Semiprimitive Rings and Faithful Semisimple Modules

Q: Why is a faithful semisimple module the right notion, rather than a faithful module of some other kind?

Because rad R is characterised as the set of elements killing all simple modules. A faithful module of arbitrary type says nothing — the regular module _R R is always faithful, for every ring. Semisimplicity is what forces the annihilator to be an intersection of annihilators of simple modules, which is exactly rad R.

Q: Can the faithful semisimple module always be taken finitely generated?

No. For Z the annihilator of any finite direct sum _i Z/p_iZ is generated by p_1 … p_n ≠ 0, so infinitely many summands are unavoidable. A semiprimitive ring with a finitely generated faithful semisimple module is a subdirect product of finitely many left primitive rings, which is a genuine extra hypothesis.

Q: Is the condition left-right symmetric?

Yes, but only because rad R is. The intersection of the maximal right ideals equals the intersection of the maximal left ideals, so R has a faithful semisimple left module exactly when it has a faithful semisimple right module. The corresponding statement with *simple* in place of *semisimple* is false: Bergman constructed a left primitive ring that is not right primitive.

Q: How does this relate to Maschke's Theorem?

Maschke's Theorem says kG is a semisimple ring when G is finite and char k |G| — much stronger than semiprimitive, since the regular module itself decomposes. In the modular case char k |G|, rad(kG) ≠ 0 and by (11.1) no semisimple kG-module is faithful at all.

Q: If R is semiprimitive, is every subring semiprimitive?

No. T_2(k) sits inside M_2(k), which is semisimple hence semiprimitive, yet rad T_2(k) ≠ 0. Faithfulness of a module restricts to a subring, but semisimplicity of the module does not.

Executive Summary

The Jacobson radical is defined by an intersection — of maximal left ideals, or equivalently of annihilators of simple modules. Setting that intersection to zero is a negative condition, and negative conditions are awkward to work with. Lam's $(11.1)$ converts it into a positive one: ** $R$ is semiprimitive if and only if some semisimple left $R$ -module is faithful.**

The gain is not cosmetic. Once semiprimitivity is a statement about the existence of a module, the natural next question is how small that module can be taken — and demanding a faithful simple module instead of a faithful semisimple one is exactly the definition of a left primitive ring. The whole of §11 grows out of that one substitution.

$rad R = 0$ Semiprimitive

$ann (M) = 0$ Faithful

$⨁_{i} M_{i}$ The witness module

(11.1)Lam's number

Overview

Fix a ring $R$ with identity; modules are unital and written on the left unless stated otherwise. Recall from the Jacobson Radical: Definition and Characterisations page that

rad R = ⋂_{M simple} ann (M),

(4.2)

The radical annihilates every simple left module, and nothing else does.

Read from left to right this says $rad R$ is what all simple modules agree to kill. Read from right to left it says: if you can find enough simple modules that between them they kill nothing, then $rad R = 0$ . Packaging "enough simple modules" into a single direct sum turns the criterion into a statement about one module, and a direct sum of simple modules is precisely a semisimple module.

Two boundary points are worth fixing immediately. First, semiprimitive is much weaker than semisimple: the ring $ℤ$ is semiprimitive but $R R = ℤ$ is not a semisimple module. Second, semiprimitive is much weaker than left primitive: $ℤ$ again, since a faithful simple $ℤ$ -module would have to be some $ℤ / p ℤ$ , whose annihilator is $p ℤ \neq 0$ .

Learning Objectives

State $(11.1)$ with both implications and identify which uses $(4.1)$ and which uses $(4.2)$ .
Build the faithful semisimple module as $⨁_{i} M_{i}$ over a complete set of simple left modules.
Justify that the isomorphism classes of simple left $R$ -modules form a set, not a proper class.
Separate three conditions: $R$ semisimple, $R$ semiprimitive, $R$ left primitive.
Show $⨁_{p} ℤ / p ℤ$ is a faithful semisimple $ℤ$ -module and that no simple one is faithful.
Explain why semiprimitivity is equivalent to being a subdirect product of left primitive rings.

Definitions

Definition(4.x)Semiprimitive ring

$R$ is semiprimitive if $rad R = 0$ . The synonyms Jacobson semisimple and J-semisimple are equally standard; older literature, including parts of Jacobson's own, uses the bare word semisimple for this condition, which is a genuine reading hazard.

$ann (M)$: For a left $R$ -module $M$ , the two-sided ideal ${r \in R : r M = 0}$ .
Faithful: $ann (M) = 0$ ; equivalently the structure map $R \to End (M_{ℤ})$ is injective.
Semisimple module: A direct sum of simple submodules. Equivalently: every submodule is a direct summand. Zero counts as semisimple (empty sum).
Complete set ${M_{i}}$: One representative from each isomorphism class of simple left $R$ -modules. Every simple module is $R / 𝔪$ for a maximal left ideal $𝔪$ , so these classes form a set.
Semisimple ring: $R R$ is a semisimple module. Strictly stronger than semiprimitive: it forces $R$ to be left artinian.

Simple modules are nonzero by convention, so a nonzero ring always has at least one — take any maximal left ideal, which exists by Zorn's Lemma.

Core Concepts

Why annihilators, not submodules

The radical is measured by what a module kills, not by how the module decomposes. That is why the annihilator of a direct sum behaves so well:

ann (⨁_{i \in I} M_{i}) = ⋂_{i \in I} ann (M_{i}) .

(A)

An element kills a direct sum exactly when it kills every summand — true for arbitrary index sets.

Combining $(A)$ with $(4.2)$ gives the whole theorem in one line, provided the index set can be chosen to exhaust all isomorphism types. It can, and that is the only set-theoretic point in the proof.

The set-theoretic step

If $M$ is simple and $0 \neq m \in M$ , then $R m = M$ , so $M ≅ R / 𝔪$ where $𝔪 = ann (m)$ is a maximal left ideal. Isomorphism classes of simple left $R$ -modules are therefore indexed by a quotient of the set of maximal left ideals — a set. Without this remark, "the direct sum of all simple modules" would be meaningless.

Multiplicities are irrelevant

Since $ann (M \oplus M) = ann (M)$ , repeating a simple module changes nothing. A faithful semisimple module can always be trimmed to one whose isotypic components are single copies, and — going the other way — inflated arbitrarily. Faithfulness is a property of the set of isomorphism types occurring, not of the module.

Maximal left ideals

𝔪

Simple modules

R / 𝔪

Annihilators

ann (R / 𝔪)

rad R

Key Results

Proposition(11.1)Semiprimitivity via faithful semisimple modules

Let $R$ be a ring with identity. Then $R$ is semiprimitive — that is, $rad R = 0$ — if and only if there exists a faithful semisimple left $R$ -module $M$ .

Proof

**( $\Leftarrow$ )** Suppose $M$ is semisimple and faithful. Write $M = ⨁_{j} S_{j}$ with each $S_{j}$ simple. By $(4.1)$ , every element of $rad R$ annihilates every simple left $R$ -module, so $(rad R) S_{j} = 0$ for each $j$ and hence $(rad R) M = 0$ . Thus $rad R \subseteq ann (M) = 0$ .

**( $\Rightarrow$ )** Suppose $rad R = 0$ . If $R = 0$ take $M = 0$ , which is (vacuously) semisimple and faithful. Otherwise choose a complete set ${M_{i}}_{i \in I}$ of pairwise non-isomorphic simple left $R$ -modules; this is a genuine set because every simple module is $R / 𝔪$ for some maximal left ideal $𝔪$ . Put $M = ⨁_{i \in I} M_{i}$ , a semisimple module. Then

ann (M) = ⋂_{i \in I} ann (M_{i}) = ⋂_{N simple} ann (N) = rad R = 0,

the middle equality because every simple $N$ is isomorphic to some $M_{i}$ and isomorphic modules have equal annihilators, and the third equality by $(4.2)$ . Hence $M$ is faithful.

Corollary—Radical quotients always qualify

For any ring $R$ , the quotient $R / rad R$ admits a faithful semisimple left module. Indeed $rad (R / rad R) = 0$ by $(4.6)$ , so $(11.1)$ applies. Concretely the module is $⨁_{i} M_{i}$ over the simple $R$ -modules, which are exactly the simple $R / rad R$ -modules.

Corollary—Subdirect decomposition

$R$ is semiprimitive if and only if the natural map $R \to \prod_{i} R / ann (M_{i})$ , taken over a complete set of simple left modules, is injective. Since each $R / ann (M_{i})$ is left primitive by $(11.4)$ , a semiprimitive ring is exactly a subdirect product of left primitive rings.

Remark—What cannot be improved

Semisimple cannot be replaced by simple. A ring with a faithful simple left module is by definition left primitive, and $ℤ$ is semiprimitive without being left primitive. The gap between the two conditions is the subject of the rest of §11.

Proof Techniques and Method

How the argument works, and which move transfers to other proofs.

Move 1

Assemble a test module

To prove an intersection of annihilators vanishes, form the direct sum of the modules and prove faithfulness. Intersections of annihilators are annihilators of direct sums — always, with no finiteness hypothesis.

Move 2

Bound the class first

Before summing over "all" objects of a kind, exhibit a set that indexes them. Here: every simple module is a cyclic quotient $R / 𝔪$ .

Move 3

Trade a negative for a positive

Replace "this intersection is zero" by "a witnessing object exists". The existential form is what admits strengthening — to a simple module, to a finitely generated one, to one with extra structure.

Move 3 is the methodological content of $(11.1)$ . Almost every subsequent definition in §11 is obtained by imposing a further condition on the witnessing module: simple gives left primitivity; simple with $R$ acting as a dense ring of linear transformations gives the Density Theorem's conclusion.

Worked Example

The integers

The simple $ℤ$ -modules are the fields $ℤ / p ℤ$ for $p$ prime, pairwise non-isomorphic, and there are no others: a simple $ℤ$ -module is $ℤ / 𝔪$ for a maximal ideal $𝔪 = p ℤ$ . Take

M = ⨁_{p prime} ℤ / p ℤ, ann (M) = ⋂_{p} p ℤ = 0,

(E.1)

An integer divisible by every prime is $0$ ; hence $M$ is faithful and $ℤ$ is semiprimitive.

No finite sub-family works: $⨁_{p \leq N} ℤ / p ℤ$ has annihilator generated by the product of those primes. So the witnessing module is genuinely infinite here, and in particular is not finitely generated.

…and why $ℤ$ is not left primitive

A faithful simple $ℤ$ -module would be some $ℤ / p ℤ$ with $ann = p ℤ = 0$ , which is absurd. This is the smallest possible illustration of the fact that $(11.1)$ cannot be sharpened, and it is also an instance of $(11.8)$ : a commutative ring is primitive only if it is a field.

A ring that fails the test

Let $R = T_{2} (k)$ be the upper triangular $2 \times 2$ matrices over a field $k$ . Up to isomorphism $R$ has exactly two simple left modules, both one-dimensional over $k$ : on $S_{1}$ a matrix acts through its $(1, 1)$ entry, on $S_{2}$ through its $(2, 2)$ entry. Then

ann (S_{1} \oplus S_{2}) = ann (S_{1}) \cap ann (S_{2}) = (\begin{matrix} 0 & k \\ 0 & 0 \end{matrix}) = rad R \neq 0 .

(E.2)

Every semisimple left $R$ -module is a sum of copies of $S_{1}$ and $S_{2}$ , so every semisimple $R$ -module is killed by the strictly upper triangular matrices. No faithful semisimple module exists, exactly as $(11.1)$ predicts.

Process and Workflow

List the simple modulesClassify

R / 𝔪

for maximal left ideals

𝔪

, up to isomorphism. For algebras this usually means classifying the irreducible representations.

Compute each annihilator

ann (M_{i})

is a two-sided ideal — in fact a left primitive ideal by

(11.4)

IntersectThe intersection is

rad R

. Zero intersection means the direct sum is a faithful semisimple module.

Ask whether one sufficesIf a single

ann (M_{i})

is already zero,

R

is left primitive — a strictly stronger conclusion.

Does $R$ have a faithful semisimple left module?

Yes, and one summand suffices

R

is left primitive: it embeds densely in

End (V_{k})

for a division ring

k

, by the Structure Theorem

(11.19)

Yes, but only with several summands

R

is semiprimitive but perhaps not primitive — a subdirect product of left primitive rings.

ℤ

and

k [x]

are the model cases.

rad R \neq 0

. Pass to

R / rad R

, which always has one, and lift what you can.

Comparison and Classification

Three conditions, three witnessing modules
Condition	Witness	Extra requirement	Model example
Semiprimitive	faithful semisimple left module	none	$ℤ$
Left primitive	faithful simple left module	one isotype suffices	$End (V_{k})$ , $\dim V$ infinite
Semisimple ring	$R R$ itself is semisimple	the regular module must work	$M_{n} (D)$
Left artinian, radical zero	$R R$ semisimple	chain condition forces it	finite products of $M_{n} (D)$

Which rings satisfy which condition
	Semiprimitive	Left primitive	Semisimple ring
$ℤ$	yes	no	no
$k [x]$ , $k$ a field	yes	no	no
$M_{n} (D)$ , $D$ a division ring	yes	yes	yes
$End (V_{k})$ , $\dim_{k} V$ infinite	yes	yes	no
$k [[x]]$	no	no	no
$T_{2} (k)$ upper triangular	no	no	no
$ℤ / 6 ℤ$	yes	no	yes

Which rings satisfy which condition

Relationship Map

Reading downwards, each band strengthens the demand made on the witnessing module.

All rings

R / rad R

always has a faithful semisimple module

Semiprimitivesome faithful semisimple left module exists

Left primitivesome faithful simple left module exists

Simpleevery nonzero module is faithful

Simple artinian

R ≅ M_{n} (D)

R

simple

R

left primitive

R

semiprimitive

Neither arrow reverses. $End (V_{k})$ with $\dim_{k} V$ infinite is left primitive but not simple — the finite-rank endomorphisms form a proper nonzero ideal — and $ℤ$ is semiprimitive but not left primitive. The details are on the Primitive versus Simple and Prime page.

Applications and Industry Use

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

Representation theory

Faithfulness as a design goal

A faithful semisimple representation is what makes a group or algebra recoverable from its irreducible representations. Maschke's Theorem supplies one for $k G$ whenever $char k ∤ | G |$ , which is why ordinary character theory determines the group algebra.

Operator algebras

Why C*-algebras behave

A C*-algebra is semiprimitive, so it always admits a faithful semisimple-like representation theory in the appropriate topological sense. This is why Banach-algebraic arguments transfer to abstract rings in results such as Rickart's and Amitsur's.

Symbolic computation

Radical-first algorithms

Computer algebra systems decompose a finite-dimensional algebra by computing $rad A$ and then splitting $A / rad A$ . The output is precisely a faithful semisimple module for the quotient, presented as a list of irreducible representations.

Coding theory

Codes over semisimple quotients

Cyclic codes over a finite chain ring are analysed through the residue ring, which is semiprimitive; the faithful semisimple module is the direct sum of the constituent fields.

The honest summary: $(11.1)$ is infrastructure inside algebra. Its practical value is that it licenses the reflex "to understand $R$ , find enough irreducible representations", and tells you exactly when that reflex loses no information.

Design Considerations

Design considerations here means the choices made when modelling a problem with these algebraic structures.

Which side. Semiprimitivity is left-right symmetric because $rad R$ is, so you may test with right modules whenever they are more convenient. This licence disappears the moment you strengthen semisimple to simple.
How big a witness. Prefer the complete-set construction when you need existence, and a hand-picked finite family when you need something computable. The two agree on the answer but not on the cost.
Quotient early. If $rad R \neq 0$ , replace $R$ by $R / rad R$ before looking for simple modules: the simple modules are the same, and the quotient is guaranteed to be semiprimitive.
Do not assume finite generation. For $ℤ$ the faithful semisimple module is necessarily infinitely generated. Any argument that quietly assumes a finitely generated witness has assumed a semilocal hypothesis.

Standards and Notation

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

Preferred termsemiprimitive (Lam, Rowen)

Common variantsJacobson semisimple, J-semisimple

Hazardous variantsemisimple, meaning

rad R = 0

, in much pre-1970 writing

Radical notation

rad R

; also

J (R)

Annihilator

ann (M)

, occasionally

{Ann}_{R} (M)

(0 : M)

GAP / SageRadicalOfAlgebra, A.radical() — test for zero to certify semiprimitivity

Failure Modes and Common Mistakes

Do not write "the direct sum of all simple modules" without noting that the isomorphism classes form a set; the sum over a proper class is not a module.
Do not confuse $ann (M)$ with $ann (m)$ for a single element: the first is two-sided, the second is only a left ideal.
Do not assume that a semiprimitive ring has finitely many simple modules. That is a semilocal condition and fails for $ℤ$ and for $k [x]$ .
Do not conclude semiprimitivity from having some faithful module. Every ring has a faithful module, namely $R R$ ; the force of $(11.1)$ is entirely in the word semisimple.

Quick Reference

Statement

rad R = 0 \Leftrightarrow

some semisimple left

R

-module is faithful

Witness

M = ⨁_{i} M_{i}

, complete set of simple left modules

Key identity

ann (⨁_{i} M_{i}) = ⋂_{i} ann (M_{i})

SymmetryHolds equally for right modules

Strengtheningfaithful simple module

\Rightarrow

left primitive

Always true

R / rad R

satisfies the condition

Subdirect formsemiprimitive

\Leftrightarrow

subdirect product of left primitive rings

Checklist for applying (11.1)
Step	What to verify	Failure signal
Enumerate	Every simple left module appears up to isomorphism	A missed isotype inflates the intersection
Annihilate	$ann (M_{i})$ computed as a two-sided ideal	A left ideal answer means an element annihilator was used
Intersect	The intersection is $0$	Nonzero intersection $= rad R \neq 0$
Interpret	One factor zero $\Rightarrow$ left primitive	None zero $\Rightarrow$ semiprimitive only

Frequently Asked Questions

Why is a faithful semisimple module the right notion, rather than a faithful module of some other kind?

Because $rad R$ is characterised as the set of elements killing all simple modules. A faithful module of arbitrary type says nothing — the regular module $R R$ is always faithful, for every ring. Semisimplicity is what forces the annihilator to be an intersection of annihilators of simple modules, which is exactly $rad R$ .

Can the faithful semisimple module always be taken finitely generated?

No. For $ℤ$ the annihilator of any finite direct sum $⨁_{i} ℤ / p_{i} ℤ$ is generated by $p_{1} \dots p_{n} \neq 0$ , so infinitely many summands are unavoidable. A semiprimitive ring with a finitely generated faithful semisimple module is a subdirect product of finitely many left primitive rings, which is a genuine extra hypothesis.

Does the result hold for rings without identity?

Not in the form stated. Without an identity, maximal left ideals may fail to exist and simple modules may be absent altogether, so the direct sum in the proof can be empty while $rad R$ , defined by quasi-regularity, is nonzero. The whole of this collection assumes an identity.

Is the condition left-right symmetric?

Yes, but only because $rad R$ is. The intersection of the maximal right ideals equals the intersection of the maximal left ideals, so $R$ has a faithful semisimple left module exactly when it has a faithful semisimple right module. The corresponding statement with simple in place of semisimple is false: Bergman constructed a left primitive ring that is not right primitive.

How does this relate to Maschke's Theorem?

Maschke's Theorem says $k G$ is a semisimple ring when $G$ is finite and $char k ∤ | G |$ — much stronger than semiprimitive, since the regular module itself decomposes. In the modular case $char k ∣ | G |$ , $rad (k G) \neq 0$ and by $(11.1)$ no semisimple $k G$ -module is faithful at all.

R

is semiprimitive, is every subring semiprimitive?

No. $T_{2} (k)$ sits inside $M_{2} (k)$ , which is semisimple hence semiprimitive, yet $rad T_{2} (k) \neq 0$ . Faithfulness of a module restricts to a subring, but semisimplicity of the module does not.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §11, (11.1), pp. 182–183.
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapter I.
N. Jacobson, “The radical and semi-simplicity for arbitrary rings”, American Journal of Mathematics 67 (1945), 300–320.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §15.
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988, Chapter 2.

AI Suggested Questions

Give a ring that is semiprimitive but whose every faithful semisimple module has uncountably many isotypic components.
Show that a semiprimitive ring with only finitely many isomorphism classes of simple left modules is semilocal, and find a converse.
Which subrings of a semiprimitive ring are semiprimitive, and what hypothesis on the extension repairs the failure?
Prove that a direct product of semiprimitive rings is semiprimitive, and decide the same question for infinite products of left primitive rings.
Trace exactly where the proof of (11.1) uses the existence of an identity element.
How does the faithful semisimple module for $k G$ change as $char k$ ranges over the primes dividing the order of the group?
Construct a semiprimitive ring whose faithful semisimple module cannot be chosen with all summands isomorphic.

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Why annihilators, not submodules

The set-theoretic step

Multiplicities are irrelevant

06Key Results

07Proof Techniques and Method

Assemble a test module

Bound the class first

Trade a negative for a positive

08Worked Example

The integers

…and why ℤ is not left primitive

A ring that fails the test

09Process and Workflow

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

Faithfulness as a design goal

Why C*-algebras behave

Radical-first algorithms

Codes over semisimple quotients

13Design Considerations

14Standards and Notation

15Failure Modes and Common Mistakes

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

…and why $ℤ$ is not left primitive

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Design Considerations

Standards and Notation

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions