of Cauchy’s Theorem, which we will prove later for 

arbitrary groups: whenever p ∣ ∣ |G| then G has an element of order p. 

Theorem (Cauchy). If G is a finite group, and p ∣ ∣ |G| is a prime, then G has an element of 

order p (or, equivalently, a subgroup of order p). 

Proof when G is abelian. First note that if |G| is prime, then G ∼ = Z p and we are done. 

In general, we work by induction. If G has no nontrivial proper subgroups, it must be a 

prime cyclic group, the case we’ve already handled. So we can suppose there is a nontrivial 

subgroup H smaller than G. Either p ∣ ∣ |H| or p ∣|G/H|. In the first case, by induction, H 

has an element of order p which is also order p in G so we’re done. In the second case, if 

g + H has order p in G/H then |g + H| ∣ ∣|g|, so 〈g〉 ∼ = Z kp for some k, and then kg ∈ G has 

order p. Note that we write our abelian groups additively. 

□ 

Definition. Given a prime p, a p-group is a group in which every element has order p k for 

some k. 

Corollary. A finite group is a p-group if and only if its order is a power of p. 

Proof. If |G| = p n then by Lagrange’s theorem, for any g ∈ G, its order divides p n , and thus 

is a (smaller) power of p. Conversely, if |G| is not a power of p, then it has some other prime 

divisor q, so by Cauchy’s theorem, G has an element of order q and thus is not a p-group. □ 

We know that in a cyclic group, any subgroup is determined uniquely by its order. Our 

first lemma proves a partial converse for p-groups. 

Lemma. If G is a finite abelian p-group and G has a unique subgroup H of order p, then G 

is cyclic. 

Proof. Again we proceed by induction on |G|, noting that the case |G| = p is obvious. Define 

φ : G → G by φ(g) = pg, and let K = ker(φ), which consists exactly of those elements of 

order p (or 1). We find that H ≤ K, so K is nontrivial. But for any nontrivial g ∈ K, the 

cyclic group 〈g〉 has order p, and thus must be H. Thus we see K = H. If K = G, then 

G ∼ = Z p is cyclic and we are done. Otherwise, φ(G) is a nontrivial proper subgroup of G, 

isomorphic to G/K. By Cauchy’s theorem, φ(G) has a subgroup of order p. Since any such 

subgroup is also a subgroup of G, there is a unique one (namely H = K). Thus we can 

apply the inductive hypothesis to the group φ(G) ∼ = G/K, and we conclude that this group 

is cyclic. If we write G/K as 〈g + K〉 for some g ≠ e, we claim that g generates G. To check 

this, it suffices to prove that K ≤ 〈g〉. But by Cauchy, 〈g〉 ≤ G has a subgroup of order p, 

which by uniqueness must be K. 

□

Math 317 C1 John Sullivan Spring 2003 p. 2 

Combining this lemma with Cauchy’s theorem, we see that a noncyclic finite abelian 

p-group has more than one subgroup of order p, which is the key to the next lemma. 

Lemma. If G is a finite abelian p-group and C is a cyclic subgroup of maximal order, then 

G = C ⊕ H for some subgroup H. 

Proof. Again, we proceed by induction on |G|, noting that when G is cyclic, C = G and 

H = {e}. When G is not cyclic, we have just shown it has more than one subgroup of 

order p, while the cyclic group C has a unique such subgroup. So let K ≤ G be a subgroup 

of order p not contained in C. Because K has prime order, K ∩ C = {e}, which implies 

(C + K)/K ∼ = C. 

Given any g ∈ G, the order of g + K in G/K divides |g|, which is at most |C|. Thus the 

cyclic subgroup (C +K)/K ∼ = C has maximal order in G/K, and we can apply the inductive 

hypothesis to prove that G/K = (C + K)/K ⊕ H ′ for some H ′ ≤ G/K. The preimage of H ′ 

under the map G → G/K is a group H with K ≤ H ≤ G. But G/K = (C + K)/K ⊕ H/K 

means that G = (C + K) + H = C + (K + H) = C + H. Since H ∩ (C + K) = K, we have 

H ∩ C = {e}, so by definition G = C ⊕ H. 

□ 

Theorem. Any finite abelian group is a direct sum of cyclic subgroups of prime-power order. 

Proof. For any prime p dividing |G|, we set G p := {g : |g| = p k } and G p ′ := {g : p ∤ |g|}. 

Then by Cauchy’s theorem, G p is nontrivial and is a p-group. Now if g ∈ G has order p k m 

(with p ̸ |m), then p k g ∈ G p ′ and mg ∈ G p . Since p k and m are relatively prime, there are r 

and s with rp k + sm = 1, so we can write g = r(p k g) + s(mg) as a sum of elements in G p ′ 

and G p . This shows that G = G p ⊕ G p ′. 

Repeating this process for the remaining primes dividing the order of G p ′ we can decompose 

G as a direct sum of p-groups for different p. So it suffices to prove the theorem for p-groups 

like G p , which have order p k . We do this by induction on k. 

Let C be a cyclic subgroup of G p of maximal order. By the last lemma, G = C ⊕ H with 

|H| < |G|. By the inductive hypothesis, H is a direct sum of cyclic subgroups, and we are 

done. 

□ 

We note that the decomposition of G given in the theorem is unique. Certainly, the 

subgroup G p is uniquely defined for any p. Now suppose a p-group G p has been expressed 

as a product of cyclic groups in two ways: as H 1 × · · · × H m and as K 1 × · · · × K n , with 

|H i | ≥ |H j | and |K i | ≥ |K j | when i < j. Then |H 1 | = |K 1 | since each of these must equal 

the maximal order of an element of G p . Proceeding by induction, we find that the two 

decompositions are really the same. 

However, we should note, for instance, that although G = Z p × Z p has no other expression 

as a product of cyclic groups, there are many pairs of subgroups H and K of order p for 

which G = H ⊕ K. In this example, for any nonzero elements a and b, we have G = 〈a〉 ⊕ 〈b〉 

unless a is a multiple of b.

Classification of Finite Abelian Groups

Create successful ePaper yourself

Delete template?

Save as template?