The Wedderburn–Artin Theorem: Structure of Semisimple Rings

Q: Why is the number of factors equal to the number of simple modules?

Because each factor M_n_i(D_i) has exactly one simple module by (3.3)(2), the other factors annihilate it, and modules over a finite product decompose along the factors. So the simple modules of the product are exactly the simple modules of the factors, one apiece.

Q: Is the decomposition unique?

The data (r; n_1, D_1; …; n_r, D_r) are unique up to permutation and isomorphism, and the simple components are unique as ideals of R. The *isomorphism* R ≅ _i M_n_i(D_i) is not unique: composing with an inner automorphism of any factor gives another one.

Q: What happens if the group algebra is not semisimple?

Then Maschke fails, char k divides |G|, and rad kG ≠ 0. Wedderburn–Artin still applies to the quotient kG/rad kG, which is semisimple because kG is finite-dimensional hence artinian; recovering information about kG itself from that quotient is modular representation theory.

Executive Summary

The Wedderburn–Artin Theorem is the complete classification of left semisimple rings. Every one of them is a finite direct product of full matrix rings over division rings, and conversely every such product is semisimple. The number of factors and the pairs $(n_{i}, D_{i})$ are determined by the ring up to permutation.

Two consequences are as important as the statement: a left semisimple ring is automatically right semisimple, so the adjective can be dropped; and the number of factors equals the number of isomorphism classes of simple modules.

$\prod M_{n_{i}} (D_{i})$ Every semisimple ring

$r$ Factors = simple modules

1907 / 1927Wedderburn / Artin

(3.5)Lam's numbering

Overview

A ring is left semisimple if its left regular module decomposes as a direct sum of simple submodules. That definition, adopted in Semisimple Rings: Definition and Equivalent Characterisations, is intrinsic and side-dependent in appearance. The theorem on this page shows it is neither mysterious nor side-dependent.

R left semisimple ⟺ R ≅ M_{n_{1}} (D_{1}) \times M_{n_{2}} (D_{2}) \times \dots \times M_{n_{r}} (D_{r})

(3.5)

$D_{1}, \dots, D_{r}$ division rings, $n_{1}, \dots, n_{r} \geq 1$ , $r$ finite. Right to left is $(3.3)$ plus $(3.4)$ ; left to right is the theorem.

The proof is short because two earlier results do the work. Schur's Lemma supplies the division rings, and the identification $End (_{R} R) ≅ R$ turns a module decomposition into a ring decomposition. That is the entire argument: decompose the regular module, take endomorphism rings of both sides.

What the theorem does not do is classify division rings. That problem is genuinely hard — it contains the Brauer group of every field — so Wedderburn–Artin reduces semisimple ring theory to division ring theory rather than eliminating it.

Learning Objectives

State $(3.5)$ with the finiteness of $r$ and the uniqueness of the pairs $(n_{i}, D_{i})$ .
Prove $(3.4)$ : a finite product of left semisimple rings is left semisimple.
Give the endomorphism-ring proof of the decomposition, citing Schur's Lemma where it is used.
Prove the uniqueness half using the Jordan–Hölder theorem.
Deduce $(3.7)$ : left semisimple is equivalent to right semisimple.
Apply the dimension identity $\dim_{k} R = \sum_{i} n_{i}^{2} \dim_{k} D_{i}$ to a group algebra.

Definitions

Left semisimple ring: $_{R} R$ is a direct sum of simple left ideals. Equivalently every left $R$ -module is semisimple; equivalently every short exact sequence of left $R$ -modules splits.
$n \cdot V$: The direct sum of $n$ copies of the module $V$ ; also written $V^{\oplus n}$ .
Simple components: The ideals $B_{1}, \dots, B_{r}$ with $R = B_{1} \oplus \dots \oplus B_{r}$ and each $B_{i} ≅ M_{n_{i}} (D_{i})$ as a ring.
Jordan–Hölder theorem: Any two composition series of a module of finite length have the same length and the same multiset of composition factors up to isomorphism.
$End (_{R} R)$: The endomorphism ring of the left regular module. With endomorphisms written on the right it is isomorphic to $R$ via $r \mapsto (x \mapsto x r)$ .

Rings have an identity. The zero ring is vacuously semisimple with r = 0; all statements below assume R is nonzero.

Core Concepts

Grouping the minimal left ideals

Let $R$ be left semisimple. Then $_{R} R$ is a sum of minimal left ideals, and since $1$ lies in a finite subsum, the sum can be taken finite. Collect the summands by isomorphism type:

_{R} R ≅ n_{1} V_{1} \oplus n_{2} V_{2} \oplus \dots \oplus n_{r} V_{r}, V_{i} not ≅ V_{j} for i \neq j,

(A)

The isotypic decomposition of the regular module. $r$ and the multiplicities $n_{i}$ are finite.

Every simple left $R$ -module is a quotient of $_{R} R$ , hence a composition factor of it, hence isomorphic to one of the $V_{i}$ by Jordan–Hölder. So ${V_{1}, \dots, V_{r}}$ is a complete and irredundant list of the simple left $R$ -modules — the first appearance of the number $r$ .

Taking endomorphism rings

Apply $End$ to both sides of $(A)$ . On the left, $End (_{R} R) ≅ R$ : with endomorphisms written opposite the scalars, every endomorphism of the regular module is right multiplication by a unique ring element, and composition corresponds to multiplication in the correct order.

On the right, Schur's Lemma makes $D_{i} = End (_{R} V_{i})$ a division ring, kills all homomorphisms between non-isomorphic $V_{i}$ and $V_{j}$ , and leaves the block-diagonal matrix rings:

R ≅ End (⨁_{i = 1}^{r} n_{i} V_{i}) ≅ \prod_{i = 1}^{r} End (n_{i} V_{i}) ≅ \prod_{i = 1}^{r} M_{n_{i}} (D_{i}) .

The whole proof, in one line.

Why the multiplicities are matrix sizes

It is worth pausing on the coincidence that makes the theorem clean: the multiplicity $n_{i}$ of $V_{i}$ in the regular module becomes the size of the $i$ -th matrix block. Read back through $(3.3)$ this is consistent — for $R = M_{n} (D)$ , the regular module is $n$ copies of the column space, and the block size is $n$ .

Why sides stop mattering

Each $M_{n_{i}} (D_{i})$ is right semisimple as well as left semisimple by the right-handed half of $(3.3)$ , and a finite product of right semisimple rings is right semisimple by the right-handed half of $(3.4)$ . So the class of rings characterised on the left coincides with the class characterised on the right, and one may speak of semisimple rings without qualification.

Key Results

Proposition(3.4)Finite products stay semisimple

Let $R_{1}, \dots, R_{r}$ be left semisimple rings, each with identity. Then the direct product $R = R_{1} \times \dots \times R_{r}$ is left semisimple. (Finiteness of $r$ is essential.)

Proof

Write $_{R_{i}} R_{i} = 𝔄_{i 1} \oplus \dots \oplus 𝔄_{i m_{i}}$ with each $𝔄_{i j}$ a minimal left ideal of $R_{i}$ . View $R_{i}$ as an ideal of $R$ via the $i$ -th coordinate embedding. Because the other factors of $R$ annihilate $R_{i}$ , a subset of $R_{i}$ is an $R$ -submodule exactly when it is an $R_{i}$ -submodule; hence each $𝔄_{i j}$ is also a minimal left ideal of $R$ .

Since $R = R_{1} \oplus \dots \oplus R_{r}$ as left $R$ -modules, we get $_{R} R = ⨁_{i, j} 𝔄_{i j}$ , a finite direct sum of minimal left ideals. So $R$ is left semisimple.

Theorem(3.5)Wedderburn–Artin

Let $R$ be a nonzero left semisimple ring. Then there exist an integer $r \geq 1$ , division rings $D_{1}, \dots, D_{r}$ and positive integers $n_{1}, \dots, n_{r}$ with

R ≅ M_{n_{1}} (D_{1}) \times M_{n_{2}} (D_{2}) \times \dots \times M_{n_{r}} (D_{r}) .

The integer $r$ is uniquely determined, and so are the pairs $(n_{1}, D_{1}), \dots, (n_{r}, D_{r})$ up to a permutation of the indices and isomorphism of the division rings. Moreover $R$ has exactly $r$ isomorphism classes of simple left modules, and $n_{i}$ is the multiplicity of the $i$ -th of them in $_{R} R$ .

Proof

Existence. Decompose $_{R} R$ into a finite direct sum of minimal left ideals and group them by isomorphism type as in $(A)$ : $_{R} R ≅ n_{1} V_{1} \oplus \dots \oplus n_{r} V_{r}$ with the $V_{i}$ pairwise non-isomorphic simple modules. Any simple left $R$ -module is isomorphic to $R / 𝔪$ for a maximal left ideal $𝔪$ , hence is a composition factor of $_{R} R$ , hence isomorphic to some $V_{i}$ by Jordan–Hölder; so the list is complete, giving the count of simple modules.

Now compute endomorphism rings, writing endomorphisms of left modules on the right. First, $End (_{R} R) ≅ R$ via $r \mapsto ρ_{r}$ , $x \mapsto x r$ : this map is additive, bijective (an endomorphism is determined by the image of $1$ ), and multiplicative in the right-operator convention since $ρ_{r} ρ_{s} = ρ_{r s}$ .

Second, put $D_{i} = End (_{R} V_{i})$ , a division ring by Schur's Lemma $(3.6)$ . Since ${Hom}_{R} (V_{i}, V_{j}) = 0$ for $i \neq j$ , again by Schur, the endomorphism ring of the direct sum is the product of the endomorphism rings of the isotypic pieces, and $End (n_{i} V_{i}) ≅ M_{n_{i}} (D_{i})$ . Combining,

R ≅ End (_{R} R) ≅ \prod_{i = 1}^{r} M_{n_{i}} (D_{i}) .

Uniqueness. Suppose also $R ≅ M_{m_{1}} (D'_{1}) \times \dots \times M_{m_{s}} (D'_{s})$ . Let $V'_{i}$ be the unique simple left module of the factor $M_{m_{i}} (D'_{i})$ , given by $(3.3) (2)$ , regarded as a left $R$ -module through the projection. The remaining factors annihilate $V'_{i}$ , so $V'_{i}$ is a simple $R$ -module, and $V'_{i} not ≅ V'_{j}$ for $i \neq j$ because they have different annihilators. By $(3.3)$ and $(3.4)$ ,

_{R} R ≅ m_{1} V'_{1} \oplus \dots \oplus m_{s} V'_{s} .

Comparing with $(A)$ through Jordan–Hölder: the two decompositions of $_{R} R$ have the same composition factors with the same multiplicities, so $s = r$ and, after reindexing, $m_{i} = n_{i}$ and $V'_{i} ≅ V_{i}$ for all $i$ . Finally, by $(3.3) (3)$ applied inside the factor $R_{i} : = M_{m_{i}} (D'_{i})$ , and because $R$ -endomorphisms of $V'_{i}$ coincide with $R_{i}$ -endomorphisms (the other factors act as zero),

D'_{i} ≅ End (_{R_{i}} V'_{i}) = End (_{R} V'_{i}) ≅ End (_{R} V_{i}) = D_{i} .

Corollary(3.7)Left–right symmetry

A ring is left semisimple if and only if it is right semisimple. Consequently the adjectives left and right may be omitted, and one speaks simply of semisimple rings.

Proof

If $R$ is left semisimple then $R ≅ \prod_{i} M_{n_{i}} (D_{i})$ by $(3.5)$ . Each $M_{n_{i}} (D_{i})$ is right semisimple by the right-handed statement of $(3.3) (1)$ — the proof is the transpose of the left-handed one, using rows instead of columns — and a finite product of right semisimple rings is right semisimple by the right-handed form of $(3.4)$ . So $R$ is right semisimple. The converse is the mirror argument.

Corollary(3.5a)Finite-dimensional algebras (Wedderburn, 1907)

Let $k$ be a field and $R$ a finite-dimensional semisimple $k$ -algebra. Then $R ≅ \prod_{i = 1}^{r} M_{n_{i}} (D_{i})$ with each $D_{i}$ a finite-dimensional division algebra over $k$ , and

\dim_{k} R = \sum_{i = 1}^{r} n_{i}^{2} \dim_{k} D_{i} .

If moreover $k$ is algebraically closed then every $D_{i} = k$ , so $R ≅ M_{n_{1}} (k) \times \dots \times M_{n_{r}} (k)$ and $\dim_{k} R = \sum_{i} n_{i}^{2}$ .

Proof

Each $D_{i} ≅ End (_{R} V_{i})$ contains $k$ centrally and is a $k$ -subquotient of ${End}_{k} (V_{i})$ , hence finite-dimensional over $k$ . The dimension identity is the dimension of a product of matrix algebras. If $k$ is algebraically closed and $D$ is a finite-dimensional division $k$ -algebra, then for $d \in D$ the commutative subalgebra $k [d]$ is a finite field extension of $k$ , hence equals $k$ ; so $D = k$ .

Proof Techniques and Method

How this proof works, and which moves to reuse.

1. Decompose the regular moduleSemisimplicity plus finite generation gives

_{R} R =

a finite direct sum of minimal left ideals.

2. Group by isomorphism typeCollect the summands into isotypic blocks

n_{i} V_{i}

. Jordan–Hölder guarantees the grouping is well defined and that the

V_{i}

exhaust the simple modules.

3. Apply Schur

D_{i} = End (V_{i})

is a division ring and cross-homomorphisms vanish, so the endomorphism ring is block diagonal.

4. Use $End (_{R} R) ≅ R$ Reading the computation backwards turns the module statement into a ring isomorphism. This step is where the right-operator convention earns its keep.

5. Uniqueness by Jordan–HölderAny second decomposition produces a second isotypic decomposition of the same module; compare composition factors.

The reusable move is step 4. Whenever an object is recoverable as the endomorphism ring of a well-understood module over itself, structural information about that module transfers to the object. Morita theory is the systematic version, and Deriving Wedderburn–Artin from the Density Theorem is a different route to the same conclusion.

Worked Example

The rational group algebra of $S_{3}$

$| S_{3} | = 6$ is invertible in $ℚ$ , so $ℚ S_{3}$ is semisimple by Maschke's theorem. It has three simple modules over $ℚ$ : the trivial module, the sign module, and the two-dimensional standard module $V$ (the permutation module on three points modulo the diagonal). The standard module is absolutely irreducible, so $End (V) = ℚ$ . Hence

ℚ S_{3} ≅ ℚ \times ℚ \times M_{2} (ℚ), 1 + 1 + 4 = 6 = \dim_{ℚ} ℚ S_{3} .

(E.1)

Here $r = 3$ , $(n_{i}) = (1, 1, 2)$ , and every $D_{i} = ℚ$ .

When the division rings are not the ground field

Take $G = C_{4} = ⟨ g ⟩$ and $k = ℚ$ . Then $ℚ C_{4} ≅ ℚ [x] / (x^{4} - 1)$ , and $x^{4} - 1 = (x - 1) (x + 1) (x^{2} + 1)$ with all three factors irreducible over $ℚ$ . So

ℚ C_{4} ≅ ℚ \times ℚ \times ℚ (i), 1 + 1 + 2 = 4,

(E.2)

All $n_{i} = 1$ ; the third division ring is the field $ℚ (i)$ , strictly larger than the ground field.

Over $ℂ$ the same group algebra becomes $ℂ^{4}$ : the number of factors changed from $3$ to $4$ when the field grew. The decomposition depends on the ground field, not only on the group.

A semisimple ring from linear algebra

Let $T$ be a linear operator on a finite-dimensional $k$ -vector space with minimal polynomial $m (x) = m_{1} (x)^{e_{1}} \dots m_{t} (x)^{e_{t}}$ , the $m_{i}$ distinct irreducibles. Then

k [T] ≅ k [x] / (m (x)) ≅ \prod_{i = 1}^{t} k [x] / (m_{i} (x)^{e_{i}}),

(3.14)

and each factor is a field precisely when $e_{i} = 1$ . So $k [T]$ is semisimple exactly when $m (x)$ is squarefree — the classical criterion for a semisimple operator, which over an algebraically closed field reads diagonalisable.

Comparison and Classification

What is true for which class of rings
	$rad R = 0$	left artinian	product of matrix rings	left = right
Semisimple	yes	yes	yes	yes
Semiprimitive, e.g. $ℤ$	yes	no	no	yes
Left artinian, e.g. $T_{2} (k)$	no	yes	no	no
Simple artinian	yes	yes	yes, $r = 1$	yes
Simple non-artinian, e.g. $A_{1} (ℂ)$	yes	no	no	yes
Commutative semisimple	yes	yes	product of fields	yes

What is true for which class of rings

Reading the decomposition off a semisimple ring
Question	Answer in terms of the decomposition
How many simple left modules?	$r$ , one for each factor
What are they?	$V_{i} = D_{i}^{n_{i}}$ , the column space of the $i$ -th factor
Composition length of $_{R} R$ ?	$n_{1} + n_{2} + \dots + n_{r}$
When is $R$ simple?	$r = 1$
When is $R$ commutative?	all $n_{i} = 1$ and all $D_{i}$ fields
What is the centre?	$Z (D_{1}) \times \dots \times Z (D_{r})$ , a product of $r$ fields
How many two-sided ideals?	$2^{r}$ , one for each subset of the factors

Relationship Map

Wedderburn–Artin sits at the junction of three chains of results: the construction of examples, the classification, and the reduction of the general case to the semisimple one.

Upstream

What it needs

Schur's Lemma $(3.6)$ — supplies the $D_{i}$
$(3.3)$ — the model ring $M_{n} (D)$
$(3.4)$ — closure under finite products
Jordan–Hölder — uniqueness

Downstream

What it gives

$(3.7)$ left–right symmetry
$(3.8)$ – $(3.9)$ uniqueness of the simple components
$(3.10)$ , $(3.13)$ simple artinian rings
Maschke plus Wedderburn: representation theory of finite groups

Generalisations

Where it goes

Jacobson's structure theorem for primitive rings
Morita equivalence of module categories
Artinian rings via $R / rad R$
Finite-dimensional $C^{*}$ -algebras

R

left artinian

R / rad R

semisimple

\prod M_{n_{i}} (D_{i})

lift back through

rad R

That chain is why the theorem matters far beyond semisimple rings: for any left artinian ring, the semisimple quotient is completely known, and the remaining work is confined to the radical.

Applications and Industry Use

Applications here means where this structure is used — inside mathematics and in the engineering and computing disciplines that consume it.

Coding theory

Cyclic codes

For $\gcd (n, q) = 1$ the algebra $𝔽_{q} [x] / (x^{n} - 1)$ is semisimple, so by Wedderburn–Artin — here commutative, hence a product of fields — cyclic codes of length $n$ correspond exactly to subsets of the irreducible factors of $x^{n} - 1$ . Every classical BCH and Reed–Solomon construction is a choice of such a subset.

Quantum information

Decoherence-free subspaces

A finite-dimensional $C^{*}$ -algebra is $⨁_{i} M_{n_{i}} (ℂ)$ , the $C^{*}$ -version of $(3.5)$ . Noise-commuting subsystems are identified by decomposing the commutant of the noise operators into such blocks; the multiplicity spaces are the protected subsystems.

Representation theory

The group Fourier transform

For $char k ∤ | G |$ , $k G ≅ \prod_{i} M_{n_{i}} (D_{i})$ , and over $ℂ$ this is the statement that the Fourier transform on $G$ is an isomorphism onto a product of matrix algebras. Fast algorithms for non-abelian transforms compute exactly this decomposition.

Symbolic computation

Algebra recognition

Given a finite-dimensional algebra by structure constants, the standard pipeline is: compute the radical, quotient, then find the Wedderburn decomposition of the semisimple quotient. GAP and Magma expose this directly, and it underlies module decomposition and isomorphism testing.

The honest summary is that Wedderburn–Artin is not applied to a physical system; it is applied to the algebra that a physical or combinatorial problem generates, and it tells you that algebra has no hidden structure beyond a list of matrix sizes and division rings.

Computational Notes

Computational notes cover algorithms, cost and library behaviour rather than manufacturing process.

For a semisimple algebra of dimension $N$ over a finite field, the decomposition into simple components can be computed in polynomial time; the central primitive idempotents are found from the centre, which is a commutative algebra of dimension $r$ .
The centre is the cheapest entry point: $Z (R) ≅ Z (D_{1}) \times \dots \times Z (D_{r})$ is a product of $r$ fields, so computing $Z (R)$ and splitting it into fields immediately yields $r$ and the central idempotents.
Over $ℚ$ , deciding whether a given simple algebra is isomorphic to $M_{n} (ℚ)$ — the explicit isomorphism problem — is as hard as certain factorisation problems; existence of the decomposition is easy, exhibiting it is not.
Once the central idempotents $e_{1}, \dots, e_{r}$ are known, arithmetic parallelises: computation in $R$ splits into $r$ independent computations in the $e_{i} R$ , each a matrix algebra.

Failure Modes and Common Mistakes

$n_{i}$ is the multiplicity of $V_{i}$ in $_{R} R$ and equals $\dim_{D_{i}} V_{i}$ — not $\dim_{k} V_{i}$ unless $D_{i} = k$ .
Maschke's theorem requires $G$ finite and $char k ∤ | G |$ ; for modular group algebras Wedderburn–Artin applies only after quotienting by the radical.
$r$ is the number of simple modules, which over a non-algebraically-closed field is generally smaller than the number of conjugacy classes for a group algebra.
The theorem classifies the ring, not its modules over a subring, and it says nothing about how two semisimple rings can be embedded in one another.

Historical Notes and Lessons Learned

1893MolienTheodor Molien decomposes finite-dimensional complex algebras into matrix blocks, obtaining the algebraically closed case before the general language exists.
1898MaschkeMaschke proves that a group algebra of a finite group in non-dividing characteristic is semisimple, supplying the most important source of examples.
1907WedderburnJ. H. M. Wedderburn classifies finite-dimensional algebras over an arbitrary field: modulo the largest nilpotent ideal, they are products of matrix algebras over division algebras.
1927ArtinEmil Artin extends the structure theory to rings with the descending chain condition on left ideals, replacing finite dimension by a chain condition.
1939Hopkins and LevitzkiLeft artinian with identity is shown to imply left noetherian, removing the ascending chain condition Artin had also assumed.
1945JacobsonThe radical is defined for arbitrary rings and the density theorem generalises the structure theory to primitive rings, exhibiting Wedderburn–Artin as the artinian special case.
1958MoritaMorita's theory explains the theorem categorically: $R$ and $M_{n} (R)$ have equivalent module categories, and semisimple rings are exactly those Morita equivalent to a finite product of division rings.

The methodological lesson is the replacement of a hypothesis by a weaker one at each step: finite dimension becomes DCC, DCC becomes the existence of a faithful simple module, and finally the ring-theoretic hypothesis becomes a categorical one. Each weakening kept the same proof skeleton — decompose a module, take endomorphisms.

Quick Reference

Theorem

R

semisimple

\Leftrightarrow

R ≅ \prod_{i = 1}^{r} M_{n_{i}} (D_{i})

Uniqueness

r

and the pairs

(n_{i}, D_{i})

, up to permutation

Simple modulesexactly

r

; the

i

-th is

V_{i} = D_{i}^{n_{i}}

Multiplicity

n_{i}

= copies of

V_{i}

_{R} R

\dim_{D_{i}} V_{i}

Division rings

D_{i} ≅ End (_{R} V_{i})

, by Schur

Symmetryleft semisimple

\Leftrightarrow

right semisimple

(3.7)

Dimension identity

\dim_{k} R = \sum_{i} n_{i}^{2} \dim_{k} D_{i}

Centre

Z (R) ≅ \prod_{i} Z (D_{i})

, a product of

r

fields

Idealsexactly

2^{r}

, all sums of simple components

Commutative casea finite product of fields

Standard decompositions
Ring	Decomposition	Check
$ℂ S_{3}$	$ℂ \times ℂ \times M_{2} (ℂ)$	$1 + 1 + 4 = 6$
$ℚ S_{3}$	$ℚ \times ℚ \times M_{2} (ℚ)$	$1 + 1 + 4 = 6$
$ℚ C_{4}$	$ℚ \times ℚ \times ℚ (i)$	$1 + 1 + 2 = 4$
$ℝ Q_{8}$	$ℝ^{4} \times ℍ$	$1 + 1 + 1 + 1 + 4 = 8$
$𝔽_{2} [x] / (x^{3} - 1)$	$𝔽_{2} \times 𝔽_{4}$	$1 + 2 = 3$

Frequently Asked Questions

Does Wedderburn-Artin require the ring to be an algebra over a field?

No. It applies to any left semisimple ring with identity. Finite dimension over a field is the special case Wedderburn treated in 1907; Artin's contribution was to replace it by the descending chain condition, and the modern definition of semisimplicity removes even that from the statement.

Why is the number of factors equal to the number of simple modules?

Because each factor $M_{n_{i}} (D_{i})$ has exactly one simple module by $(3.3) (2)$ , the other factors annihilate it, and modules over a finite product decompose along the factors. So the simple modules of the product are exactly the simple modules of the factors, one apiece.

Is the decomposition unique?

The data $(r; n_{1}, D_{1}; \dots; n_{r}, D_{r})$ are unique up to permutation and isomorphism, and the simple components are unique as ideals of $R$ . The isomorphism $R ≅ \prod_{i} M_{n_{i}} (D_{i})$ is not unique: composing with an inner automorphism of any factor gives another one.

What happens if the group algebra is not semisimple?

Then Maschke fails, $char k$ divides $| G |$ , and $rad k G \neq 0$ . Wedderburn–Artin still applies to the quotient $k G / rad k G$ , which is semisimple because $k G$ is finite-dimensional hence artinian; recovering information about $k G$ itself from that quotient is modular representation theory.

Can the division rings be arbitrary?

Any finite list of division rings and positive integers occurs, since the product $\prod_{i} M_{n_{i}} (D_{i})$ is always semisimple by $(3.3)$ and $(3.4)$ . So the theorem is a genuine classification but not a finiteness statement: classifying division rings over a given field is the theory of the Brauer group, and over $ℚ$ or a function field it is a deep subject.

How does this relate to the density theorem?

Jacobson's density theorem says a primitive ring acts densely on a vector space over a division ring; when a chain condition is present, dense becomes onto, and one recovers $M_{n} (D)$ . That route is developed in Deriving Wedderburn–Artin from the Density Theorem and gives the same classification with a different proof.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §3, (3.4)–(3.7) (pp. 34–37).
J. H. M. Wedderburn, “On hypercomplex numbers”, Proceedings of the London Mathematical Society (2) 6 (1908), 77–118.
E. Artin, “Zur Theorie der hyperkomplexen Zahlen”, Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg 5 (1927), 251–260.
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapter IV.
C. W. Curtis and I. Reiner, Representation Theory of Finite Groups and Associative Algebras, Interscience, 1962, §25–§26.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §13.

AI Suggested Questions

Prove that an infinite product of fields is never semisimple, and identify which characterisation of semisimplicity fails.
Decompose $ℚ [Q_{8}]$ and compare it with $ℝ [Q_{8}]$ and $ℂ [Q_{8}]$ .
How does the number of Wedderburn factors of $k G$ vary as $k$ ranges over $ℚ$ , $ℝ$ and $ℂ$ for a fixed finite group?
State and prove Maschke's theorem, and show where the hypothesis on the characteristic is used.
What is the Brauer group of a field, and how does it parametrise the possible $D_{i}$ in a Wedderburn decomposition?
Give the density-theorem proof of Wedderburn–Artin and compare the two arguments.
Which properties of a semisimple ring can be read off from the multiset of pairs $(n_{i}, D_{i})$ , and which cannot?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Grouping the minimal left ideals

Taking endomorphism rings

Why the multiplicities are matrix sizes

Why sides stop mattering

06Key Results

07Proof Techniques and Method

08Worked Example

The rational group algebra of S3

When the division rings are not the ground field

A semisimple ring from linear algebra

09Comparison and Classification

10Relationship Map

What it needs

What it gives

Where it goes

11Applications and Industry Use

Cyclic codes

Decoherence-free subspaces

The group Fourier transform

Algebra recognition

12Computational Notes

13Failure Modes and Common Mistakes

14Historical Notes and Lessons Learned

15Quick Reference

16Frequently Asked Questions

17Related KEVOS Topics

18References

19AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Worked Example

The rational group algebra of $S_{3}$

Comparison and Classification

Relationship Map

Applications and Industry Use

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions