ProofComplete

Proof of the Jordan-Holder Theorem

The Jordan-Holder theorem establishes the uniqueness of composition factors, showing that the "building blocks" of a group are well-defined regardless of how the group is decomposed.


Statement

Theorem10.1Jordan-Holder theorem

Let GG be a group with a composition series. Any two composition series of GG have the same length, and the composition factors (up to isomorphism and permutation) are the same.


Proof

Proof

We prove the theorem by induction on the length nn of a composition series.

Base case (n=0n = 0): G={e}G = \{e\}, trivially true.

Base case (n=1n = 1): GG is simple. Any composition series is {e}G\{e\} \trianglelefteq G with one factor GG. Unique.

Inductive step: Let {e}=G0G1Gn=G\{e\} = G_0 \trianglelefteq G_1 \trianglelefteq \cdots \trianglelefteq G_n = G and {e}=H0H1Hm=G\{e\} = H_0 \trianglelefteq H_1 \trianglelefteq \cdots \trianglelefteq H_m = G be two composition series.

Case 1: Gn1=Hm1G_{n-1} = H_{m-1}. Then the two series share the same maximal normal subgroup, and by induction on Gn1G_{n-1}, the portions below Gn1G_{n-1} have the same factors. The top factors are both G/Gn1G/G_{n-1}, so the full series are equivalent.

Case 2: Gn1Hm1G_{n-1} \neq H_{m-1}. Since both are maximal normal subgroups of GG (as the quotients are simple), and Gn1Hm1G_{n-1} \neq H_{m-1}, the product Gn1Hm1G_{n-1} H_{m-1} is a normal subgroup of GG strictly containing both. By maximality: Gn1Hm1=GG_{n-1} H_{m-1} = G.

Let K=Gn1Hm1K = G_{n-1} \cap H_{m-1}. By the second isomorphism theorem:

G/Gn1=Gn1Hm1/Gn1Hm1/K,G/G_{n-1} = G_{n-1} H_{m-1}/G_{n-1} \cong H_{m-1}/K, G/Hm1=Gn1Hm1/Hm1Gn1/K.G/H_{m-1} = G_{n-1} H_{m-1}/H_{m-1} \cong G_{n-1}/K.

Now KK has composition series (being a subgroup of Gn1G_{n-1}). Let {e}=K0K=K\{e\} = K_0 \trianglelefteq \cdots \trianglelefteq K_\ell = K be one.

Then:

  • {e}=K0K=KGn1G\{e\} = K_0 \trianglelefteq \cdots \trianglelefteq K_\ell = K \trianglelefteq G_{n-1} \trianglelefteq G is a composition series (factors of KK, then Gn1/KG/Hm1G_{n-1}/K \cong G/H_{m-1}, then G/Gn1G/G_{n-1}).
  • {e}=K0K=KHm1G\{e\} = K_0 \trianglelefteq \cdots \trianglelefteq K_\ell = K \trianglelefteq H_{m-1} \trianglelefteq G is a composition series (factors of KK, then Hm1/KG/Gn1H_{m-1}/K \cong G/G_{n-1}, then G/Hm1G/H_{m-1}).

By induction on Gn1G_{n-1}: the original series through Gn1G_{n-1} has the same factors as the series through KK then Gn1G_{n-1}. By induction on Hm1H_{m-1}: similarly for the series through Hm1H_{m-1}.

Comparing: both original series have the same multiset of composition factors as the "bridge" series through KK (the factors of KK, plus G/Gn1G/G_{n-1} and G/Hm1G/H_{m-1}, which simply swap positions). \blacksquare


Corollary

RemarkWell-definedness of composition factors

The Jordan-Holder theorem means we can speak of "the composition factors of GG" without ambiguity. For a finite group GG, the composition factors (with multiplicities) are a complete invariant of the "simple building blocks" of GG. However, the composition factors do NOT determine GG up to isomorphism: many non-isomorphic groups share the same composition factors (e.g., Z/4\mathbb{Z}/4 and Z/2×Z/2\mathbb{Z}/2 \times \mathbb{Z}/2 both have factors {Z/2,Z/2}\{\mathbb{Z}/2, \mathbb{Z}/2\}). The extension problem -- classifying groups with given composition factors -- is one of the hardest problems in group theory.