Semisimple Rings: Definition and Equivalent Characterisations

Q: Why bother with five conditions when one would do?

Because they serve different purposes. Condition (5) is what you verify — it involves a single module. Condition (2) is what you use — it applies to whatever module the problem hands you. Condition (1) is what makes the homological consequences immediate, and it is the form that generalises to the statements about projectives and injectives in (2.8) and (2.9).

Q: Is every subring of a semisimple ring semisimple?

No. Z sits inside the field Q, and the upper triangular matrices T_2(k) sit inside M_2(k); neither subring is semisimple. A related question — whether every ring embeds in a semisimple ring — also has answer no: semisimple rings are Dedekind-finite, that property passes to subrings, and End_k(V) for infinite-dimensional V is not Dedekind-finite.

Q: Does semisimplicity survive base change?

Not always. If A is a semisimple k-algebra and L/k is a field extension, A ⊗_k L can fail to be semisimple when L/k is inseparable. Algebras for which it never fails are called separable, and over a perfect field every semisimple algebra is separable — so in characteristic zero and over finite fields the issue does not arise.

Executive Summary

Semisimplicity of a module is a property of that module. Semisimplicity of a ring is the statement that the property is universal. Lam's $(2.5)$ makes this economical: it is enough to check the single module $_{R} R$ , and everything else follows.

The five conditions run from the strongest-sounding — all short exact sequences of left modules split — down to the weakest-sounding — the regular module is semisimple. They are all the same condition. The corollary $(2.6)$ then extracts a chain condition for free: a left semisimple ring is left artinian and left noetherian, because $1$ can only involve finitely many of the minimal left ideals.

5Equivalent conditions in (2.5)

$1 \in R$ Why the sum is finite

$23$ Characterisations listed by Rowen

BothArtinian and noetherian

Overview

The class of semisimple rings is the one class of noncommutative rings that is completely classified: by the Wedderburn–Artin theorem a ring is semisimple exactly when it is a finite direct product of matrix rings over division rings. Everything on this page is the input to that classification — the definition, its robustness, and the chain conditions it silently carries.

R left semisimple ⟺_{R} R = 𝔞_{1} \oplus \dots \oplus 𝔞_{n}, 𝔞_{i} minimal left ideals

(2.5·5)

The working definition. Finiteness of $n$ is not an assumption — it is forced by $1 \in R$ .

The adjective left is provisional. Lam defines right semisimplicity symmetrically and proves in $(3.7)$ , after the Wedderburn–Artin theorem, that the two coincide, at which point the adjective is dropped. It is worth stressing that this is a theorem and not a triviality: many neighbouring properties, including left artinian and left primitive, are genuinely one-sided.

Two nearby notions are frequently confused with this one. Semiprimitive means $rad R = 0$ and is strictly weaker — $ℤ$ is semiprimitive. Simple means no proper nonzero two-sided ideals and is neither stronger nor weaker in general, although a simple ring that is left artinian is semisimple.

Learning Objectives

State all five conditions of $(2.5)$ and reproduce the implication diagram.
Prove the only non-trivial implication, that $_{R} R$ semisimple forces all left modules semisimple.
Prove $(2.6)$ : a left semisimple ring satisfies both chain conditions on left ideals.
Show that a semisimple ring has finitely many simple left modules, all occurring in $_{R} R$ .
Test concrete rings: $ℤ / n$ , $ℚ [x] / (f)$ , $T_{2} (k)$ , $k G$ , infinite products of fields.
Explain why subrings of semisimple rings need not be semisimple, and which rings can be embedded at all.

Definitions

Definition(2.5)Left semisimple ring

A ring $R$ with identity is left semisimple if the left regular module $_{R} R$ is a semisimple module, that is, if $R$ is the sum of its minimal left ideals. Right semisimplicity is defined by the mirror condition on $R_{R}$ .

$_{R} R$: The ring $R$ regarded as a left module over itself; its submodules are exactly the left ideals of $R$ .
Left artinian: The descending chain condition on left ideals: every descending chain stabilises.
Left noetherian: The ascending chain condition on left ideals; equivalently every left ideal is finitely generated.
$rad R$: The Jacobson radical: the intersection of the maximal left ideals, equivalently of the annihilators of the simple left modules.
Dedekind-finite: $a b = 1$ implies $b a = 1$ . Every one-sided artinian or one-sided noetherian ring has this property.

Some mid-century sources use semisimple to mean what is here called semiprimitive. Check the convention before quoting a pre-1970 theorem.

Core Concepts

Why one module suffices

Every left $R$ -module is the sum of its cyclic submodules, $M = \sum_{m \in M} R m$ , and every cyclic submodule is a quotient of $_{R} R$ via $r \mapsto r m$ . Semisimplicity is inherited by quotients and is preserved by arbitrary sums, so it propagates from $_{R} R$ to every module in two steps. This is the entire content of the implication $(5) \Rightarrow (2)$ .

_{R} R semisimple \Rightarrow R m ≅ R / ann (m) semisimple \Rightarrow M = \sum_{m \in M} R m semisimple

(2.5a)

Quotient closure from $(2.2)$ , then sum closure from $(2.4) (3)$ .

Why the decomposition is finite

Write $_{R} R = ⨁_{i \in I} 𝔞_{i}$ with each $𝔞_{i}$ a minimal left ideal. The identity element has only finitely many nonzero components, say $1 = e_{i_{1}} + \dots + e_{i_{n}}$ . For any $r \in R$ we get $r = r \cdot 1 \in 𝔞_{i_{1}} + \dots + 𝔞_{i_{n}}$ , so the finite subfamily already spans and $I$ is finite. The presence of an identity is doing real work here: without it, infinite decompositions are possible.

The stock of simple modules

If $R$ is left semisimple, every simple left $R$ -module is isomorphic to one of the $𝔞_{i}$ . Indeed a simple module $M$ is cyclic, hence a quotient of $_{R} R = ⨁_{i = 1}^{n} 𝔞_{i}$ ; the composite $𝔞_{i} \to M$ must be nonzero for some $i$ , and a nonzero map between simple modules is an isomorphism. So a semisimple ring has only finitely many isomorphism classes of simple left modules — one for each Wedderburn block.

Where semisimplicity fails

** $ℤ$ ** is semiprimitive and noetherian but not artinian; $2 ℤ$ has no complement in $ℤ$ .
** $ℤ / p^{2}$ ** is artinian with $rad = (p) \neq 0$ ; the ideal $(p)$ is the unique minimal ideal and is essential, so it cannot split off.
** $T_{2} (k)$ **, upper triangular matrices, is artinian with nonzero radical: the strictly upper triangular ideal is nilpotent.
** $\prod_{i = 1}^{\infty} k$ ** has zero radical but is neither artinian nor noetherian, and is not semisimple — the direct sum $⨁_{i} k$ is an ideal with no complement.

The last two examples separate the two halves of the standard criterion: $R$ is left semisimple iff $R$ is left artinian and $rad R = 0$ . $T_{2} (k)$ has the chain condition without the radical condition; the infinite product has the radical condition without the chain condition.

Key Results

Theorem(2.5)Theorem and Definition: five equivalent conditions

For a ring $R$ with identity the following statements are equivalent.

Every short exact sequence of left $R$ -modules splits.
Every left $R$ -module is semisimple.
Every finitely generated left $R$ -module is semisimple.
Every cyclic left $R$ -module is semisimple.
The left regular module $_{R} R$ is semisimple.

A ring satisfying these conditions is called left semisimple.

Proof

**(2) $\Rightarrow$ (1).** Let $0 \to A \overset{f}{⟶} B \overset{g}{⟶} C \to 0$ be exact. Since $B$ is semisimple, the submodule $f (A)$ has a complement: $B = f (A) \oplus B'$ . Then $g$ restricted to $B'$ is injective, because $B' \cap \ker g = B' \cap f (A) = 0$ , and surjective, because $g (B) = g (f (A)) + g (B') = g (B')$ . Its inverse followed by the inclusion $B' ↪ B$ splits $g$ .

**(1) $\Rightarrow$ (2).** Given a module $M$ and a submodule $N$ , apply (1) to $0 \to N \to M \to M / N \to 0$ ; a splitting exhibits $N$ as a direct summand of $M$ . Hence $M$ is semisimple.

**(2) $\Rightarrow$ (3) $\Rightarrow$ (4) $\Rightarrow$ (5)** are successive specialisations: finitely generated modules are modules, cyclic modules are finitely generated, and $_{R} R$ is cyclic, generated by $1$ .

**(5) $\Rightarrow$ (2).** Let $M$ be any left $R$ -module and $m \in M$ . The cyclic submodule $R m$ is the image of $_{R} R$ under $r \mapsto r m$ , so it is a quotient of a semisimple module and therefore semisimple by $(2.2)$ . Being semisimple, each $R m$ is a sum of simple submodules by $(2.4)$ . Hence $M = \sum_{m \in M} R m$ is a sum of simple submodules, and $(2.4) (3) \Rightarrow (1)$ gives that $M$ is semisimple.

Corollary(2.6)Chain conditions

A left semisimple ring $R$ is both left noetherian and left artinian. More precisely $_{R} R$ has finite length, equal to the number of factors in any decomposition of $R$ into minimal left ideals.

Proof

By $(2.4)$ write $_{R} R = ⨁_{i \in I} 𝔞_{i}$ with $𝔞_{i}$ minimal left ideals. Express $1$ in this decomposition: only finitely many components are nonzero, say those indexed by $i_{1}, \dots, i_{n}$ . For arbitrary $r \in R$ , $r = r \cdot 1$ lies in $𝔞_{i_{1}} \oplus \dots \oplus 𝔞_{i_{n}}$ , so $I = {i_{1}, \dots, i_{n}}$ and the sum is finite.

The chain $0 ⊊ 𝔞_{i_{1}} ⊊ 𝔞_{i_{1}} \oplus 𝔞_{i_{2}} ⊊ \dots ⊊ R$ has simple successive quotients, so it is a composition series of length $n$ . A module of finite length satisfies both chain conditions, and for $_{R} R$ these are the ACC and DCC on left ideals.

Corollary—Finitely many simple modules

Let $R$ be left semisimple with $_{R} R = 𝔞_{1} \oplus \dots \oplus 𝔞_{n}$ as above. Then every simple left $R$ -module is isomorphic to some $𝔞_{i}$ ; in particular there are at most $n$ isomorphism classes of simple left $R$ -modules, and at least one.

Proof

A simple module $M$ is cyclic, so there is a surjection $π :_{R} R ↠ M$ . Since $π (⨁_{i} 𝔞_{i}) = \sum_{i} π (𝔞_{i}) = M \neq 0$ , some restriction $π |_{𝔞_{i}}$ is nonzero. Its kernel is a proper submodule of the simple module $𝔞_{i}$ , hence zero, and its image is a nonzero submodule of the simple module $M$ , hence all of $M$ . So $𝔞_{i} ≅ M$ .

Corollary—Semisimple rings are Dedekind-finite

If $R$ is left semisimple and $a b = 1$ in $R$ , then $b a = 1$ . Consequently every subring of a left semisimple ring is Dedekind-finite, and a ring that is not Dedekind-finite — such as ${End}_{k} (V)$ for $V$ of infinite dimension — cannot be embedded in any semisimple ring.

Proof

Consider right multiplication $λ_{a} :_{R} R \to_{R} R$ , $x \mapsto x a$ ; it is a homomorphism of left modules. It is injective: if $x a = 0$ then $x = x (a b) = (x a) b = 0$ . By $(2.6)$ the module $_{R} R$ is artinian, and an injective endomorphism of an artinian module is surjective — the chain $im λ_{a} \supseteq im λ_{a}^{2} \supseteq \dots$ stabilises, and injectivity of $λ_{a}^{n}$ then gives surjectivity of $λ_{a}$ . So $x a = 1$ for some $x$ , whence $x = x (a b) = (x a) b = b$ and $b a = 1$ .

For the last claim: if $S \subseteq R$ is a subring with the same identity and $a b = 1$ with $a, b \in S$ , then $b a = 1$ holds in $R$ and $b a \in S$ , so it holds in $S$ .

Remark—The standard external criterion

$R$ is left semisimple iff $R$ is left artinian and $rad R = 0$ . This is the criterion used in practice, and it is the bridge to the Wedderburn–Artin theorem, which upgrades it to the classification $R ≅ M_{n_{1}} (D_{1}) \times \dots \times M_{n_{r}} (D_{r})$ with the $D_{i}$ division rings, the $n_{i}$ and $D_{i}$ unique up to permutation and isomorphism.

Proof Techniques and Method

How these proofs work, and which move to reuse.

The proof of $(2.5)$ is a cycle of cheap implications closed by one substantial step. Recognising that shape saves effort whenever a list of equivalent conditions has to be established.

1. Order the conditions by apparent strengthHere: all sequences split, all modules, finitely generated modules, cyclic modules, the regular module. Each step down is a specialisation and costs nothing.

2. Identify the single hard implicationOnly

(5) \Rightarrow (2)

requires an idea. Everything else is bookkeeping, and stating this openly makes the proof readable.

3. Climb back with a generation argumentExpress an arbitrary module as a sum of cyclic ones. Since the property in question is closed under quotients and sums, it lifts from the regular module to all modules.

4. Harvest the consequencesFiniteness of the decomposition comes from

1

; the chain conditions come from finite length; the classification of simple modules comes from surjecting the regular module onto them.

The reusable move is step 3, the cyclic reduction. A property $P$ of modules that holds for $_{R} R$ , passes to quotients and is closed under sums, automatically holds for all left $R$ -modules. Projectivity, flatness and injectivity all fail one of the three closure conditions, which is exactly why the corresponding ring-level theorems in $(2.8)$ and $(2.9)$ need separate arguments.

Worked Example

Quotients of a polynomial ring

Let $R_{f} = ℚ [x] / (f)$ for a nonconstant $f \in ℚ [x]$ . By the Chinese Remainder Theorem, if $f = p_{1}^{e_{1}} \dots p_{r}^{e_{r}}$ with the $p_{i}$ distinct monic irreducibles then

ℚ [x] / (f) ≅ \prod_{i = 1}^{r} ℚ [x] / (p_{i}^{e_{i}}) .

(E.1)

A factor $ℚ [x] / (p^{e})$ is a field when $e = 1$ and has nonzero nilpotent radical $(p) / (p^{e})$ when $e \geq 2$ . Hence $R_{f}$ is semisimple exactly when $f$ is squarefree. Three instances:

Semisimplicity of $ℚ [x] / (f)$
$f$	Decomposition	$\dim_{ℚ}$	$rad$	Semisimple?
$x^{3} - x$	$ℚ \times ℚ \times ℚ$	$3$	$0$	yes
$(x^{2} + 1) (x - 1)$	$ℚ (i) \times ℚ$	$3$	$0$	yes
$x^{3} - x^{2}$	$ℚ [x] / (x^{2}) \times ℚ$	$3$	$(x) / (x^{2})$	no

In the third case the ideal $𝔫 = (x) / (x^{2})$ of the first factor is one-dimensional, satisfies $𝔫^{2} = 0$ , and is the unique minimal ideal of that factor; being essential, it has no complement, so condition $(5)$ of $(2.5)$ fails and the ring is not semisimple.

Finite rings

$ℤ / 6 ≅ ℤ / 2 \times ℤ / 3$ is semisimple: explicitly $ℤ / 6 = ⟨ 3 ⟩ \oplus ⟨ 2 ⟩$ with $⟨ 3 ⟩ = {0, 3} ≅ ℤ / 2$ and $⟨ 2 ⟩ = {0, 2, 4} ≅ ℤ / 3$ , both minimal ideals, intersecting in $0$ . By contrast $ℤ / 12$ has minimal ideals $⟨ 6 ⟩$ and $⟨ 4 ⟩$ whose sum is $⟨ 2 ⟩ \neq ℤ / 12$ , so the regular module is not the sum of its simple submodules. In general $ℤ / n$ is semisimple iff $n$ is squarefree.

A group algebra

Take $G = S_{3}$ and $k = ℚ$ . Since $char ℚ = 0$ , Maschke's theorem applies and $ℚ S_{3}$ is semisimple. Its three irreducible rational representations — trivial, sign, and the two-dimensional standard representation — give

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

(E.2)

The dimension count is the standard check on any Wedderburn decomposition of a group algebra.

Over $𝔽_{3}$ the hypothesis of Maschke's theorem fails, since $3$ divides $| S_{3} | = 6$ . The failure is visible in one line: the element $σ = \sum_{g \in S_{3}} g$ satisfies $g σ = σ g = σ$ for all $g$ , so $k σ$ is a two-sided ideal, and $σ^{2} = | S_{3} | σ = 6 σ = 0$ in characteristic $3$ . So $k σ$ is a nonzero square-zero ideal, $rad (𝔽_{3} S_{3}) \neq 0$ , and the algebra is not semisimple.

Comparison and Classification

Where the standard examples sit
	Left semisimple	Left artinian	Left noetherian	$rad R = 0$
Division ring $D$	yes	yes	yes	yes
$M_{n} (D)$	yes	yes	yes	yes
$ℤ / 6$	yes	yes	yes	yes
$ℤ / 12$	no	yes	yes	no
$ℤ$	no	no	yes	yes
$k [x]$	no	no	yes	yes
$k [[x]]$	no	no	yes	no
$T_{2} (k)$	no	yes	yes	no
$\prod_{i = 1}^{\infty} k$	no	no	no	yes
$ℚ S_{3}$	yes	yes	yes	yes
$𝔽_{3} S_{3}$	no	yes	yes	no

Where the standard examples sit

Reading the table columns: left semisimple is exactly the conjunction of columns two and four. $T_{2} (k)$ and the infinite product show that neither column alone suffices.

Constructions that preserve semisimplicity
Construction	Preserves semisimplicity?	Reason or counterexample
Finite direct product $R \times S$	yes	Modules are pairs of modules; minimal ideals of each factor still split off.
Matrix ring $M_{n} (R)$	yes	Morita equivalent to $R$ ; semisimplicity is a Morita invariant.
Quotient $R / I$	yes	$R / I$ -modules are $R$ -modules with the same submodule lattice.
Opposite ring $R^{op}$	yes	Because left and right semisimplicity coincide, by $(3.7)$ .
Subring $S \subseteq R$	no	$ℤ \subseteq ℚ$ ; also $T_{2} (k) \subseteq M_{2} (k)$ .
Polynomial ring $R [x]$	no	$R [x]$ is never artinian for $R \neq 0$ : the chain $(x) ⊋ (x^{2}) ⊋ \dots$ never stabilises.
Infinite product $\prod_{i} R_{i}$	no	$\prod_{i = 1}^{\infty} k$ is semiprimitive but not artinian.
Group algebra $R G$ , $G$ finite	conditional	Semisimple iff $R$ is semisimple and $\| G \|$ is invertible in $R$ (Maschke).

Relationship Map

Semisimple rings sit at the bottom of the hierarchy of finiteness conditions — bottom in the sense of smallest class and strongest hypothesis.

All rings with identity

rad R

defined; nothing else guaranteed

Left artinianDCC on left ideals;

rad R

nilpotent,

R / rad R

semisimple

Left semisimple

rad R = 0

as well;

R ≅ \prod_{i} M_{n_{i}} (D_{i})

Simple artinianOne block:

R ≅ M_{n} (D)

Division rings

n = 1

Fieldscommutative division rings

_{R} R

semisimpleall cyclic modules semisimpleall modules semisimpleall sequences splitall modules projective

The final node is $(2.8)$ , developed on Projective Modules, Splitting and Semisimplicity; the classification of the class itself is The Wedderburn–Artin Theorem: Structure of Semisimple Rings.

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 and idempotents

For $\gcd (n, q) = 1$ the ring $𝔽_{q} [x] / (x^{n} - 1)$ is semisimple, so every cyclic code of length $n$ is generated by an idempotent and is a direct sum of minimal ideals. Encoders and syndrome decoders are built directly from this decomposition.

Harmonic analysis

Fourier transforms on finite groups

The Wedderburn isomorphism $ℂ G ≅ \prod_{i} M_{n_{i}} (ℂ)$ is the group Fourier transform. Fast algorithms for it — Cooley–Tukey and its non-abelian generalisations — are algorithms for computing this isomorphism efficiently.

Computational chemistry

Symmetry-adapted orbitals

Molecular point groups are finite and the ground field is $ℝ$ or $ℂ$ , so the group algebra is semisimple. Projecting onto isotypic components block-diagonalises the Hamiltonian, which is what makes symmetry-adapted basis sets worth constructing.

Symbolic computation

Algebra recognition

Deciding semisimplicity of a finite-dimensional algebra is the first step of any structural computation: compute the radical, quotient by it, then apply Wedderburn–Artin. GAP, Magma and Sage all expose this pipeline.

The honest reading is that semisimplicity is a hypothesis to be secured rather than an object of study. Almost every applied use of representation theory arranges its ground field so that Maschke's theorem applies, precisely so that this page's theorem is available.

Standards and Notation

Standards here covers notation, symbol and markup standards, and reference implementations, rather than material or design codes.

Preferred termleft semisimple ring (Lam); the adjective is dropped after

(3.7)

Synonymcompletely reducible ring, in older representation-theoretic sources

False friendsemisimple meaning

rad R = 0

in pre-1970 literature — now called semiprimitive

Regular modulewritten

_{R} R

R_{R}

; some authors write

R

with a subscript on the acting side only

MarkupPresentation MathML per ISO/IEC 40314; symbol conventions per ISO 80000-2

GAPRadicalOfAlgebra, and the Wedderga package for rational group algebras

Magma / SageIsSemisimple, JacobsonRadical, A.radical()

Computational Notes

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

Finite-dimensional algebras. Semisimplicity is decided by computing $rad A$ and testing whether it is zero. In characteristic $0$ this is the radical of the trace form, one nullspace computation costing $O (n^{3})$ field operations for $n = \dim_{k} A$ ; in characteristic $p$ the Friedl–Rónyai algorithm is used, still polynomial time.
Group algebras. No computation is needed: $k G$ with $G$ finite is semisimple iff $char k ∤ | G |$ . For infinite $G$ the algebra is not artinian, so it is never semisimple.
Commutative case. A finite commutative ring is semisimple iff it is a product of fields iff it is reduced; for $ℤ / n$ this reduces to testing $n$ squarefree.
Blocks. Once semisimplicity is known, splitting $A$ into Wedderburn blocks amounts to finding the primitive central idempotents — a computation over the centre of $A$ , and the expensive part in practice because it can require factoring polynomials over the ground field.

Failure Modes and Common Mistakes

Do not assume an infinite product of semisimple rings is semisimple. $\prod_{i = 1}^{\infty} k$ fails the chain condition even though every factor is a field.
Do not apply Maschke's theorem without checking that $G$ is finite and that $| G |$ is invertible in the coefficient ring — both hypotheses are needed.
Do not conclude from $(2.5)$ that a semisimple ring has a unique decomposition into minimal left ideals. The isotypic components and the block structure are unique; the individual minimal left ideals are not.
Do not transfer one-sided results carelessly. Left semisimple equals right semisimple, but the analogous equality is false for artinian, noetherian, primitive and perfect.

Historical Notes and Lessons Learned

1893MolienMolien determines the structure of finite-dimensional associative algebras over $ℂ$ , obtaining the decomposition into matrix algebras in the complex case.
1898MaschkeMaschke proves complete reducibility of finite group representations in characteristic zero — the first widely used sufficient condition for semisimplicity.
1907WedderburnWedderburn proves the structure theorem for finite-dimensional algebras over an arbitrary field, introducing the radical as the maximal nilpotent ideal.
1927ArtinArtin replaces finite dimensionality with the descending chain condition, extending the theory to left artinian rings.
1929NoetherNoether recasts the subject in terms of modules. The definition used on this page — every submodule splits off — is her formulation, and it is why $(2.5)$ can be proved without any dimension count.
1945JacobsonJacobson's radical removes the chain condition from the definition of the obstruction, separating semisimple from semiprimitive once and for all.

The methodological lesson is the 1929 one. Wedderburn's and Artin's arguments were about ideals inside the ring; Noether's are about the category of modules over it. The module-theoretic definition is what makes semisimplicity checkable on a single object, and it is what allowed the theory to survive the removal of the chain condition twenty years later.

Quick Reference

Definition

_{R} R

is a semisimple module

Equivalentlyevery left

R

-module is semisimple

Equivalentlyevery short exact sequence of left modules splits

Equivalentlyleft artinian and

rad R = 0

Structure

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

Free consequencesleft artinian, left noetherian, Dedekind-finite

Simple modulesfinitely many, all inside

_{R} R

Sidednessleft semisimple

\Leftrightarrow

right semisimple, by

(3.7)

Fast tests for common families
Ring	Semisimple iff	Note
$ℤ / n$	$n$ squarefree	CRT into fields $ℤ / p$
$k [x] / (f)$ , $k$ a field	$f$ squarefree	CRT into fields $k [x] / (p_{i})$
$k G$ , $G$ finite	$char k ∤ \| G \|$	Maschke; never semisimple for $G$ infinite
Commutative artinian $R$	$R$ reduced	then $R$ is a finite product of fields
$M_{n} (R)$	$R$ semisimple	Morita invariance
Finite-dimensional $k$ -algebra	$rad A = 0$	chain condition is automatic

Frequently Asked Questions

Why bother with five conditions when one would do?

Because they serve different purposes. Condition $(5)$ is what you verify — it involves a single module. Condition $(2)$ is what you use — it applies to whatever module the problem hands you. Condition $(1)$ is what makes the homological consequences immediate, and it is the form that generalises to the statements about projectives and injectives in $(2.8)$ and $(2.9)$ .

Is a left semisimple ring necessarily right semisimple?

Yes, but it is a theorem rather than an observation. Lam proves it as $(3.7)$ , after the Wedderburn–Artin classification: a finite product of matrix rings over division rings is visibly symmetric under passing to the opposite ring, since $M_{n} (D)^{op} ≅ M_{n} (D^{op})$ and $D^{op}$ is again a division ring. There is no direct module-theoretic proof at the level of $(2.5)$ .

Is every subring of a semisimple ring semisimple?

No. $ℤ$ sits inside the field $ℚ$ , and the upper triangular matrices $T_{2} (k)$ sit inside $M_{2} (k)$ ; neither subring is semisimple. A related question — whether every ring embeds in a semisimple ring — also has answer no: semisimple rings are Dedekind-finite, that property passes to subrings, and ${End}_{k} (V)$ for infinite-dimensional $V$ is not Dedekind-finite.

Why is semisimple strictly stronger than semiprimitive?

Semiprimitive says only that the radical vanishes; semisimple additionally forces the descending chain condition. $ℤ$ and $\prod_{i = 1}^{\infty} k$ are both semiprimitive and neither is artinian. In the presence of the DCC the two notions coincide, which is the standard criterion: left artinian plus zero radical equals left semisimple.

How many simple modules does a semisimple ring have?

Finitely many up to isomorphism — one for each block in the Wedderburn decomposition $R ≅ \prod_{i = 1}^{r} M_{n_{i}} (D_{i})$ , so exactly $r$ . Each occurs in $_{R} R$ with multiplicity $n_{i}$ , so $\dim$ counts and length counts both reproduce $r$ from the regular module.

Does semisimplicity survive base change?

Not always. If $A$ is a semisimple $k$ -algebra and $L / k$ is a field extension, $A \otimes_{k} L$ can fail to be semisimple when $L / k$ is inseparable. Algebras for which it never fails are called separable, and over a perfect field every semisimple algebra is separable — so in characteristic zero and over finite fields the issue does not arise.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §2, results (2.5)–(2.6), pp. 27–29.
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988 — see the list of characterisations of semisimple rings in Chapter 2.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §13 (semisimple rings).
E. Artin, “Zur Theorie der hyperkomplexen Zahlen”, Abhandlungen aus dem Mathematischen Seminar der Universität Hamburg 5 (1927), 251–260.
E. Noether, “Hyperkomplexe Größen und Darstellungstheorie”, Mathematische Zeitschrift 30 (1929), 641–692.
D. S. Passman, The Algebraic Structure of Group Rings, Wiley-Interscience, 1977, Chapter 1.

AI Suggested Questions

Reconstruct the proof that a left artinian ring with zero Jacobson radical is left semisimple.
Why does the equivalence of left and right semisimplicity require the Wedderburn–Artin theorem, and is there a shorter route?
Give a ring that is right semisimple but not left artinian, or explain why none exists.
Work out the Wedderburn decomposition of the rational group algebra of the quaternion group of order eight.
Prove that semisimplicity is a Morita invariant and identify which of the five conditions in (2.5) is most convenient for that proof.
Describe an algorithm that decides semisimplicity of an algebra given by structure constants over a finite field, and estimate its cost.
What is the correct analogue of (2.5) for rings without an identity element?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Why one module suffices

Why the decomposition is finite

The stock of simple modules

Where semisimplicity fails

06Key Results

07Proof Techniques and Method

08Worked Example

Quotients of a polynomial ring

Finite rings

A group algebra

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

Cyclic codes and idempotents

Fourier transforms on finite groups

Symmetry-adapted orbitals

Algebra recognition

12Standards and Notation

13Computational Notes

14Failure Modes and Common Mistakes

15Historical Notes and Lessons Learned

16Quick Reference

17Frequently Asked Questions

18Related KEVOS Topics

19References

20AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Worked Example

Comparison and Classification

Relationship Map

Applications and Industry Use

Standards and Notation

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions