Jacobson Semisimple (Semiprimitive) Rings

Q: Is every subring of a semiprimitive ring semiprimitive?

No. k[[x]] sits inside the field k((x)) and has radical (x). Semiprimitivity is not inherited by subrings, nor by quotients; it is inherited by products, matrix rings and polynomial extensions.

Q: How do I recognise semiprimitivity for a group algebra?

For a finite group and a field k, use Maschke: kG is semisimple, hence semiprimitive, if and only if char k does not divide |G|. For infinite groups the question is much harder; in characteristic p there are group algebras with nonzero radical and group algebras without, and in characteristic zero the general case is open.

Q: Why is a local ring almost never semiprimitive?

A local ring has a unique maximal left ideal m, which is also its unique maximal right ideal, so rad R = m. That vanishes only when R has no nonzero proper left ideal, i.e. when R is a division ring. The failure is structural, not accidental: k[[x]], Z_(p) and Z/p^nZ for n ≥ 2 are all local with nonzero radical.

Executive Summary

Semisimple rings are classified completely by Wedderburn–Artin, and they are rare. Dropping the chain condition but keeping the vanishing of the radical gives the Jacobson semisimple or semiprimitive rings, and this class is large enough to contain $ℤ$ , every polynomial and free algebra over a division ring, every simple ring, every von Neumann regular ring and every $C^{*}$ -algebra.

The reason the class is useful is $(4.8)$ : the passage $R \to R / rad R$ loses no simple modules and no units. Whatever question you were asking about representations or about invertibility, you may as well ask it in a semiprimitive ring.

$rad R = 0$ Definition (4.7)

SameSimple left modules of

R

and

R / rad R

YesClosed under products and

M_{n} (-)

NoClosed under quotients or subrings

Overview

Lam introduces the definition immediately after establishing that $R / rad R$ always has zero radical. That fact — proved on The Radical of a Quotient Ring — means semiprimitive rings are not a scarce commodity to be hunted for: every ring produces one canonically.

The name semiprimitive anticipates a later theorem. A ring is left primitive when it has a faithful simple left module, and $rad R$ is the intersection of the left primitive ideals; so $rad R = 0$ says exactly that $R$ embeds as a subdirect product of left primitive rings. That is the perspective taken on Semiprime and Semiprimitive Rings as Subdirect Products.

R semiprimitive ⟺ ⋂_{𝔪 max. left} 𝔪 = 0 ⟺ ⋂_{M simple} ann (M) = 0

(4.7)

Equivalently: the simple left $R$ -modules are jointly faithful.

Learning Objectives

State Definition $(4.7)$ and give three inequivalent-looking reformulations of it.
Prove $(4.8)$ : $R$ and $\bar{R} = R / rad R$ have the same simple left modules.
Prove that $x$ is left-invertible in $R$ if and only if $\bar{x}$ is left-invertible in $\bar{R}$ .
Deduce that $U (R) \to U (\bar{R})$ is surjective and explain why the analogous statement for idempotents fails.
Show that a ring with $U (R) \cup {0}$ a division ring is semiprimitive, and apply it to free and skew polynomial algebras.
Place semiprimitive between semisimple and von Neumann regular in the implication chain.

Definitions

Definition(4.7)Jacobson semisimple ring

A ring $R$ is Jacobson semisimple, or J-semisimple, if $rad R = 0$ . The synonym semiprimitive is used interchangeably.

$\bar{R}$: Standing notation for $R / rad R$ , the radical quotient, which is semiprimitive for every $R$ .
Semisimple: $R R$ is a direct sum of simple left modules. Strictly stronger than semiprimitive; the exact gap is a chain condition, as The Hopkins–Levitzki Theorem records.
Left primitive: $R$ has a faithful simple left module. Primitive implies semiprimitive; semiprimitive is the subdirect closure of primitive.
Reduced: No nonzero nilpotent elements. For commutative affine algebras over a field, reduced and semiprimitive coincide.
Local ring: $R / rad R$ is a division ring. A local ring is semiprimitive only in the degenerate case where it is already a division ring.

Core Concepts

What the radical quotient keeps

The radical annihilates every simple left module, so every simple left $R$ -module is already a module over $\bar{R}$ , with the same lattice of submodules. Conversely every simple $\bar{R}$ -module becomes a simple $R$ -module by inflation. The two categories of simple modules are literally the same set of objects.

Simple left

R

-modulesModules killed by

rad R

Simple left

\bar{R}

-modules

The unit statement in $(4.8)$ is subtler because invertibility is not obviously detected by a quotient. It works only because the kernel is the radical: an element congruent to $1$ modulo $rad R$ is a unit, by the maximality property of the radical.

Three ways a ring can fail to be semiprimitive

Too few maximal left ideals. A local ring has exactly one, so $rad R$ is that ideal; $k [[x]]$ , $ℤ_{(p)}$ and $ℤ / p^{n} ℤ$ all fail this way.
Nilpotent structure. Any nonzero nil one-sided ideal lands inside the radical, so $k [x] / (x^{2})$ , the exterior algebra $Λ (V)$ with $\dim V \geq 1$ , and $T_{n} (k)$ are all non-examples.
Characteristic obstruction. $k G$ for a finite group $G$ with $char k$ dividing $| G |$ has nonzero radical, by Maschke's theorem read backwards.

Key Results

Proposition(4.8)What survives the radical quotient

Let $R$ be a ring and $\bar{R} = R / rad R$ , with $x \mapsto \bar{x}$ the quotient map. Then:

$R$ and $\bar{R}$ have the same simple left modules: a left $R$ -module is simple if and only if it is annihilated by $rad R$ and simple as an $\bar{R}$ -module.
$x \in R$ is left-invertible in $R$ if and only if $\bar{x}$ is left-invertible in $\bar{R}$ .
$x \in U (R)$ if and only if $\bar{x} \in U (\bar{R})$ .

Proof

(1) Every simple left $R$ -module $M$ satisfies $(rad R) M = 0$ , because $rad R$ is the intersection of the annihilators of the simple modules. So $M$ is a module over $\bar{R}$ , and its $R$ -submodules and $\bar{R}$ -submodules coincide; simplicity therefore transfers. In the other direction, inflating a simple $\bar{R}$ -module along $R ↠ \bar{R}$ gives a simple $R$ -module.

(2) If $y x = 1$ in $R$ then $\bar{y} \bar{x} = 1$ . Conversely suppose $\bar{y} \bar{x} = \bar{1}$ for some $y \in R$ . Then $1 - y x \in rad R$ , so $y x \in 1 + rad R \subseteq U (R)$ . Put $u = y x$ ; then $(u^{- 1} y) x = 1$ , so $x$ is left-invertible in $R$ .

(3) If $\bar{x}$ is a unit it is left- and right-invertible, so by (2) and its right-handed mirror $x$ has a left inverse and a right inverse in $R$ , whence $x \in U (R)$ . The converse is clear.

Corollary—Units lift

The group homomorphism $U (R) \to U (\bar{R})$ is surjective, with kernel $1 + rad R$ . Indeed if $\bar{x} \in U (\bar{R})$ then $x \in U (R)$ by $(4.8) (3)$ , and $x$ maps to $\bar{x}$ .

Proposition(4.16)–(4.17)The unit test for semiprimitivity

Let $R$ be a ring such that $S : = U (R) \cup {0}$ is a division ring. Then $R$ is semiprimitive. In particular, for a division ring $k$ the following are all semiprimitive: the free ring $k ⟨ x_{i} : i \in I ⟩$ on any set of indeterminates over $k$ , the commutative polynomial ring $k [x_{i} : i \in I]$ , and the skew polynomial rings $k [x; σ]$ and $k [x; δ]$ for any endomorphism $σ$ or derivation $δ$ of $k$ .

Proof

Suppose $0 \neq y \in rad R$ . Then $1 + y \in U (R) \subseteq S$ by the maximality property of the radical, and $1 \in S$ , so $y = (1 + y) - 1 \in S$ because $S$ is closed under subtraction. As $y \neq 0$ and $S$ is a division ring, $y \in U (R)$ . But then $rad R$ contains a unit and hence equals $R$ , contradicting $1 \notin rad R$ . So $rad R = 0$ .

For the listed algebras a degree argument gives $U (R) \cup {0} = k$ . In each case the ring is graded or filtered by degree with $\deg (f g) = \deg f + \deg g$ — for $k [x; σ]$ this uses that $σ$ is injective, automatic for a ring endomorphism of a division ring — so a product can be $1$ only if both factors have degree $0$ .

Proposition(4.24)Von Neumann regular rings are semiprimitive

If for every $a \in R$ there exists $x \in R$ with $a = a x a$ , then $rad R = 0$ .

Proof

Let $a \in rad R$ and choose $x$ with $a = a x a$ . Then $a (1 - x a) = 0$ . Since $x a \in rad R$ , the element $1 - x a$ is a unit, so $a = 0$ .

Remark—Two families worth knowing

Every simple ring is semiprimitive, since $rad R$ is a two-sided ideal and cannot be all of $R$ . And Amitsur proved that $rad (R [x]) = N [x]$ , where $N = R \cap rad (R [x])$ is a nil ideal of $R$ ; since a semiprimitive $R$ has no nonzero nil ideals, $R$ semiprimitive forces $R [x]$ semiprimitive. The converse fails: $R = k [[x]]$ has no nonzero nil ideals, so $R [y]$ is semiprimitive while $R$ is not.

Worked Example

Verifying $(4.8)$ in $ℤ / p^{n} ℤ$

Let $p$ be prime, $n \geq 2$ and $R = ℤ / p^{n} ℤ$ . The unique maximal ideal is $(p)$ , so $rad R = (p)$ and $\bar{R} = ℤ / p ℤ = 𝔽_{p}$ . The ring is not semiprimitive; the quotient is.

Simple modules. $R$ has exactly one simple module up to isomorphism, namely $R / (p) ≅ 𝔽_{p}$ , and $\bar{R} = 𝔽_{p}$ has exactly one, itself. They agree, as $(4.8) (1)$ predicts.
Units. $x \in U (R)$ if and only if $p ∤ x$ , if and only if $\bar{x} \neq 0$ in $𝔽_{p}$ , if and only if $\bar{x} \in U (𝔽_{p})$ . This is $(4.8) (3)$ made arithmetic.
Kernel of the unit map. $1 + (p)$ has $p^{n - 1}$ elements, and $| U (R) | = p^{n - 1} (p - 1) = | 1 + (p) | \cdot | U (𝔽_{p}) |$ , confirming surjectivity of $U (R) \to U (\bar{R})$ .

A semiprimitive ring with no chain conditions at all

Let $k$ be a division ring and $R = k ⟨ x, y ⟩$ , the free $k$ -ring on two indeterminates commuting with the coefficients. Every nonzero $f \in R$ has a well-defined total degree, and the top-degree components multiply without cancellation, so $\deg (f g) = \deg f + \deg g$ . Hence $f g = 1$ forces $\deg f = \deg g = 0$ , so $U (R) = k ∖ {0}$ and $U (R) \cup {0} = k$ is a division ring.

rad k ⟨ x, y ⟩ = 0,

(E.1)

even though $R$ is neither left nor right noetherian and has no nonzero idempotents other than $1$ .

Frameworks and Models

Family 1

Arithmetic rings

$ℤ$ and the ring of integers of any number field are semiprimitive: there are infinitely many nonzero prime ideals, each nonzero element lies in only finitely many, so no nonzero element survives the intersection.

Family 2

Affine commutative algebras

For a finitely generated commutative $k$ -algebra, the Nullstellensatz gives $rad R = Nil R$ . Such an $R$ is semiprimitive exactly when it is reduced.

Family 3

Free and polynomial algebras

Over a division ring $k$ : $k [x_{i}]$ , $k ⟨ x_{i} ⟩$ , $k [x; σ]$ , $k [x; δ]$ . All semiprimitive by the unit test, with no chain condition in sight.

Family 4

Regular and operator-theoretic rings

Von Neumann regular rings, including $End (V_{k})$ for any vector space, all $C^{*}$ -algebras, and the ring of continuous real functions on a compact Hausdorff space.

The families overlap only partly: $ℤ$ is not regular, $End (V_{k})$ for infinite-dimensional $V$ is regular but not noetherian, and simple rings such as the Weyl algebra $A_{1} (k)$ in characteristic zero belong to none of the first three.

Comparison and Classification

Semisimple, regular, semiprimitive and local across standard rings
	Semisimple	von Neumann regular	Semiprimitive	Local
Division ring $D$	yes	yes	yes	yes
$M_{2} (D)$	yes	yes	yes	no
$ℤ$	no	no	yes	no
$k [x]$	no	no	yes	no
$k ⟨ x, y ⟩$	no	no	yes	no
$\prod_{i = 1}^{\infty} k$	no	yes	yes	no
$End (V_{k})$ , $\dim V$ infinite	no	yes	yes	no
$k [[x]]$	no	no	no	yes
$ℤ / p^{n} ℤ$ , $n \geq 2$	no	no	no	yes
$T_{2} (k)$	no	no	no	no

Semisimple, regular, semiprimitive and local across standard rings

Closure properties of the semiprimitive class
Operation	Preserves semiprimitivity?	Reason or counterexample
Finite and infinite products	yes	$rad \prod R_{i} = \prod rad R_{i}$
Matrix rings $M_{n} (-)$	yes	$rad M_{n} (R) = M_{n} (rad R)$
Polynomial extension $R [x]$	yes	Amitsur: the radical of a polynomial ring is nil-generated
Quotients	no	$ℤ ↠ ℤ / 4 ℤ$
Subrings	no	$k [[x]] \subseteq k ((x))$
Passing to the radical quotient	yes, always	$rad (R / rad R) = 0$

Relationship Map

Semisimplevon Neumann regularSemiprimitive

Neither implication reverses: $\prod_{i = 1}^{\infty} k$ is regular and not semisimple; $ℤ$ is semiprimitive and not regular. Adding a chain condition collapses the chain — semisimple equals semiprimitive plus left artinian, and equals von Neumann regular plus left noetherian.

Semiprimitive rings — $rad R = 0$
- contains
  - all semisimple rings
  - all von Neumann regular rings
  - all simple rings
  - all left primitive rings
  - all reduced affine commutative algebras
- is contained in
  - all semiprime rings
  - rings with no nonzero nil one-sided ideals
- excludes
  - every local ring that is not a division ring
  - every nonzero ring with a nonzero nil ideal

The containment in the semiprime rings is strict and is the reason the two notions have separate subdirect decomposition theories; the comparison is made on Semiprime and Semiprimitive Rings as Subdirect Products.

Applications and Industry Use

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

Group rings

Semiprimitivity of $k G$

For $char k = p$ and $G$ finite, $k G$ is semiprimitive exactly when $p ∤ | G |$ , which is Maschke's theorem. For infinite $G$ and $char k = 0$ , Amitsur proved semiprimitivity whenever $k$ is not algebraic over $ℚ$ ; the case of fields algebraic over $ℚ$ , such as $ℚ G$ itself, is a long-standing open problem.

Operator algebras

Analytic semiprimitivity

Every $C^{*}$ -algebra has zero Jacobson radical. This is what allows purely ring-theoretic arguments about simple modules and units to be imported into functional analysis without extra hypotheses.

Symbolic computation

Recognising the reduced case

For a commutative affine algebra a CAS decides semiprimitivity by testing reducedness, a radical-ideal computation on the defining ideal. In the noncommutative finite-dimensional case the test is whether the computed radical is zero.

Continuous geometry

Regular rings

Von Neumann introduced regular rings to coordinatise continuous geometries; semiprimitivity of every regular ring is the ring-theoretic residue of that programme and is used in the theory of rings of operators.

Design Considerations

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

Reduce first, or keep the radical? If the question concerns simple modules or units, reduce immediately by $(4.8)$ . If it concerns extensions, projective covers or Loewy structure, the radical is the data you need and reducing destroys the problem.
Which side? Semiprimitivity is left-right symmetric because $rad R$ is, so the choice of side is free — unlike primitivity, which genuinely is not symmetric.
Which vanishing condition? Semiprime, semiprimitive and reduced are three different vanishing hypotheses with three different subdirect decompositions. Pick the one matching the objects you want in the decomposition: prime, primitive, or domain.
Model with a regular ring when you need pseudo-inverses. Von Neumann regularity gives $a = a x a$ and hence semiprimitivity for free; it is the natural setting for generalised inverses.

Standards and Notation

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

Lam's termJacobson semisimple, J-semisimple

Modern defaultsemiprimitive

Older usagesemisimple, meaning

rad R = 0

— check the source

Radical notation

rad R

, or

J (R)

in module-theoretic texts

Quotient notation

\bar{R}

for

R / rad R

; some authors write

R^{ss}

GAPRadicalOfAlgebra(A) returns the zero ideal exactly when

A

is semiprimitive

MarkupPresentation MathML per ISO/IEC 40314; symbols per ISO 80000-2

Failure Modes and Common Mistakes

Do not conclude semiprimitivity from the absence of nilpotent elements. $k [[x]]$ is a domain with nonzero radical.
Do not conclude non-semiprimitivity from the presence of nilpotent elements. $M_{2} (k)$ is full of them and has zero radical.
Do not assume a semiprimitive ring has many idempotents; the free algebra has none besides $0$ and $1$ .

Quick Reference

Definition

rad R = 0

Module formthe simple left modules are jointly faithful

Always semiprimitive

R / rad R

, for every

R

Unit test

U (R) \cup {0}

a division ring

\Rightarrow

semiprimitive

Implicationssemisimple

\Rightarrow

regular

\Rightarrow

semiprimitive

With left artiniansemiprimitive plus left artinian equals semisimple

Shared with $\bar{R}$ simple left modules, left-invertibility, units

Not sharedidempotents, projectives, Loewy length

Fast decisions
If your ring is…	then it is…	because
simple	semiprimitive	the radical is a proper two-sided ideal
von Neumann regular	semiprimitive	$a = a x a$ and $1 - x a$ a unit force $a = 0$
local, not a division ring	not semiprimitive	the unique maximal left ideal is the radical
a nonzero nil ring extension	not semiprimitive	nil one-sided ideals lie in the radical
$k [x_{i}]$ over a division ring	semiprimitive	units are the nonzero constants
a finite product of semiprimitive rings	semiprimitive	the radical of a product is the product of radicals

Frequently Asked Questions

Why bother with a second notion of semisimplicity?

Because the Wedderburn–Artin notion is useless without a chain condition and most rings of interest do not have one. $ℤ$ , free algebras and $C^{*}$ -algebras are all semiprimitive; none is semisimple. Semiprimitivity is what remains of semisimplicity when the finiteness is stripped away, and it is still strong enough to support a structure theory via subdirect products of primitive rings.

Does

(4.8)

mean

R

and

\bar{R}

have the same representation theory?

Only at the level of simple modules. Extensions, projective covers, injective hulls and the whole homological picture differ. For $R = k [x] / (x^{2})$ and $\bar{R} = k$ the simple modules agree but $R$ has infinite global dimension and $\bar{R}$ has dimension zero.

Is every subring of a semiprimitive ring semiprimitive?

No. $k [[x]]$ sits inside the field $k ((x))$ and has radical $(x)$ . Semiprimitivity is not inherited by subrings, nor by quotients; it is inherited by products, matrix rings and polynomial extensions.

How do I recognise semiprimitivity for a group algebra?

For a finite group and a field $k$ , use Maschke: $k G$ is semisimple, hence semiprimitive, if and only if $char k$ does not divide $| G |$ . For infinite groups the question is much harder; in characteristic $p$ there are group algebras with nonzero radical and group algebras without, and in characteristic zero the general case is open.

Why is a local ring almost never semiprimitive?

A local ring has a unique maximal left ideal $𝔪$ , which is also its unique maximal right ideal, so $rad R = 𝔪$ . That vanishes only when $R$ has no nonzero proper left ideal, i.e. when $R$ is a division ring. The failure is structural, not accidental: $k [[x]]$ , $ℤ_{(p)}$ and $ℤ / p^{n} ℤ$ for $n \geq 2$ are all local with nonzero radical.

What replaces the Wedderburn decomposition for semiprimitive rings?

The subdirect product decomposition into left primitive rings, followed by the Jacobson density theorem, which describes each primitive ring as a dense ring of linear transformations of a vector space over a division ring. That is the semiprimitive analogue of Wedderburn–Artin.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §4, (4.7)–(4.8) and Examples (1)–(9), (4.16)–(4.24) (pp. 55–66).
N. Jacobson, “The radical and semi-simplicity for arbitrary rings”, American Journal of Mathematics 67 (1945), 300–320.
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapters I–II.
D. S. Passman, The Algebraic Structure of Group Rings, Wiley-Interscience, 1977, Chapter 4.
K. R. Goodearl, Von Neumann Regular Rings, 2nd edition, Krieger, 1991, Chapter 1.
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988, Chapter 2.

AI Suggested Questions

Prove that $rad (\prod_{i} R_{i}) = \prod_{i} rad R_{i}$ and deduce that products of semiprimitive rings are semiprimitive.
Give a full proof that the ring of continuous real-valued functions on a compact Hausdorff space is semiprimitive.
State Amitsur's theorem on $rad (R [x])$ precisely and outline its proof.
Exhibit a semiprimitive ring with a non-semiprimitive matrix subring, or explain why none exists.
How does semiprimitivity of $k G$ depend on the characteristic of $k$ for locally finite groups $G$ ?
Which semiprimitive rings are subdirectly irreducible, and how does that relate to primitivity?

Jacobson Semisimple Rings

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

What the radical quotient keeps

Three ways a ring can fail to be semiprimitive

Key Results

Worked Example

Verifying $(4.8)$ in $ℤ / p^{n} ℤ$

A semiprimitive ring with no chain conditions at all

Frameworks and Models

Arithmetic rings

Affine commutative algebras

Free and polynomial algebras

Regular and operator-theoretic rings

Comparison and Classification

Relationship Map

Applications and Industry Use

Semiprimitivity of $k G$

Analytic semiprimitivity

Recognising the reduced case

Regular rings

Design Considerations

Standards and Notation

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

References

AI Suggested Questions

Continue learning

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

What the radical quotient keeps

Three ways a ring can fail to be semiprimitive

06Key Results

07Worked Example

Verifying (4.8) in ℤ/pnℤ

A semiprimitive ring with no chain conditions at all

08Frameworks and Models

Arithmetic rings

Affine commutative algebras

Free and polynomial algebras

Regular and operator-theoretic rings

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

Semiprimitivity of kG

Analytic semiprimitivity

Recognising the reduced case

Regular rings

12Design Considerations

13Standards and Notation

14Failure Modes and Common Mistakes

15Quick Reference

16Frequently Asked Questions

17Related KEVOS Topics

18References

19AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Worked Example

Verifying $(4.8)$ in $ℤ / p^{n} ℤ$

Frameworks and Models

Comparison and Classification

Relationship Map

Applications and Industry Use

Semiprimitivity of $k G$

Design Considerations

Standards and Notation

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions