Ring Class Hierarchy: Semisimple to Semilocal and Everything Between

Q: Where do local rings sit in the hierarchy?

Every local ring is semiperfect, because R/rad R is a division ring and division rings have only the idempotents 0 and 1, which lift trivially. Local rings can be artinian (k[x]/(x^2)), perfect but not artinian, or merely semiperfect (k[[x]]). Locality is a condition on the shape of the quotient, orthogonal to the nilpotence conditions on the radical.

Executive Summary

There are two independent gradings of rings in this subject and they should not be confused. The finiteness axis runs semisimple $\Rightarrow$ artinian $\Rightarrow$ semiprimary $\Rightarrow$ perfect $\Rightarrow$ semiperfect $\Rightarrow$ semilocal, and measures how much of the Wedderburn picture survives. The primeness axis runs division ring $\Rightarrow$ simple $\Rightarrow$ primitive $\Rightarrow$ prime $\Rightarrow$ semiprime, and measures how far the ring is from having a nontrivial two-sided ideal structure.

Each consecutive inclusion on both axes is strict, and this page supplies a witness for each. The two axes meet only under a chain condition: for left artinian rings, semisimple, semiprimitive and semiprime coincide, and simple, primitive and prime coincide $(11.7)$ .

6Classes on the finiteness axis

5Classes on the primeness axis

0Inclusions that reverse

$(11.7)$ Where the axes collapse

Overview

The hierarchy exists because Wedderburn–Artin is too rigid to be applied directly. A semisimple ring is completely known, but semisimplicity is a very strong hypothesis, so the theory proceeds by weakening it in a controlled sequence. At each stage two things are given up in step: something about $rad R$ , and something about how much of $R / rad R$ can be pulled back into $R$ .

What each weakening gives up
Class	Condition on $rad R$	Condition on $R / rad R$	Lifting available
Semisimple	$rad R = 0$	Equal to $R$ ; semisimple	Nothing to lift
Left artinian	Nilpotent $(4.12)$	Semisimple	Idempotents lift
Semiprimary	Nilpotent	Semisimple	Idempotents lift
Right perfect	Right T-nilpotent	Semisimple	Idempotents lift; projective covers exist
Semiperfect	No condition	Semisimple	Idempotents lift
Semilocal	No condition	Semisimple	None guaranteed

For the fine structure of the radical conditions see The Radicals of a Ring Compared; for the finiteness conditions see Chain Conditions: A Reference.

Learning Objectives

State the definitions of semilocal, semiperfect, perfect, semiprimary, artinian and semisimple in a single common format.
Prove that a semiprimary ring is both left and right perfect.
Produce a witness for every strict inclusion on the finiteness axis.
State Bass's Theorem P with its four equivalent conditions and note the side-switch.
Explain why the primeness axis collapses under the descending chain condition.
Classify a given concrete ring against both axes.

Definitions

Definition(20.1)Semilocal

$R$ is semilocal if $R / rad R$ is left artinian, equivalently if $R / rad R$ is semisimple. The equivalence is $(4.14)$ applied to the quotient, whose radical is zero by $(4.6)$ .

Definition(23.1)Semiperfect

$R$ is semiperfect if $R$ is semilocal and every idempotent of $R / rad R$ lifts to an idempotent of $R$ .

Definition(23.13), (23.18)T-nilpotent and perfect

A subset $A \subseteq R$ is right T-nilpotent if for every sequence $a_{1}, a_{2}, a_{3}, \dots$ of elements of $A$ there exists $n$ with $a_{1} a_{2} \dots a_{n} = 0$ ; left T-nilpotent if instead $a_{n} \dots a_{2} a_{1} = 0$ for some $n$ . The ring $R$ is right perfect if $R / rad R$ is semisimple and $rad R$ is right T-nilpotent, and left perfect if $R / rad R$ is semisimple and $rad R$ is left T-nilpotent. $R$ is semiprimary if $R / rad R$ is semisimple and $rad R$ is nilpotent.

Local ring: $R \neq 0$ and $R / rad R$ is a division ring; equivalently the non-units form an additive subgroup.
Simple ring: $R \neq 0$ with no two-sided ideals other than $0$ and $R$ .
Left primitive ring: $R$ has a faithful simple left module.
Semiprime ring: $𝔄^{2} = 0$ implies $𝔄 = 0$ for two-sided ideals; equivalently ${Nil}_{*} R = 0$ .
Semiprimitive ring: $rad R = 0$ ; also called Jacobson semisimple.

Core Concepts

Why nilpotence is replaced by T-nilpotence

Nilpotence of $J = rad R$ asserts that some fixed $n$ kills all products of length $n$ . T-nilpotence asks much less: for each individually chosen sequence, some initial product vanishes, with $n$ depending on the sequence. The weaker condition is exactly what is needed for the two facts that matter — Nakayama's Lemma for arbitrary (not just finitely generated) modules, and the existence of projective covers.

J

nilpotent

J

right T-nilpotent

J

nilno condition

The middle implication is genuine but not reversible; and right T-nilpotent does not imply left T-nilpotent, which is exactly why one-sided perfectness is a real distinction.

Why lifting idempotents matters

If $R$ is semilocal then $R / rad R ≅ \prod_{i} M_{n_{i}} (D_{i})$ , so the quotient has a rich supply of orthogonal idempotents. Being able to lift them means the decomposition $1 = {\bar{e}}_{1} + \dots + {\bar{e}}_{m}$ into primitive orthogonal idempotents can be realised in $R$ itself, and then $R = ⨁_{i} e_{i} R$ decomposes the ring. Without lifting, the quotient's decomposition is information about $R / rad R$ only, and tells you nothing about $R$ .

Key Results

Proposition(23.19)Semiprimary implies perfect on both sides

Let $R$ be semiprimary: $J = rad R$ is nilpotent and $R / J$ is semisimple. Then $R$ is both left perfect and right perfect. In particular every left or right artinian ring is left and right perfect.

Proof

Suppose $J^{n} = 0$ . Given any sequence $a_{1}, a_{2}, \dots$ in $J$ , the product $a_{1} a_{2} \dots a_{n}$ lies in $J^{n} = 0$ , so it vanishes for the fixed index $n$ ; hence $J$ is right T-nilpotent. The same computation read in the other order gives $a_{n} \dots a_{1} \in J^{n} = 0$ , so $J$ is left T-nilpotent. Since $R / J$ is semisimple by hypothesis, both definitions of perfectness are met. For the last claim, a left artinian ring has $rad R$ nilpotent by $(4.12)$ , and $R / rad R$ is left artinian with zero radical by $(4.6)$ , hence semisimple by $(4.14)$ .

Theorem(23.20)Bass's Theorem P

For any ring $R$ the following are equivalent:

$R$ is right perfect — that is, $R / rad R$ is semisimple and $rad R$ is right T-nilpotent;
$R$ satisfies the descending chain condition on principal left ideals;
every left $R$ -module satisfies the descending chain condition on cyclic submodules;
$R$ contains no infinite set of nonzero orthogonal idempotents, and every nonzero left $R$ -module contains a simple submodule.

A fifth equivalent condition, proved separately in $(24.25)$ , is that every flat right $R$ -module is projective.

Remark

The switch from right in (1) to left in (2) and (3) is not a typographical accident. The name right perfect is chosen so that right perfect rings are exactly those over which every right module has a projective cover, which is the property the definition is designed to capture.

Theorem(23.10)Semiperfect with simple quotient

For a ring $R$ the following are equivalent: (1) $R$ is semiperfect and $R / rad R$ is a simple ring; (2) $R ≅ M_{n} (k)$ for some local ring $k$ . When these hold, $n$ is uniquely determined, $k$ is unique up to isomorphism, and $R$ is indecomposable as a ring.

Proposition(11.7)The axes collapse under DCC

Let $R$ be left artinian. Then $R$ is semisimple $\Leftrightarrow$ $R$ is semiprimitive $\Leftrightarrow$ $R$ is semiprime; and $R$ is simple $\Leftrightarrow$ $R$ is left primitive $\Leftrightarrow$ $R$ is right primitive $\Leftrightarrow$ $R$ is prime.

Proof

For the first chain, semisimple $\Leftrightarrow$ semiprimitive is $(4.14)$ given the DCC, and semiprimitive $\Leftrightarrow$ semiprime holds because a left artinian ring has $rad R$ nilpotent $(4.12)$ : a nilpotent ideal lies in ${Nil}_{*} R$ , so ${Nil}_{*} R = 0$ forces $rad R = 0$ , and the reverse inclusion ${Nil}_{*} R \subseteq rad R$ is $(10.14)$ . For the second chain, simple $\Rightarrow$ left primitive $\Rightarrow$ prime holds in any ring $(11.6)$ , and the return implication prime $\Rightarrow$ simple uses the DCC: in a prime left artinian ring $rad R$ is a nilpotent ideal, hence zero, so $R$ is semisimple and therefore a finite product of simple rings; primeness rules out more than one factor.

Worked Example

Placing five concrete rings

Take $k$ a field and $p \neq q$ distinct primes. Each ring below is classified against both axes; the arithmetic is elementary and worth checking.

Five rings, fully classified
Ring	$rad$	Finiteness class	Primeness class
$M_{2} (ℚ)$	$0$	Semisimple	Simple, primitive, prime
$T_{2} (k)$ , upper triangular	strictly upper triangular	Left and right artinian; semiprimary; perfect; semiperfect	Not semiprime: the strictly upper triangular ideal is nonzero and squares to zero
$k [[x]]$	$(x)$	Local, hence semiperfect; not perfect	Prime (a domain); not primitive, since it is commutative and not a field
$S = ℤ_{(p)} \cap ℤ_{(q)}$	$p q S$	Semilocal; not semiperfect	Prime (a domain)
$ℤ$	$0$	None of the classes; semiprimitive	Prime; not primitive

Verifying the semilocal but not semiperfect entry

Let $S$ be the localisation of $ℤ$ at the multiplicative set of integers coprime to $p q$ . Then $S$ is a principal ideal domain with exactly two maximal ideals $p S$ and $q S$ , so

rad S = p S \cap q S = p q S, S / rad S ≅ 𝔽_{p} \times 𝔽_{q},

(E.1)

The quotient is semisimple, so $S$ is semilocal.

The quotient contains the idempotent $(1, 0)$ , which is neither $0$ nor $1$ . But $S$ is an integral domain, so $e^{2} = e$ forces $e (e - 1) = 0$ and hence $e \in {0, 1}$ . The only idempotents available to lift to are $0$ and $1$ , whose images are $(0, 0)$ and $(1, 1)$ . Therefore $(1, 0)$ does not lift and $S$ is not semiperfect. This also matches $(23.11)$ : $S$ is commutative and is not a finite direct product of local rings, being a domain with two maximal ideals.

Process and Workflow

Where does your ring sit on the finiteness axis?

$R / rad R$ not semisimpleThe ring is below the whole hierarchy. Nothing on this page applies; use the prime and primitive theory of §10–§12 instead.

Semisimple quotient, idempotents do not liftSemilocal only. Typical of commutative domains with finitely many maximal ideals.

Idempotents lift, radical not T-nilpotentSemiperfect. Projective covers exist for finitely generated modules but not for all modules.

Radical right T-nilpotent, not nilpotentRight perfect. Every right module has a projective cover; flat right modules are projective

(24.25)

Radical nilpotentSemiprimary. Hopkins–Levitzki applies, so noetherian, artinian and finite length agree for modules.

Radical zero and DCCSemisimple. Read off the Wedderburn data

(n_{i}, D_{i})

Compute $rad R$ Use the unit criterion, or the trace form for a finite-dimensional algebra.

Test the quotientIs

R / rad R

semisimple? If not, stop: the ring is not semilocal.

Test liftingNil radical implies lifting

(21.28)

. Otherwise look for an explicit idempotent obstruction, as in the localisation example above.

Test T-nilpotenceExhibit a sequence with no vanishing product to refute it; find a uniform bound to prove nilpotence instead.

Record the sidePerfectness and the chain conditions are one-sided. Note which side you verified.

Comparison and Classification

Properties across the finiteness axis
	$rad R$ nilpotent	$rad R$ T-nilpotent	Idempotents lift	$R / rad R$ semisimple	Left-right symmetric
Semisimple	yes	yes	yes	yes	yes
Left artinian	yes	yes	yes	yes	no
Semiprimary	yes	yes	yes	yes	yes
Right perfect	no	yes	yes	yes	no
Semiperfect	no	no	yes	yes	yes
Semilocal	no	no	no	yes	yes
Local	no	no	yes	yes	yes

Properties across the finiteness axis

A “no” marks that the property is not implied by membership in the class, not that it always fails. Local rings, for instance, have nilpotent radical whenever they are artinian.

Every inclusion is strict — the witnesses
Inclusion	Witness in the larger class only	Why it fails the smaller condition
Semisimple $⊊$ left artinian	$k [x] / (x^{2})$	$rad = (x) \neq 0$
Left artinian $⊊$ semiprimary	$k \oplus V$ with $V$ a $k$ -space of infinite dimension and $V^{2} = 0$	Subspaces of $V$ give an infinite descending chain of ideals
Semiprimary $⊊$ right perfect	$k \cdot 1 + J$ , $J$ the strictly upper triangular infinite matrices over $k$ with finitely many nonzero entries $(23.22)$	$J$ is right T-nilpotent but not nilpotent — and not left T-nilpotent, so the ring is right perfect only
Right perfect $⊊$ semiperfect	$k [[x]]$	$rad = (x)$ and the sequence $x, x, x, \dots$ has no vanishing product
Semiperfect $⊊$ semilocal	$ℤ$ localised at the complement of $p ℤ \cup q ℤ$ , for distinct primes $p \neq q$	A domain has only trivial idempotents, so the idempotents of $𝔽_{p} \times 𝔽_{q}$ cannot lift
Semilocal $⊊$ all rings	$ℤ$	$rad ℤ = 0$ and $ℤ$ is not semisimple

The primeness axis and its witnesses
Inclusion	Witness	Comment
Division ring $⊊$ simple	$M_{2} (ℚ)$	Simple, but has zero divisors
Simple $⊊$ left primitive	A free algebra $k ⟨ x, y ⟩$	Left primitive by $(11.26)$ , far from simple
Left primitive $⊊$ prime	$ℤ$	Prime; primitive commutative rings are fields $(11.8)$
Prime $⊊$ semiprime	$ℤ \times ℤ$	Reduced hence semiprime; the two factor ideals multiply to zero

Relationship Map

Semilocal

R / rad R

semisimple

Semiperfect…and idempotents lift modulo the radical

Right perfect…and

rad R

right T-nilpotent

Semiprimary…and

rad R

nilpotent

Left artinian…and DCC on left ideals

Semisimple…and

rad R = 0

Local

R / rad R

a division ring; automatically semiperfect

Division ringSimpleLeft primitivePrimeSemiprime

Where the axes intersect — Conditions that force membership on both axes at once
- Left artinian and prime
  - $\Rightarrow$ simple artinian, hence $M_{n} (D)$ $(11.7)$
  - $\Rightarrow$ left and right primitive
- Left artinian and semiprime
  - $\Rightarrow$ semisimple
  - $\Rightarrow$ a finite product of $M_{n_{i}} (D_{i})$
- Local and prime
  - No collapse: $ℤ_{(p)}$ is local and prime but far from simple

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

Blocks and principal indecomposables

Finite-dimensional algebras are semiprimary, so the entire semiperfect apparatus — projective covers, principal indecomposables, the Cartan matrix — is available and is the standard toolkit for modular representation theory.

Homological algebra

Where flat means projective

Bass's characterisation identifies exactly the rings over which flatness and projectivity coincide, which controls when derived-category computations may substitute one for the other.

Commutative algebra

Semilocal rings in valuation theory

Rings with finitely many maximal ideals arise as intersections of localisations and as completions; the semiperfect obstruction here is the reason such rings do not decompose as products.

Computation

Normal forms in CAS

The semiperfect case is exactly where a computer algebra system can return a matrix presentation $M_{n} (k)$ over a local ring $k$ , which is the standard normal form used for basic algebras and quiver presentations.

Design Considerations

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

Ask for the weakest class that works. If a construction only needs projective covers of finitely generated modules, require semiperfect rather than artinian; the class is much larger and includes all local rings.
Choose the side deliberately. If your modules are right modules, right perfect is the hypothesis you want, and it is equivalent to DCC on principal left ideals. Getting this backwards produces theorems about the wrong class.
Prefer nilpotence when you can get it. Semiprimary is symmetric, easy to verify, and unlocks Hopkins–Levitzki. T-nilpotence buys generality at the cost of symmetry.
Do not over-specify. Requiring artinian when semiprimary suffices excludes natural infinite-dimensional examples such as the trivial extension $k \oplus V$ .

Failure Modes and Common Mistakes

Do not read semiprimary as a chain condition: it is a nilpotence condition, and the trivial extension $k \oplus V$ with $\dim_{k} V$ infinite is semiprimary and neither artinian nor noetherian.
Do not assume the classes on the primeness axis have anything to do with the finiteness axis without a chain condition; only $(11.7)$ links them, and it needs the DCC.
Do not conclude that a semiperfect ring decomposes as a product of local rings — that is the commutative statement $(23.11)$ , and it fails in general because the centrally primitive idempotents of the quotient need not lift centrally.

Quick Reference

Semilocal

R / rad R

semisimple

SemiperfectSemilocal

+

idempotents lift

Right perfectSemilocal

+

rad R

right T-nilpotent

SemiprimarySemilocal

+

rad R

nilpotent

Left artinianDCC on left ideals; implies semiprimary

Semisimple

rad R = 0

and left artinian

BassRight perfect

\Leftrightarrow

DCC on principal left ideals

CollapseLeft artinian: prime

=

primitive

=

simple

One witness per boundary
Boundary	Witness
Artinian, not semisimple	$k [x] / (x^{2})$
Semiprimary, not artinian	$k \oplus V$ , $\dim_{k} V$ infinite, $V^{2} = 0$
Right perfect, not semiprimary	Lam's infinite triangular ring $(23.22)$
Semiperfect, not perfect	$k [[x]]$
Semilocal, not semiperfect	$ℤ$ localised away from ${p, q}$
Semiprimitive, not semilocal	$ℤ$

Frequently Asked Questions

Why is semiperfect defined by a lifting property rather than a chain condition?

Because the lifting property is what the applications need. Semiperfect rings are exactly those over which every finitely generated module has a projective cover $(24.15)$ , and that statement is about lifting decompositions across $R \to R / rad R$ , not about chains. Chain conditions are sufficient for lifting but far from necessary — every local ring lifts trivially, since the quotient has no nontrivial idempotents.

Is a perfect ring necessarily semiperfect?

Yes. Right T-nilpotence implies the radical is nil, and idempotents lift modulo any nil ideal $(21.28)$ . Combined with the semisimplicity of the quotient, which is part of the definition of perfect, this gives semiperfect. The converse fails: $k [[x]]$ is semiperfect and not perfect on either side.

Why does Bass's theorem mix left and right?

The definition of right perfect is engineered so that right modules have projective covers. It happens that this is equivalent to a descending chain condition on principal left ideals — a genuine theorem, not a convention. The naming was chosen to match the module-theoretic conclusion rather than the chain condition, and Lam remarks explicitly that this is the right choice.

Where do local rings sit in the hierarchy?

Every local ring is semiperfect, because $R / rad R$ is a division ring and division rings have only the idempotents $0$ and $1$ , which lift trivially. Local rings can be artinian ( $k [x] / (x^{2})$ ), perfect but not artinian, or merely semiperfect ( $k [[x]]$ ). Locality is a condition on the shape of the quotient, orthogonal to the nilpotence conditions on the radical.

Does the hierarchy interact with Morita equivalence?

All six classes on the finiteness axis are Morita invariant, since each is defined by conditions on $rad R$ and $R / rad R$ that are preserved by passage to matrix rings and by categorical equivalence of module categories. Being local is not Morita invariant: $M_{2} (k)$ is semiperfect with simple radical quotient but not local.

What is the largest useful class beyond semilocal?

There is no single answer, and that is the point at which this hierarchy stops being the right tool. Beyond semilocal one changes axes entirely and works with semiprime and semiprimitive rings, using subdirect decomposition into prime or primitive factors instead of a Wedderburn-style product.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §19–§25, especially (20.1)–(20.7), (23.1)–(23.24) and (24.25).
H. Bass, “Finitistic dimension and a homological generalization of semi-primary rings”, Transactions of the American Mathematical Society 95 (1960), 466–488.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §27–§28.
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988, Chapter 2.
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964.

AI Suggested Questions

Construct a semiperfect ring that is not a direct product of local rings, and identify which central idempotent fails to lift.
Is there a ring that is right perfect and left semiperfect but not left perfect? Describe the general shape of such examples.
How does the hierarchy behave under passing to $M_{n} (R)$ , to $R [x]$ and to $R [[x]]$ ?
Which of these classes are closed under taking corner rings $e R e$ for an idempotent $e$ ?
Give the module-theoretic characterisation of each class in terms of projective covers and explain the pattern.
What replaces the semiperfect condition for rings without identity?
How does the hierarchy for group rings depend on the group and on the characteristic of the coefficient field?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Why nilpotence is replaced by T-nilpotence

Why lifting idempotents matters

06Key Results

07Worked Example

Placing five concrete rings

Verifying the semilocal but not semiperfect entry

08Process and Workflow

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

Blocks and principal indecomposables

Where flat means projective

Semilocal rings in valuation theory

Normal forms in CAS

12Design Considerations

13Failure Modes and Common Mistakes

14Quick Reference

15Frequently Asked Questions

16Related KEVOS Topics

17References

18AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Worked Example

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Design Considerations

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions