← LibraryExamples of Noncommutative Local RingsEngineering · Engineering MathematicsLesson 396/812← PrevNext →
ArticlePublished 8 Aug 2026Updated 9 Aug 202621 min readBy KEVOS®
Skip to content

Engineering Mathematics Core Local rings

Examples of Local Rings

Division rings, twisted power series, constant-diagonal triangular matrices, exterior algebras and group algebras of finite p-groups — the standing supply of local rings that no localisation process produces.

Page ID
KEVOS-ENG-MATH-NCR-0140
Taxonomy
ENG / ENG-MATH
Collection
noncommutative-rings-core
Source
(19.4)–(19.11), §19 (pp. 296–300)
Reviewed
2026-08-08
Version
1.0.0

Executive Summary

Commutative algebra manufactures local rings by localising at a prime. Noncommutative algebra cannot, so its local rings have to be recognised where they occur. Five families cover almost every example that arises in practice: division rings; power series and twisted power series over a local ring; triangular matrices with constant diagonal; exterior algebras; and group algebras of finite p-groups in characteristic p.

Each is verified the same way — exhibit a two-sided ideal 𝔪 with R/𝔪 a division ring and 𝔪 nil or nilpotent, then invoke (19.1)(3). The differences lie in whether 𝔪 is nilpotent (completely primary) or merely maximal, and that difference is what separates the artinian examples from the rest.

5Standard families
𝔪n+1=0Exterior algebra of dimV=n
|G|Nilpotency bound for kG
kResidue ring of kG, G a p-group

Overview

Local Rings: Definition and Equivalent Characterisations reduces locality to a single test: R0 is local exactly when R/radR is a division ring. This page applies that test repeatedly.

It is worth being explicit about what is not available. Localising a noncommutative ring at a prime ideal requires an Ore condition that generically fails, so there is no functorial machine turning an arbitrary ring into a local one. Every example below is therefore a construction that happens to produce a division-ring quotient, not the output of a localisation.

Find a candidate ideal 𝔪Check R/𝔪 is a division ringCheck 𝔪radR𝔪=radR, so R is local

The third step is usually the only work, and it is almost always done by showing 𝔪 is nil or nilpotent and quoting (4.11): every nil one-sided ideal lies in the radical.

Learning Objectives

  • Recall the commutative examples — localisations, local rings of varieties, germ rings, valuation rings — and why they do not transfer.
  • Prove that R[[x]] and R[[x;σ]] are local whenever R is.
  • Compute radA for the constant-diagonal triangular matrix ring of (19.8).
  • Prove that the exterior algebra of a finite-dimensional vector space is local with nilpotent radical.
  • State and prove (19.11) for group algebras RG with G a finite p-group.
  • Separate the completely primary examples from the merely local ones.

Core Concepts

The commutative stock, for orientation

(19.4) Localisation
For a commutative ring R and a prime 𝔭, the localisation R𝔭 is local with maximal ideal 𝔭R𝔭.
(19.5) Local ring of a variety
At a point x of an algebraic variety X, the rational functions regular at x form a local ring 𝒪x; the maximal ideal is the functions vanishing at x.
(19.6) Germs of holomorphic functions
At a point x of a Riemann surface, germs of functions holomorphic at x form a local ring, with maximal ideal the germs vanishing at x.
(19.7) Valuation rings
Every valuation ring of a field is local. The p-adic integers p are the rank-one discrete case inside p, with rad(p)=pp.

These four are the intuition. None of them survives verbatim: (19.4) needs Ore localisation, and (19.5)(19.6) are geometric. What does survive is (19.7), because the valuation criterion (19.3)(b) makes no use of commutativity.

Power series: the first noncommutative supply

Let R be any ring and A=R[[x]]. A power series is a unit in A exactly when its constant term is a unit in R: if f=a0g with a0U(R) and gxR[[x]], then f=a0(1a01g) and 1a01g is inverted by the geometric series i0(a01g)i, which converges coefficientwise because a01gxR[[x]].

radR[[x]]={a0+a1x+a2x2+:a0radR}
(19.7a)

The variable x always lies in the radical; the constant term is the only obstruction.

Nothing in that argument uses commutativity of x with the coefficients. If σAut(R) and A=R[[x;σ]] is the twisted power series ring with xr=σ(r)x, the same computation applies verbatim, because xR[[x;σ]]=R[[x;σ]]x is still a two-sided ideal whose powers shrink.

Graded algebras with one-dimensional degree zero

If A=A0A1An is a graded algebra with A0 a division ring and Ai=0 for i>n, then 𝔪=A1An is a two-sided ideal with 𝔪n+1=0 and A/𝔪A0. So A is local and completely primary. Exterior algebras (19.9) are the standard instance; so are truncated tensor algebras.

Key Results

Proposition(19.7b)Power series over a local ring

Let R be a ring and σ an automorphism of R. Then radR[[x;σ]] consists of the series whose constant term lies in radR, and R[[x;σ]]/radR[[x;σ]]R/radR. In particular, if R is local then so is R[[x;σ]], with the same residue division ring.

Proof

Write A=R[[x;σ]] and let 𝔘={fA:f(0)radR}, where f(0) denotes the constant term. Since σ(radR)=radR (an automorphism preserves the radical) and ff(0) is a ring homomorphism AR, the set 𝔘 is the preimage of radR and hence a two-sided ideal, with A/𝔘R/radR.

For f𝔘, the series 1f has constant term 1f(0)1+radRU(R), so 1fU(A) by the unit criterion for power series. The same applies to 1gfh for any g,hA, since 𝔘 is an ideal. By the maximality property of the radical, 𝔘radA.

Conversely, because 𝔘radA, the quotient rule (4.6) gives rad(A/𝔘)=(radA)/𝔘. But A/𝔘R/radR has zero radical, so radA=𝔘. If R is local, A/radAR/radR is a division ring and (19.1)(3) makes A local.

Example(19.8)Constant-diagonal triangular matrices

Let k be a division ring, T=Tn(k) the ring of upper triangular n×n matrices over k, and JT the strictly upper triangular matrices. It is standard that J=radT and Jn=0. Let

A={a1+N:ak,NJ}=k1JT.
(19.8)

Then A is a subring of T, J is a two-sided ideal of A with Jn=0, and A/Jk. Hence A is local with radA=J, and completely primary. For n3 it is noncommutative even when k is a field.

Proposition(19.9)Exterior algebras are local

Let k be a field and V a k-vector space with dimkV=n<. The exterior algebra

R=Λ(V)=kΛ1(V)Λn(V)
(19.9)

is a local ring with radR=𝔪=Λ1(V)Λn(V), and 𝔪n+1=0, R/𝔪k. It is noncommutative precisely when n2 and chark2.

Proof

𝔪 is the ideal of elements with zero component in degree 0, so it is two-sided and R/𝔪k is a field. A product of n+1 elements of 𝔪 lands in degrees n+1, all of which vanish, so 𝔪n+1=0. A nilpotent ideal is nil, hence 𝔪radR by (4.11); and radR𝔪 because 𝔪 is a maximal ideal with R/𝔪 a field. So radR=𝔪 and (19.1)(3) applies.

Example(19.10)Group algebras of finite p-groups

Let k be a field of characteristic p>0 and G a finite p-group. Then rad(kG) is the augmentation ideal 𝔞=ker(kGk), with 𝔞|G|=0, and kG/𝔞k. So kG is an artinian local ring, indeed completely primary.

Proposition(19.11)Group algebras over a local coefficient ring

Let (R,𝔪) be a commutative local ring whose residue field k=R/𝔪 has characteristic p>0, and let G be a **finite p-group**. Then A=RG is a local ring with A/radAk, and radA is the ideal I generated by 𝔪 together with all g1 for gG — the kernel of the reduced augmentation map RGRk.

Proof

Let V be any simple left A-module. Being simple, V is cyclic over A; since A is generated as an R-module by the finitely many elements of G, V is a finitely generated R-module. Nakayama's Lemma (4.22) then gives 𝔪VV.

Because R is commutative and central in RG, the subset 𝔪V is an A-submodule of V. Simplicity and 𝔪VV force 𝔪V=0, so V is a module over A/𝔪AkG, and it is simple as such.

Now (8.8) applies: for a finite p-group G over a field of characteristic p, the only simple kG-module is the trivial one, so G acts trivially on V, i.e. (g1)V=0 for all gG.

Thus I annihilates every simple left A-module, giving IradA. On the other hand A/Ik: killing all g1 collapses RG onto R, and killing 𝔪 then gives k. Hence I is a maximal ideal, and since radA is proper we get radA=I and A/radAk, a field. By (19.1)(3), A is local.

Remark(19.11′)Dropping commutativity of the coefficients

The proposition remains true for (R,𝔪) a noncommutative local ring with residue division ring of characteristic p. The reduction is the same: Nakayama kills 𝔪V, leaving a simple module over kG for k a division ring of characteristic p, and the argument of (8.8) still shows that a finite p-group acts trivially. This form of the result is what is used in the integral representation theory of finite groups, where R is typically a complete discrete valuation ring.

RemarkThe converse for group algebras

For a field k and a nontrivial finite group G, kG is local only in the situation of (19.10): kG local is equivalent to chark=p>0 and G a p-group, and equally to kG being indecomposable as a module over itself. That converse is developed in When Is a Group Algebra a Local Ring?

Proof Techniques and Method

How these verifications work, and which move to reuse.

Move 1

Guess the ideal from the grading

If the ring is graded or filtered with degree-zero part a division ring, the positive part is the only candidate for the radical — and its nilpotence is immediate from the grading.

Move 2

Nil goes down, maximal comes back

(4.11) gives 𝔪radR when 𝔪 is nil; maximality of 𝔪 gives the reverse. Two one-line inclusions replace any direct computation of the radical.

Move 3

Compute units by a convergent series

In power series and other complete settings, 1g is inverted by gi whenever the powers of g shrink. This is what makes non-nilpotent radicals manageable.

Move 4

Nakayama to strip the coefficients

Over RG with R local, a simple module is finitely generated over R, so Nakayama forces 𝔪V=0 and reduces the question to the residue field. This is the whole content of (19.11).

Move 5

p-groups act trivially

In characteristic p a finite p-group has only the trivial simple module. The reason is that g1 is nilpotent for g of p-power order: (g1)pe=gpe1=0.

Move 6

Transport along a quotient

If 𝔘radA then rad(A/𝔘)=(radA)/𝔘 by (4.6). Identify the quotient as something with zero radical and the radical is pinned down exactly.

Move 5 is worth spelling out because it is where characteristic p enters. Over , g1 for g of order p is a semisimple element with p distinct eigenvalues, and G is semisimple by Maschke's theorem — the opposite extreme. Locality of kG is a purely modular phenomenon.

Worked Example

A: the exterior algebra on two generators

Take k=, V=ke1ke2, and R=Λ(V), a 4-dimensional -algebra with basis 1,e1,e2,e1e2 and relations e12=e22=0, e2e1=e1e2.

𝔪=e1e2e1e2,𝔪2=e1e2,𝔪3=0.
(E.1)

Dimensions 3,1,0 — the Loewy series of a completely primary algebra of Loewy length 3.

Check 𝔪2 by hand: the products of degree-one basis elements are e1e1=0, e2e2=0, e1e2, and e2e1=e1e2. So 𝔪2 is spanned by e1e2 alone, of dimension 1. Multiplying once more lands in degree 3>dimV, which is zero.

Units: an element u=a+w with a× and w𝔪 is invertible, with u1=a1a1wa1+a1wa1wa1, the series terminating because 𝔪3=0. Concretely (1+e1)1=1e1, since e12=0; and (1+e1+e2)1=1e1e2, since (e1+e2)2=e1e2+e2e1=0.

So RU(R)=𝔪, an ideal, and R is local by (19.1)(4) with residue field . It is noncommutative because e1e2=e2e1e2e1.

B: a group algebra with non-nilpotent radical

Take R=(2)={a/b:b odd}, a commutative local ring with 𝔪=2(2) and residue field k=𝔽2 of characteristic 2. Let G={1,g} be cyclic of order 2 — a 2-group. By (19.11), A=RG is local.

Make it explicit. Put t=g1. Then t2=g22g+1=22g=2t, so

A(2)[t]/(t2+2t),radA=(2,t),A/radA𝔽2.
(E.2)

The two halves of the example bracket the phenomenon: (19.9) and (19.10) produce completely primary artinian rings, while (19.7) and (19.11) over an infinite coefficient ring produce local rings with no chain condition at all.

Frameworks and Models

The examples organise by how the division-ring quotient is produced.

  • Sources of local rings
    • Degenerate — the radical is zero
      • division rings, including fields
      • the real quaternions
      • cyclic algebras that happen to be division algebras
    • Complete / filtered — the radical is not nil but powers shrink
      • R[[x]] over a local R
      • twisted power series R[[x;σ]]
      • p and other discrete valuation rings
      • valuation rings of a division ring (19.3)(b)
    • Graded / nilpotent — the radical is nilpotent — completely primary
      • exterior algebras Λ(V)
      • constant-diagonal triangular matrices (19.8)
      • kG for G a finite p-group, chark=p
      • truncated polynomial and tensor algebras
    • Endomorphism rings — locality is a theorem, not a construction
      • End(MR) for M indecomposable of finite length
      • End(MR) for M indecomposable injective

The fourth branch is the one with no commutative shadow, and it is the reason local rings matter here at all: it supplies the hypothesis of the Krull–Schmidt–Azumaya theorem. See Strongly Indecomposable Modules and Local Endomorphism Rings.

Comparison and Classification

The standard examples, side by side
RingradResidue objectrad nilpotent?Commutative?
Division ring D0Dyes (trivially)only if D is a field
ppp𝔽pno, not even nilyes
k[[x]](x)kno, not even nilyes
k[[x;σ]], σid(x)kno, not even nilno
A=k1JTn(k)Jkyes, index nno for n3
Λ(V), dimV=nΛ1(V)kyes, index n+1no for n2, chark2
kG, G a p-group, chark=paugmentation idealkyes, index |G|only if G is abelian
(p)G, G a p-group(p,g1)𝔽pnoonly if G is abelian
Which additional properties each family enjoys
ArtinianNoetherianDomainCompletely primary
Division ringyesyesyesyes
k[[x;σ]]noyesyesno
pnoyesyesno
Λ(V), dimV1yesyesnoyes
kG, G a nontrivial p-groupyesyesnoyes
Valuation ring of infinite ranknonoyesno

Which additional properties each family enjoys

Relationship Map

The constructions interlock: each preserves or creates locality in a predictable way.

k a fieldk[[x;σ]] local, noncommutativek[[x;σ]][[y;τ]] localiterate
  • Quotients. R local and IR an ideal R/I local. This is why 𝔽2G inherits locality from (2)G.
  • Power series. R local R[[x;σ]] local, with the same residue division ring. Polynomial rings do not behave this way: k[x] has infinitely many maximal ideals.
  • Matrices. R local Mn(R) is not local for n2; instead it is semilocal, with semisimple quotient Mn(R/radR).
  • Group algebras. R local with residue characteristic p and G a finite p-group RG local; for any other finite G the conclusion fails.
  • Centres. R local Z(R) is a commutative local ring, so every noncommutative example above has a commutative local ring hiding inside it.

Applications and Industry Use

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

Modular representation theory

Blocks and vertices

kG for G a p-group being local is the base case of block theory: every block of kH for general H is analysed by restriction to p-subgroups, where the group algebra is completely primary.

Integral representations

Coefficient rings

(19.11) over a complete discrete valuation ring R with residue characteristic p underlies the reduction-mod-p machinery relating characteristic-zero and modular representations.

Coding theory

Galois and chain rings

Codes over /pn and Galois rings live over finite chain rings, which are exactly the finite local rings with principal maximal ideal; the residue field carries the associated code.

Physics and geometry

Exterior algebras

Λ(V) is the algebra of differential forms at a point and the fermionic side of a supersymmetric algebra. Its locality is the statement that a form is invertible iff its scalar part is nonzero.

Control and signals

Skew series

Twisted power series k[[x;σ]] model linear time-varying systems where the shift operator does not commute with the coefficients; invertibility of a transfer element reduces to its constant term.

Number theory

Completions

p and, more generally, valuation rings inside division algebras over local fields are the arithmetic prototypes; the noncommutative case classifies division algebras over p-adic fields.

Computational Notes

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

  • For a finite-dimensional algebra A over a field given by structure constants, testing locality is one radical computation plus a check that dimkA/radA equals the dimension of a single division algebra factor. In characteristic 0 the radical is the kernel of the trace form, an O(n3) nullspace computation for n=dimkA.
  • In characteristic p the trace form is inadequate — the Friedl–Rónyai algorithm handles it in polynomial time. This matters exactly for the group algebras of (19.10), where the trace form is identically degenerate.
  • For kG with G a p-group in characteristic p, no computation is needed: locality is automatic and the radical is the augmentation ideal, of dimension |G|1, computable as the kernel of a single linear map.
  • Power series rings are not finitely presented; software handles them by truncation. Deciding whether a truncated series is a unit is one test on the constant term, and the inverse to precision N costs O(N2) coefficient multiplications by the geometric-series recursion, or O(NlogN) by Newton iteration in the untwisted commutative case.
  • Exterior algebras are supported natively in GAP, Macaulay2 and Singular; their radical is the ideal of positive-degree elements and needs no algorithm at all.

Failure Modes and Common Mistakes

  • Do not take the whole upper triangular ring Tn(k) for n2 to be local: its diagonal idempotents forbid it. Only the constant-diagonal subring A of (19.8) is local.
  • Do not assume Λ(V) is noncommutative: in characteristic 2, or when dimV1, it is commutative. Locality holds regardless.
  • Do not extend (19.11) to a coefficient ring of residue characteristic p: with |G| invertible in R, Maschke's argument applies and RG acquires idempotents.
  • Do not confuse the augmentation ideal with the radical in general. They coincide for p-groups in characteristic p; for S3 the augmentation ideal is a direct summand and the radical is zero.

Quick Reference

R[[x;σ]]rad={f:f(0)radR}; local iff R is
Tn(k) constant diagonalrad= strictly upper triangular, Jn=0
Λ(V), dimV=nrad=Λ1(V), 𝔪n+1=0
kG, G a p-grouprad= augmentation ideal, 𝔞|G|=0
RG, (R,𝔪) locallocal if char(R/𝔪)=p and G is a p-group
Unit test in R[[x]]fU iff f(0)U(R)
Valuation criterionRD with d or d1 in R R local
Never localMn(R) for n2; Tn(k) for n2; G for G1
Where each example is proved
ExampleReferenceRadical nilpotent?
Localisation R𝔭(19.4)no in general
Local ring of a variety(19.5)no
Germs on a Riemann surface(19.6)no
Valuation rings, p(19.7)no
Constant-diagonal triangular(19.8)yes
Exterior algebra(19.9)yes
kG, G a p-group(19.10)yes
RG over a local ring(19.11)only if radR is

Frequently Asked Questions

Why is there no noncommutative localisation producing local rings on demand?

Forming a ring of fractions with respect to a multiplicative set S requires the Ore condition — every pair (r,s)R×S must satisfy rs=sr for some rR, sS. For a general ring, and in particular for the complement of a prime ideal, that condition fails, and the universal construction that ignores it can collapse. So the localisation route to local rings simply is not available in general, and the noncommutative examples come from completions, gradings and endomorphism rings instead.

Is a twisted power series ring k[[x;σ]] ever commutative?

Exactly when σ is the identity. The relation xa=σ(a)x shows that x commutes with a iff σ(a)=a, so nontrivial σ gives a genuinely noncommutative ring. It is always a local domain with maximal ideal (x) and residue division ring k, so this is the cheapest way to produce a noncommutative local ring.

Why does the exterior algebra fail to be a domain even though it is local?

Because ee=0 for every eV, so every degree-one element is a nonzero square-zero element. Locality says nothing about zero divisors: the two conditions are independent. k[[x]] is a local domain, Λ(V) is a local ring with plenty of zero divisors, and is a domain that is not local.

How large can the nilpotency index of rad(kG) be for a p-group?

Lam's bound is 𝔞|G|=0, which is generally not sharp. For G cyclic of order pn the augmentation ideal is generated by t=g1 with tpn=0 and tpn10, so the index is exactly |G| there. For non-cyclic p-groups the index is strictly smaller: for the Klein four group in characteristic 2 the augmentation ideal has cube zero, well below |G|=4.

Does (19.11) need the group to be finite?

Yes, in two places. Finiteness makes A=RG a finitely generated R-module, which is what lets Nakayama's Lemma apply to a cyclic A-module; and the triviality of the G-action on simple modules in (8.8) is proved for finite p-groups. For infinite locally finite p-groups the augmentation ideal is nil but not nilpotent, and the conclusion has to be argued differently.

Which of these examples are artinian?

Exactly the completely primary ones with finite dimension: the constant-diagonal triangular ring, the exterior algebra, and kG for G a finite p-group over a field. Division rings are trivially artinian. The power series rings, p, valuation rings and (p)G are not — their radicals are not nil.

References

  1. T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §19 (pp. 293–310), especially (19.4)–(19.11).
  2. D. S. Passman, The Algebraic Structure of Group Rings, Wiley-Interscience, 1977, Chapters 1 and 3.
  3. C. W. Curtis and I. Reiner, Methods of Representation Theory, Volume I, Wiley-Interscience, 1981, §6.
  4. N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964.
  5. O. F. G. Schilling, The Theory of Valuations, Mathematical Surveys 4, American Mathematical Society, 1950.
  6. J.-P. Serre, Local Fields, Graduate Texts in Mathematics 67, Springer-Verlag, 1979, Chapters I–II.

AI Suggested Questions

  • Compute the Loewy length of 𝔽2Q8 and compare it with the bound of order |G|.
  • Give an example of a local ring whose maximal ideal is nil but not nilpotent, and explain where it sits among these families.
  • Which finite local rings have principal maximal ideal, and why are these exactly the finite chain rings used in coding theory?
  • Work out the units and radical of R[[x;σ,δ]], the skew power series ring with a derivation as well as an automorphism.
  • Classify the local rings of dimension at most 4 over an algebraically closed field.
  • For which finite groups G and fields k is kG semilocal but not local, and what is the semisimple quotient?
  • Explain why the invariance condition on a valuation ring of a division ring is needed to recover a valuation, and give a non-invariant example.
Page
KEVOS-ENG-MATH-NCR-0140
Path
Engineering / Mathematics
Template
kevos-knowledge-article-v2
KEVOS® Knowledge Library — reviewed 2026-08-08

Continue learning

Local Rings: Definition and Equivalent CharacterisationsArticle · Engineering MathematicsNEXT LESSON →Strongly Indecomposable Modules and Local Endomorphism RingsArticle · Engineering MathematicsPolynomial Equations over Formally Real Division RingsArticle · Engineering MathematicsFitting’s Lemma and Endomorphism Rings of Modules of Finite LengthArticle · Engineering Mathematics