Subdirectly Irreducible Rings and the Little Ideal

Q: How big can the Birkhoff family be, and can it be trimmed?

As stated (12.3) uses one factor for every nonzero element of R, which is far more than necessary — the map factors through any subfamily of the m_a still meeting in zero. In practice one replaces it by a structured family: prime ideals for a semiprime ring (12.5), minimal primes for a reduced ring (12.7). Birkhoff's version is the fallback that needs no hypotheses.

Q: Why must ZG fail to be subdirectly irreducible for every finite group G?

For G = 1 this is the statement that Z has no smallest nonzero ideal. For G ≠ 1, QG is semisimple by Maschke's theorem and has at least two simple components, so at least two distinct projections have nonzero kernel; restricting them to ZG gives two nonzero ideals meeting in zero.

Q: Does subdirect irreducibility pass to quotients, subrings or matrix rings?

Not to subrings: Z Q and only the larger ring is irreducible. Not to quotients in general either, since Z/(p^2) → Z/(p) preserves it but R → R × R-style quotients need not. It does pass to matrix rings in the following sense: the ideals of M_n(R) are exactly M_n(A) for A R, so M_n(R) is subdirectly irreducible iff R is, with little ideal M_n(L).

Executive Summary

A subdirect decomposition is only informative when no factor hands $R$ straight back. Subdirectly irreducible rings are those for which no informative decomposition exists: every attempt returns $R$ itself in some coordinate. Lam's $(12.2)$ shows this happens precisely when the nonzero ideals of $R$ intersect in something nonzero, so the whole obstruction is carried by a single ideal — the smallest one there is.

Birkhoff's theorem $(12.3)$ then guarantees that every nonzero ring is a subdirect product of these atoms, and $(12.4)$ settles the commutative reduced case completely: there the atoms are exactly the fields. The two results together are the reason subdirect products are worth setting up at all.

$⋂ 𝔄 \neq 0$ Defining criterion

UniqueThe little ideal

1944Birkhoff's theorem

FieldsCommutative reduced case

Overview

Throughout, rings are associative with identity and ideal means two-sided ideal. A subdirect representation of $R$ is the same data as a family ${𝔄_{i}}$ of ideals with $⋂_{i} 𝔄_{i} = 0$ , the factors being the quotients $R / 𝔄_{i}$ ; the representation is trivial exactly when some $𝔄_{i} = 0$ . Whether $R$ admits a nontrivial decomposition is therefore a question purely about the lattice of ideals of $R$ , which is what makes $(12.2)$ possible.

R subdirectly irreducible iff ⋂ {𝔄 : 0 \neq 𝔄 ⊴ R} \neq 0

(12.2)

For a simple ring the family on the right has the single member $R$ , so the intersection is $R \neq 0$ and simple rings qualify.

The criterion is easy to test in practice because it only asks for a smallest nonzero ideal. Chains of ideals decreasing to zero — $ℤ \supset (p) \supset (p^{2}) \supset \dots$ , or $R \supset R x \supset R x^{2} \supset \dots$ in a skew polynomial ring — are the standard obstruction, and they are what makes most familiar rings subdirectly reducible.

The companion page Subdirect Products of Rings sets up the representations themselves; this page is about the atoms. Semiprime and Semiprimitive Rings as Subdirect Products and Reduced Rings as Subdirect Products of Domains then give the two decompositions that actually get used.

Learning Objectives

State the three equivalent conditions of $(12.2)$ and prove their equivalence.
Show the little ideal is unique, minimal, and generated by any of its nonzero elements.
Prove Birkhoff's theorem $(12.3)$ using ideals maximal with respect to avoiding an element.
Prove that a commutative reduced ring is subdirectly irreducible iff it is a field $(12.4)$ .
Classify $ℤ / (p^{n})$ , $ℤ$ , $ℤ G$ and prime rings with nonzero socle.
Explain why left primitive rings need not be subdirectly irreducible.

Definitions

Definition(12.2)Subdirectly irreducible ring

A nonzero ring $R$ is subdirectly irreducible if every representation of $R$ as a subdirect product of rings is trivial, i.e. one of the coordinate maps $R \to R_{i}$ is an isomorphism. Otherwise $R$ is subdirectly reducible. The zero ring is excluded by convention.

Little ideal $L$: A nonzero ideal contained in every nonzero ideal of $R$ . It exists iff $R$ is subdirectly irreducible, and is then unique. Called the heart in much of the ring-theoretic literature and the monolith in universal algebra.
$𝔄 ⊴ R$: $𝔄$ is a two-sided ideal of $R$ . One-sided ideals play no role in $(12.2)$ .
Trivial representation: A subdirect representation $ε : R ↪ \prod_{i} R_{i}$ with $\ker (R \to R_{i}) = 0$ for some $i$ .
Completely meet-irreducible: An ideal $𝔄$ for which the intersection of all ideals strictly containing $𝔄$ is strictly larger than $𝔄$ ; equivalently $R / 𝔄$ is subdirectly irreducible.
Reduced: $a^{2} = 0 \Rightarrow a = 0$ ; equivalently $R$ has no nonzero nilpotent elements.

The definition is a statement about two-sided ideals only, so it is left-right symmetric on its face — unlike primitivity, which is not.

Core Concepts

The little ideal and what it determines

Suppose $R$ is subdirectly irreducible with little ideal $L$ . Then $L$ is the intersection of all nonzero ideals, hence unique. It is a minimal ideal: any nonzero ideal inside $L$ contains $L$ . More sharply, if $0 \neq a \in L$ then the ideal $(a)$ generated by $a$ is nonzero and sits inside $L$ , so $L \subseteq (a) \subseteq L$ .

L = ⋂_{0 \neq 𝔄 ⊴ R} 𝔄 = (a) for every a \in L ∖ {0}

(12.2′)

The little ideal is principal as a two-sided ideal, in every possible way at once. As an $(R, R)$ -bimodule it is simple.

Direct indecomposability

If $e \in R$ is a central idempotent with $e \neq 0, 1$ , then $R ≅ R e \times R (1 - e)$ and the ideals $R e$ , $R (1 - e)$ are nonzero with zero intersection. So a subdirectly irreducible ring has no nontrivial central idempotents; in particular it is directly indecomposable. The converse fails badly — $ℤ$ is directly indecomposable and subdirectly reducible.

Which quotients are irreducible

For an ideal $𝔄 ⊴ R$ , the quotient $R / 𝔄$ is subdirectly irreducible exactly when $𝔄$ is completely meet-irreducible: the intersection of the ideals strictly containing $𝔄$ must be strictly larger than $𝔄$ . Birkhoff's proof manufactures such ideals in bulk by taking, for each nonzero $a$ , an ideal maximal among those missing $a$ .

a \neq 0

𝔪_{a}

maximal with

a \notin 𝔪_{a}

R / 𝔪_{a}

subdirectly irreducible

⋂_{a} 𝔪_{a} = 0

Key Results

Proposition(12.2)Characterisation of subdirect irreducibility

Let $R$ be a nonzero ring. The following are equivalent:

every representation of $R$ as a subdirect product of rings is trivial;
the intersection of all nonzero ideals of $R$ is nonzero;
$R$ possesses a nonzero ideal contained in every other nonzero ideal.

Proof

**(2) $\Leftrightarrow$ (3).** If the intersection $L$ of all nonzero ideals is nonzero then $L$ itself is an ideal contained in every nonzero ideal. Conversely such an ideal is contained in the intersection, which is therefore nonzero.

**(1) $\Rightarrow$ (2).** Contrapositive. Suppose the nonzero ideals ${𝔄_{i} : i \in I}$ satisfy $⋂_{i} 𝔄_{i} = 0$ . The induced map $R \to \prod_{i} R / 𝔄_{i}$ has kernel $⋂_{i} 𝔄_{i} = 0$ , so it is injective, and each coordinate map is the quotient map, hence onto. This is a subdirect representation, and no coordinate map is injective because every $𝔄_{i} \neq 0$ . So the representation is nontrivial, contradicting (1).

**(2) $\Rightarrow$ (1).** Let $ε : R ↪ \prod_{i} R_{i}$ be any subdirect representation and put $𝔄_{i} = \ker (R \to R_{i})$ . If every $𝔄_{i}$ were nonzero, then $⋂_{i} 𝔄_{i}$ would contain the nonzero intersection of all nonzero ideals, hence be nonzero — contradicting injectivity of $ε$ . So $𝔄_{i} = 0$ for some $i$ , and since $R \to R_{i}$ is onto with zero kernel it is an isomorphism.

Corollary(12.2′)The little ideal

A nonzero ring $R$ is subdirectly irreducible iff it has a little ideal $L$ , i.e. a nonzero ideal contained in every nonzero ideal. In that case $L$ is unique, equals the intersection of all nonzero ideals, is the unique minimal ideal of $R$ , and satisfies $L = (a)$ for every nonzero $a \in L$ .

Theorem(12.3)Birkhoff

Every nonzero ring $R$ can be represented as a subdirect product of subdirectly irreducible rings.

Proof

Fix $a \neq 0$ in $R$ . The set of ideals not containing $a$ is nonempty (it contains $0$ ) and is closed under unions of chains, since a union of a chain of ideals is an ideal and still misses $a$ . By Zorn's Lemma choose $𝔪_{a}$ maximal among ideals with $a \notin 𝔪_{a}$ .

Write $\bar{R} = R / 𝔪_{a}$ and $\bar{a} = a + 𝔪_{a} \neq 0$ . If $\bar{𝔅}$ is a nonzero ideal of $\bar{R}$ , its preimage $𝔅$ strictly contains $𝔪_{a}$ , so maximality forces $a \in 𝔅$ , i.e. $\bar{a} \in \bar{𝔅}$ . Hence $\bar{a}$ lies in every nonzero ideal of $\bar{R}$ , and the ideal $(\bar{a}) \neq 0$ is contained in all of them: $\bar{R}$ is subdirectly irreducible with little ideal $(\bar{a})$ .

Finally $⋂_{a \neq 0} 𝔪_{a} = 0$ , because $a \notin 𝔪_{a}$ for each $a \neq 0$ . By the ideal-theoretic criterion for subdirect representations, $R ↪ \prod_{a \neq 0} R / 𝔪_{a}$ exhibits $R$ as a subdirect product of subdirectly irreducible rings.

R ↪ \prod_{0 \neq a \in R} R / 𝔪_{a}, 𝔪_{a} maximal with respect to a \notin 𝔪_{a}

(12.3)

The canonical Birkhoff representation. It is wildly redundant — one factor per nonzero element — but requires no hypotheses at all.

Proposition(12.4)Commutative reduced case

Let $R$ be a nonzero commutative reduced ring. Then $R$ is subdirectly irreducible if and only if $R$ is a field.

Proof

( $\Leftarrow$ ) A field is simple, and a simple ring has $R$ as its unique nonzero ideal, hence is subdirectly irreducible.

( $\Rightarrow$ ) Let $L$ be the little ideal and fix $0 \neq a \in L$ . First, $R$ has no idempotent $e \neq 0, 1$ : such an $e$ would give $R ≅ R e \times R (1 - e)$ , a nontrivial subdirect representation. Since $R$ is reduced, $a^{2} \neq 0$ , so $(a^{2}) = a^{2} R$ is a nonzero ideal and therefore contains $L ∋ a$ . Write $a = a^{2} b$ with $b \in R$ .

Then $(a b)^{2} = a^{2} b^{2} = (a^{2} b) b = a b$ , so $a b$ is idempotent, and $a b \neq 0$ because $a \cdot a b = a^{2} b = a \neq 0$ . By the previous paragraph $a b = 1$ , so $a \in U (R)$ .

Now take any $c \neq 0$ . The ideal $R c$ is nonzero, so $L \subseteq R c$ and in particular $a = r c$ for some $r \in R$ . Since $a$ is a unit, $c$ is a unit. Every nonzero element of $R$ is invertible, so $R$ is a field.

PropositionEx. 12.1Transfer along the little ideal

Let $R$ be subdirectly irreducible with little ideal $L$ . Then: (a) if $R$ is semiprime, $R$ is prime; (b) if $R$ is semiprimitive, $R$ is left primitive and right primitive; (c) if $R$ is reduced, $R$ is a domain.

Proof

(a) Semiprimeness says no nonzero ideal squares to zero, so $L^{2} \neq 0$ . But $L^{2}$ is a nonzero ideal, hence $L \subseteq L^{2} \subseteq L$ and $L^{2} = L$ . Now let $𝔄, 𝔅$ be nonzero ideals: both contain $L$ , so $𝔄 𝔅 \supseteq L^{2} = L \neq 0$ . Thus $R$ is prime.

(b) $rad R = 0$ is the intersection of the left primitive ideals of $R$ . If every left primitive ideal were nonzero, that intersection would contain $L \neq 0$ . So some left primitive ideal is $0$ , i.e. $R$ is left primitive. The same argument with right primitive ideals — whose intersection is also $rad R$ , by left-right symmetry of the radical — gives right primitivity.

(c) In a reduced ring the minimal prime ideals intersect in the lower nilradical ${Nil}_{*} R = 0$ . If all of them were nonzero their intersection would contain $L$ , so some minimal prime is $0$ ; by the Lemma $(12.6)$ minimal primes of a reduced ring are completely prime, so $R ≅ R / 0$ is a domain.

Proof Techniques and Method

The reusable moves behind these proofs.

Move 1

Zorn against a single element

To split off a subdirectly irreducible quotient, pick $a \neq 0$ and take an ideal maximal with respect to not containing $a$ . Maximality forces $a$ into every larger ideal, which is exactly the little-ideal condition downstairs.

Move 2

Test everything on $L$

A claim about all nonzero ideals reduces to a claim about $L$ , since every nonzero ideal contains it. This is how semiprime upgrades to prime: one only has to know $L^{2} \neq 0$ .

Move 3

Idempotents split rings

A central idempotent $e \neq 0, 1$ produces two nonzero ideals meeting in zero. So subdirect irreducibility immediately buys you trivial central idempotents — the entry point to $(12.4)$ .

Move 4

Kernels, not maps

Every question about subdirect representations translates into a question about a family of ideals with zero intersection. Do the translation first; the ring maps rarely help.

Move 1 is Birkhoff's contribution and is not special to rings: the same argument, with congruences in place of ideals, proves that every algebra in any variety is a subdirect product of subdirectly irreducible algebras. Move 2 is the one specific to this section and reappears in the semiprime and reduced decompositions.

Worked Example

A subdirectly irreducible ring: $ℤ / (p^{n})$

The ideals of $R = ℤ / (p^{n})$ are the $(p^{j}) / (p^{n})$ for $0 \leq j \leq n$ , and they form a chain. The smallest nonzero one is $(p^{n - 1}) / (p^{n})$ , a group of order $p$ .

L (ℤ / (p^{n})) = (p^{n - 1}) / (p^{n}) ≅ ℤ / (p)

(E.1)

For $n \geq 2$ this ring is subdirectly irreducible but not simple, and it is not reduced — consistent with $(12.4)$ , which would otherwise force it to be a field.

Concretely for $p = 2$ , $n = 3$ : the ideals of $ℤ / 8$ are $0 \subset {0, 4} \subset {0, 2, 4, 6} \subset ℤ / 8$ , so $L = {0, 4}$ . Note $4^{2} = 16 = 0$ , so $L^{2} = 0$ and the ring is not semiprime — again matching Ex. 12.1(a), since $ℤ / 8$ is certainly not prime.

A subdirectly reducible ring: $ℤ C_{3}$

Let $G = C_{3} = ⟨ g ⟩$ and $R = ℤ G$ . Since $ℚ G ≅ ℚ \times ℚ (ω)$ with $ω$ a primitive cube root of unity, $R$ maps onto two rings of characteristic zero:

ε : ℤ C_{3} ↠ ℤ, g \mapsto 1; ϕ : ℤ C_{3} ↠ ℤ [ω], g \mapsto ω .

(E.2)

Both kernels are nonzero: $g - 1 \in \ker ε$ and $1 + g + g^{2} \in \ker ϕ$ . Their intersection is zero. Indeed if $a + b g + c g^{2}$ dies under both, then $a + b + c = 0$ and, using $ω^{2} = - 1 - ω$ and the fact that ${1, ω}$ is a $ℤ$ -basis of $ℤ [ω]$ ,

a + b ω + c ω^{2} = (a - c) + (b - c) ω = 0 ⟹ a = c, b = c,

(E.3)

whence $3 a = a + b + c = 0$ , so $a = b = c = 0$ . Therefore $ℤ C_{3} ↪ ℤ \times ℤ [ω]$ is a nontrivial subdirect representation and $ℤ C_{3}$ is subdirectly reducible. The same argument runs for $ℤ C_{n}$ , giving a subdirect embedding into $\prod_{d ∣ n} ℤ [ζ_{d}]$ .

A left primitive ring that is subdirectly reducible

Let $k$ be a division ring and $σ$ an endomorphism of $k$ that is not an automorphism of finite inner order — for instance $k = ℚ (t)$ with $σ (f (t)) = f (t^{2})$ , which is injective but not onto. In $R = k [x; σ]$ Lam's $(11.12)$ shows the nonzero ideals are exactly the $R x^{m}$ , $m \geq 0$ , and $(11.13)$ shows $R$ is left primitive. Since every nonzero element of $R x^{m}$ has all terms of degree $\geq m$ , $⋂_{m \geq 0} R x^{m} = 0$ : no little ideal, so $R$ is subdirectly reducible even though it is primitive.

Frameworks and Models

Subdirect irreducibility arises from a short list of causes. Recognising which one is in play usually identifies the little ideal immediately.

Subdirectly irreducible rings — smallest nonzero ideal $L$ exists
- Simple rings — $L = R$
  - division rings and fields
  - $M_{n} (D)$ for $D$ a division ring
  - the Weyl algebra $A_{1} (k)$ , $char k = 0$
- Prime rings with nonzero socle — $L = soc (R)$
  - $End (V_{D})$ for $V$ infinite-dimensional
  - any prime ring containing a minimal left ideal
- Local rings with simple socle — $L$ is the last nonzero power of the maximal ideal
  - $ℤ / (p^{n})$ and Galois rings
  - $k [t] / (p (t)^{n})$
  - commutative artinian local Gorenstein rings
- Birkhoff quotients — $R / 𝔪_{a}$ for $𝔪_{a}$ maximal missing $a$
  - little ideal generated by the image of $a$
  - produced from any ring at all

Comparison and Classification

Subdirect irreducibility across familiar rings
Ring	Subdirectly irreducible?	Little ideal	Why
Field $k$ , division ring $D$	yes	$R$ itself	simple
$M_{n} (D)$	yes	$R$ itself	simple
Semisimple ring	iff one simple component	$R$ itself	distinct components give ideals meeting in $0$
$ℤ / (p^{n})$ , $n \geq 1$	yes	$(p^{n - 1}) / (p^{n})$	ideals form a chain
$k [t] / (p (t)^{n})$ , $p$ irreducible	yes	$(p (t)^{n - 1}) / (p (t)^{n})$	ideals form a chain
$ℤ$	no	—	$⋂_{i} (p_{i}) = 0$ for distinct primes
$k [t]$ , $k$ a field	no	—	infinitely many primes
$ℤ G$ , $G$ finite	no	—	$ℚ G$ semisimple with $\geq 2$ components
Prime with $soc (R) \neq 0$	yes	$soc (R)$	every nonzero ideal contains each minimal left ideal
Non-artinian simple ring	yes	$R$ itself	simple, though $soc (R) = 0$
$k [x; σ]$ as in $(11.12)$	no	—	$⋂_{m} R x^{m} = 0$ , yet $R$ is left primitive

Which hypotheses force which conclusions for a subdirectly irreducible ring
	Prime	Left primitive	Domain	Simple
No extra hypothesis	no	no	no	no
Semiprime	yes	no	no	no
Semiprimitive	yes	yes	no	no
Reduced	yes	no	yes	no
Commutative reduced	yes	yes	yes	yes

Which hypotheses force which conclusions for a subdirectly irreducible ring

Read each row as: subdirectly irreducible plus the row hypothesis implies the marked columns. Semiprimitive implies semiprime, which is why the first column fills in.

Relationship Map

The place of these rings in the wider hierarchy is easiest to read off the implications they satisfy — and, just as usefully, the ones they do not.

SimpleSubdirectly irreducibleDirectly indecomposableNo nontrivial central idempotent

**Simple $\Rightarrow$ subdirectly irreducible**, with $L = R$ ; the converse fails for $ℤ / (p^{2})$ .
**Subdirectly irreducible $+$ semiprime $\Rightarrow$ prime**; the converse fails for $ℤ$ viewed as a prime ring.
**Left primitive $not \Rightarrow$ subdirectly irreducible**, by the skew polynomial example; and **subdirectly irreducible $not \Rightarrow$ left primitive**, since $ℤ / (p^{2})$ is neither.
**Subdirectly irreducible $+$ semiprimitive $\Rightarrow$ left and right primitive**, so within this class the two primitivities agree.

Birkhoff's theorem is the bridge from this page to the rest of the section: it says the class is large enough to build everything, while $(12.5)$ and $(12.7)$ show that for semiprime, semiprimitive and reduced rings one can do far better than the canonical Birkhoff factors.

Applications and Industry Use

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

Universal algebra

The prototype decomposition theorem

Birkhoff's argument is stated for arbitrary algebras: every algebra is a subdirect product of subdirectly irreducible ones. The ring case is the instance where congruences are ideals, and it is the model for the same statement about groups, lattices and modules.

Commutative algebra

Gorenstein local rings

A commutative artinian local ring is subdirectly irreducible exactly when its socle is one-dimensional over the residue field — the Gorenstein condition. Macaulay duality and the theory of inverse systems are built on that unique minimal ideal.

Coding theory

Codes over finite chain rings

$ℤ / (p^{n})$ and the Galois rings $GR (p^{n}, m)$ have linearly ordered ideals, hence a little ideal. Codes over them — the setting of the Kerdock and Preparata results over $ℤ / 4$ — use that unique minimal ideal as the bottom of the torsion filtration.

Module theory

Indecomposable injectives

The module-theoretic analogue is a module with essential simple socle. These are exactly the modules whose injective envelopes are indecomposable, which is how the injective spectrum of a noetherian ring is catalogued.

The honest summary: subdirect irreducibility is infrastructure. It is rarely the property one wants to prove about a ring; it is the property that certifies a decomposition has run out of road.

Standards and Notation

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

Lam's termlittle ideal, written

L

Common variantheart of the ring; also core in some sources

Universal algebramonolith — the unique minimal congruence

Ideal symbol

𝔄 ⊴ R

for two-sided ideals;

\subseteq

alone is ambiguous

Socle

soc (R)

means the left socle unless stated; in a semiprime ring the two agree

ComputationGAP TwoSidedIdeals, Magma MinimalIdeals, Sage ideal lattices for finite rings

Failure Modes and Common Mistakes

Do not assume the Birkhoff factors are canonical — the ideals $𝔪_{a}$ depend on choices, and different Zorn choices give genuinely different factorisations.
Do not expect $(12.4)$ without reduced: $ℤ / (p^{n})$ is commutative and subdirectly irreducible but is not a field for $n \geq 2$ .
Do not expect $(12.4)$ without commutative: Lam's Exercise 12.2 asks for a nonsimple subdirectly irreducible domain, and such a ring is reduced and subdirectly irreducible but not a division ring.
Do not read subdirect irreducibility as a finiteness condition. It says nothing about chain conditions, and non-artinian simple rings satisfy it.

Historical Notes and Lessons Learned

1934Macaulay's inverse systemsIn the commutative graded setting Macaulay studies artinian rings with a one-dimensional socle — the objects later recognised as the subdirectly irreducible artinian local rings.
1944Birkhoff's subdirect unionsBirkhoff introduces subdirect products for arbitrary algebras and proves that every algebra is a subdirect union of subdirectly irreducible ones. The Zorn argument used here is his.
1945McCoy on commutative ringsMcCoy studies subdirectly irreducible commutative rings systematically, obtaining structural information well beyond the reduced case treated in $(12.4)$ .
1945Jacobson's radicalThe radical-theoretic side develops in parallel; combining it with Birkhoff's decomposition is what produces $(12.5)$ and the reduction technique of $(12.8)$ .
1965Divinsky's surveyDivinsky's Rings and Radicals collects the subdirect and radical machinery in one place and continues the classification programme for subdirectly irreducible rings.

The methodological lesson is worth stating plainly. Birkhoff's theorem is cheap — it needs no hypotheses — and correspondingly weak, because the subdirectly irreducible rings it produces can be as complicated as the ring one started with. Every later result in this section trades generality for control: restrict to semiprime or reduced rings, and the factors become prime rings or domains, objects one can actually name.

Quick Reference

Definition

R \neq 0

and every subdirect representation of

R

is trivial

Working test

R

has a smallest nonzero two-sided ideal

L

Little ideal

L = ⋂ {𝔄 \neq 0} = (a)

for any

0 \neq a \in L

Birkhoff

R ↪ \prod_{a \neq 0} R / 𝔪_{a}

, all factors irreducible

Commutative reducedsubdirectly irreducible

iff

field

Semiprime case

L^{2} = L

and

R

is prime

Prime with socle

L = soc (R)

Not implied byleft primitivity, artinian-ness, or being a domain

Results on this page
Reference	Statement	Hypotheses
(12.2)	Three equivalent conditions; definition of subdirectly irreducible	$R \neq 0$
(12.2′)	Little ideal exists, is unique and minimal	$R$ subdirectly irreducible
(12.3)	Subdirect product of subdirectly irreducible rings	$R \neq 0$ ; uses Zorn's Lemma
(12.4)	Subdirectly irreducible $iff$ field	$R$ commutative, reduced, nonzero
Ex. 12.1	Semiprime $\to$ prime; semiprimitive $\to$ primitive; reduced $\to$ domain	$R$ subdirectly irreducible

Frequently Asked Questions

Why does the definition use two-sided ideals when so much of this theory is one-sided?

Because subdirect representations are built from ring homomorphisms, and the kernel of a ring homomorphism is a two-sided ideal. There is no useful one-sided version: a family of left ideals meeting in zero does not give an embedding of rings. This is also why subdirect irreducibility is automatically left-right symmetric, unlike primitivity.

Is a subdirectly irreducible ring the same as a ring with a unique minimal ideal?

Almost, and the difference is worth knowing. A little ideal is a minimal ideal contained in every nonzero ideal. A ring can have exactly one minimal ideal without that ideal being contained in all nonzero ideals — but only if some nonzero ideal contains no minimal ideal at all, which requires a descending chain with no bottom. For rings where every nonzero ideal contains a minimal one, the two notions coincide.

How big can the Birkhoff family be, and can it be trimmed?

As stated $(12.3)$ uses one factor for every nonzero element of $R$ , which is far more than necessary — the map factors through any subfamily of the $𝔪_{a}$ still meeting in zero. In practice one replaces it by a structured family: prime ideals for a semiprime ring $(12.5)$ , minimal primes for a reduced ring $(12.7)$ . Birkhoff's version is the fallback that needs no hypotheses.

Why must

ℤ G

fail to be subdirectly irreducible for every finite group

G

For $G = {1}$ this is the statement that $ℤ$ has no smallest nonzero ideal. For $G \neq {1}$ , $ℚ G$ is semisimple by Maschke's theorem and has at least two simple components, so at least two distinct projections have nonzero kernel; restricting them to $ℤ G$ gives two nonzero ideals meeting in zero.

Does subdirect irreducibility pass to quotients, subrings or matrix rings?

Not to subrings: $ℤ \subset ℚ$ and only the larger ring is irreducible. Not to quotients in general either, since $ℤ / (p^{2}) \to ℤ / (p)$ preserves it but $R \to R \times R$ -style quotients need not. It does pass to matrix rings in the following sense: the ideals of $M_{n} (R)$ are exactly $M_{n} (𝔄)$ for $𝔄 ⊴ R$ , so $M_{n} (R)$ is subdirectly irreducible iff $R$ is, with little ideal $M_{n} (L)$ .

What replaces

(12.4)

for commutative rings that are not reduced?

There is no equally clean answer, which is why Lam stops where he does. McCoy and Divinsky obtained partial structure results: a commutative subdirectly irreducible ring has a little ideal $L$ annihilated by a maximal ideal, so $L$ is a one-dimensional vector space over a field, but the ring above it can be complicated. In the artinian local case the condition is exactly Gorenstein.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991 — §12, results (12.2), (12.3) and (12.4), pp. 204–206.
G. Birkhoff, Subdirect unions in universal algebra, Bulletin of the American Mathematical Society 50 (1944).
N. H. McCoy, Subdirectly irreducible commutative rings, Duke Mathematical Journal 12 (1945).
N. Divinsky, Rings and Radicals, University of Toronto Press, 1965.
S. Burris and H. P. Sankappanavar, A Course in Universal Algebra, Springer, 1981 — Chapter II for Birkhoff's subdirect representation theorem in full generality. https://www.math.uwaterloo.ca/~snburris/htdocs/ualg.html
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988 — subdirect decompositions and radical theory.

AI Suggested Questions

Prove that $M_{n} (R)$ is subdirectly irreducible if and only if $R$ is, and identify its little ideal.
Construct a nonsimple subdirectly irreducible domain, as requested by Lam's Exercise 12.2.
Show that a commutative artinian local ring is subdirectly irreducible exactly when it is Gorenstein.
Work out the little ideal of the Galois ring $GR (4, 3)$ and describe the ideal chain above it.
Compare Birkhoff's theorem for rings with its version for groups and for lattices — what plays the role of the little ideal?
Give an example of a subdirectly irreducible ring whose little ideal is not finitely generated as a one-sided ideal.
How does the class of subdirectly irreducible rings behave under passage to $R [x]$ and to $R [[x]]$ ?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

The little ideal and what it determines

Direct indecomposability

Which quotients are irreducible

06Key Results

07Proof Techniques and Method

Zorn against a single element

Test everything on L

Idempotents split rings

Kernels, not maps

08Worked Example

A subdirectly irreducible ring: ℤ/(pn)

A subdirectly reducible ring: ℤC3

A left primitive ring that is subdirectly reducible

09Frameworks and Models

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

The prototype decomposition theorem

Gorenstein local rings

Codes over finite chain rings

Indecomposable injectives

13Standards and Notation

14Failure Modes and Common Mistakes

15Historical Notes and Lessons Learned

16Quick Reference

17Frequently Asked Questions

18Related KEVOS Topics

19References

20AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Test everything on $L$

Worked Example

A subdirectly irreducible ring: $ℤ / (p^{n})$

A subdirectly reducible ring: $ℤ C_{3}$

Frameworks and Models

Comparison and Classification

Relationship Map

Applications and Industry Use

Standards and Notation

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions