The Socle of a Module and of a Ring

Q: Can the socle of a nonzero module be zero?

Yes. soc(Z) = 0 because no subgroup nZ is minimal, and soc(Q) = 0 for the same reason. It cannot happen over a left artinian ring: there every nonzero module has a simple submodule, since every cyclic submodule is artinian.

Q: Is the left socle of a ring equal to its right socle?

Not in general. For R = T_2(k) the left socle is the top row and the right socle is the second column, two different two-sided ideals of the same dimension. They agree when R is commutative and when R is semisimple, since then both are R. This is a real difference from the Jacobson radical, which is always side-neutral.

Q: Is the socle preserved by ring homomorphisms or base change?

Not in general. Restricting scalars along S → R can change which submodules are simple, and extending the base field can split a simple module into several. The socle is functorial in the module argument only, for a fixed ring.

Executive Summary

Most modules are not semisimple. The socle is the canonical device for extracting the part that is: $soc (M)$ is the sum of all simple submodules of $M$ , it is semisimple by $(2.4)$ , and it contains every semisimple submodule. So semisimplicity of $M$ is the single equation $soc (M) = M$ .

The socle appears implicitly in Lam's $(2.3)$ and in the proof of $(2.4)$ , where the sum of all simple submodules is the object that has to be shown to exhaust $M$ . Two further facts make it a working tool: it is the intersection of all essential submodules, and for a ring the left socle is a two-sided ideal — although it need not equal the right socle.

$\sum S$ Definition

$⋂ E$ Essential submodules

Two-sidedSocle of a ring

$\neq$ Left versus right

Overview

Every module has a largest semisimple submodule. It is obtained without any choice: take all the simple submodules there are and add them up.

soc (M) = \sum {S \subseteq M : S is a simple submodule of M}

(soc)

The empty sum convention applies: if $M$ has no simple submodule then $soc (M) = 0$ , as happens for $M = ℤ$ over $ℤ$ .

The socle is one of a dual pair. The radical $rad (M)$ — the intersection of the maximal submodules — is the smallest submodule with semisimple quotient in good cases; the socle is the largest submodule that is itself semisimple. Over a semilocal ring the two are tied together by $soc (M) = {ann}_{M} (rad R)$ , which turns the socle into a kernel and makes it computable.

For a ring one may take the socle of $_{R} R$ or of $R_{R}$ . Both are two-sided ideals, but they can be different ideals — a genuine asymmetry, in contrast with the Jacobson radical.

Learning Objectives

Define $soc (M)$ and prove it is the largest semisimple submodule.
Prove that $soc (M)$ is the intersection of the essential submodules of $M$ .
Prove that the left socle of a ring is a two-sided ideal.
Prove that over a left artinian ring the socle of any module is essential in it.
Compute $soc$ for $ℤ / 12$ , $ℤ (p^{\infty})$ , $\prod_{p} ℤ / p$ and for $T_{2} (k)$ on both sides.
Build the socle series of $ℤ / p^{n}$ and identify its Loewy length.

Definitions

Definition(soc)Socle of a module and of a ring

For a left $R$ -module $M$ , the socle $soc (M)$ is the sum of all simple submodules of $M$ , with $soc (M) = 0$ when there are none. The left socle of a ring $R$ is $soc (_{R} R)$ , the sum of the minimal left ideals; the right socle is $soc (R_{R})$ .

Essential submodule: $N \subseteq M$ with $N \cap K \neq 0$ for every nonzero submodule $K \subseteq M$ . Written $N \subseteq_{e} M$ .
Complement submodule: A submodule $C$ maximal with respect to $C \cap N = 0$ for a fixed $N$ ; it exists by Zorn's Lemma and makes $N \oplus C$ essential.
Isotypic component: ${soc}_{S} (M)$ , the sum of the submodules isomorphic to a fixed simple $S$ ; the socle is the direct sum of these.
$rad (M)$: The intersection of the maximal submodules of $M$ ; the dual of the socle, and equal to $M$ when $M$ has no maximal submodule.
Socle series: $0 = {soc}_{0} (M) \subseteq {soc}_{1} (M) \subseteq \dots$ with ${soc}_{n + 1} (M) / {soc}_{n} (M) = soc (M / {soc}_{n} (M))$ .

Some authors write Soc with a capital S. The lower-case operator is used throughout this collection, and soc R without a side always means the left socle here.

Core Concepts

The socle is a functor

If $f : M \to N$ is an $R$ -homomorphism and $S \subseteq M$ is simple, then $f (S)$ is a homomorphic image of a simple module, hence zero or simple. Summing, $f (soc (M)) \subseteq soc (N)$ . So $soc$ is an additive functor, and for a submodule $N \subseteq M$ it satisfies $soc (N) = N \cap soc (M)$ .

Simple submodules of

M

soc (M)

Largest semisimple submodule

⋂

essential submodules

Behaviour under sums and products

The socle commutes with direct sums, $soc (⨁_{i} M_{i}) = ⨁_{i} soc (M_{i})$ , because a simple submodule of a direct sum lies in a finite subsum. It does not commute with infinite direct products: only the inclusion $soc (\prod_{i} M_{i}) \subseteq \prod_{i} soc (M_{i})$ holds in general, and the inclusion is strict for $\prod_{p} ℤ / p$ over $ℤ$ .

Socle and radical as a dual pair

The radical of a module is the intersection of its maximal submodules; the socle is the sum of its minimal ones. The duality is genuine but imperfect: $M / rad (M)$ is the largest semisimple quotient when it exists, while $soc (M)$ is always the largest semisimple submodule. Over a semilocal ring the two are linked by the annihilator formula $soc (M) = {ann}_{M} (rad R)$ .

The socle series

Iterating the socle produces the ascending Loewy filtration: ${soc}_{1} (M) = soc (M)$ and ${soc}_{n + 1} (M)$ is the preimage in $M$ of $soc (M / {soc}_{n} (M))$ . The layers are semisimple by construction, so a module with a finite socle series is built from finitely many semisimple slabs. The number of slabs is the Loewy length, and it is finite whenever $M$ has finite length.

Key Results

Proposition—The socle is the largest semisimple submodule

Let $M$ be a left $R$ -module. Then $soc (M)$ is a semisimple submodule of $M$ , and every semisimple submodule of $M$ is contained in it. Consequently $M$ is semisimple if and only if $soc (M) = M$ .

Proof

By construction $soc (M)$ is a sum of simple submodules, hence semisimple by $(2.4) (3) \Rightarrow (1)$ . If $N \subseteq M$ is semisimple then $N$ is a sum of simple submodules of $N$ by $(2.4)$ , and each of these is a simple submodule of $M$ , so $N \subseteq soc (M)$ . The final claim is immediate: $M$ semisimple gives $M \subseteq soc (M)$ , and conversely $soc (M) = M$ makes $M$ semisimple.

Theorem—Socle as an intersection of essential submodules

For every left $R$ -module $M$ , $soc (M)$ equals the intersection of all essential submodules of $M$ .

Proof

Containment in every essential submodule. Let $S \subseteq M$ be simple and $E \subseteq M$ essential. Then $S \cap E \neq 0$ , and $S \cap E$ is a nonzero submodule of the simple module $S$ , so $S \cap E = S$ and $S \subseteq E$ . Summing over all simple $S$ gives $soc (M) \subseteq E$ .

The reverse inclusion. Fix $x \notin soc (M)$ ; we produce an essential submodule omitting $x$ . By Zorn's Lemma choose $C \subseteq M$ maximal with respect to $C \cap soc (M) = 0$ . Then $soc (M) \oplus C$ is essential in $M$ : if $K \neq 0$ met it trivially, then $(C \oplus K) \cap soc (M) = 0$ — for $c + κ = s$ forces $κ = s - c \in K \cap (soc (M) \oplus C) = 0$ and then $c = s \in C \cap soc (M) = 0$ — contradicting maximality of $C$ .

Note also $soc (C) = C \cap soc (M) = 0$ , so $C$ contains no simple submodule. Write $x = s + c$ with $s \in soc (M)$ and $c \in C$ if $x$ lies in $soc (M) \oplus C$ ; if it does not, that submodule is itself an essential submodule omitting $x$ and we are done. So assume $x = s + c$ , and $c \neq 0$ because $x \notin soc (M)$ .

By Zorn's Lemma choose $E' \subseteq C$ maximal with respect to $c \notin E'$ . Then $E'$ is essential in $C$ . Indeed, if $K \subseteq C$ is nonzero with $K \cap E' = 0$ , then $E' ⊊ E' \oplus K$ , so $c \in E' \oplus K$ by maximality, say $c = e + κ$ with $κ \neq 0$ (else $c \in E'$ ). For any nonzero $K_{0} \subseteq K$ the same argument gives $c = e_{0} + κ_{0}$ with $κ_{0} \in K_{0}$ , and subtracting, $κ - κ_{0} = e_{0} - e \in K \cap E' = 0$ , so $κ = κ_{0} \in K_{0}$ . Hence $R κ \subseteq K_{0}$ for every nonzero $K_{0} \subseteq R κ$ , so $R κ$ is simple — impossible, since $soc (C) = 0$ .

Now $E'$ essential in $C$ gives $soc (M) \oplus E'$ essential in $soc (M) \oplus C$ , which is essential in $M$ ; essentiality is transitive, so $soc (M) \oplus E'$ is essential in $M$ . Finally $x \notin soc (M) \oplus E'$ : an equality $s + c = s' + e'$ would give $c - e' = s' - s \in C \cap soc (M) = 0$ , hence $c = e' \in E'$ , contradicting the choice of $E'$ .

Proposition—The left socle is a two-sided ideal

For any ring $R$ , $soc (_{R} R)$ is a two-sided ideal of $R$ ; likewise $soc (R_{R})$ is a two-sided ideal.

Proof

The left socle is by definition a left ideal. Let $𝔞$ be a minimal left ideal and $r \in R$ . Right multiplication $ρ_{r} :_{R} R \to_{R} R$ , $x \mapsto x r$ , is a homomorphism of left modules, so $𝔞 r = ρ_{r} (𝔞)$ is a homomorphic image of the simple module $𝔞$ and is therefore either $0$ or again a minimal left ideal. In both cases $𝔞 r \subseteq soc (_{R} R)$ . Summing over all minimal left ideals gives $soc (_{R} R) r \subseteq soc (_{R} R)$ for every $r$ , which is the missing right-ideal condition.

Proposition—Essentiality over a left artinian ring

Let $R$ be left artinian. Then every nonzero left $R$ -module contains a simple submodule, and $soc (M)$ is an essential submodule of $M$ for every left $R$ -module $M$ .

Proof

Let $0 \neq K \subseteq M$ and pick $0 \neq m \in K$ . The cyclic module $R m$ is a quotient of $_{R} R$ , which is an artinian module by hypothesis, so $R m$ is artinian. A nonzero artinian module has a minimal nonzero submodule — take a minimal element of the nonempty set of nonzero submodules, available by the DCC — and such a submodule is simple. Hence $K$ contains a simple submodule, so $K \cap soc (M) \neq 0$ . As $K$ was arbitrary, $soc (M)$ is essential.

Proposition—Socle as an annihilator

Suppose $R / rad R$ is semisimple — for instance $R$ left artinian. Then for every left $R$ -module $M$ ,

soc (M) = {ann}_{M} (rad R) = {m \in M : (rad R) m = 0} .

(soc·ann)

For a general ring only the inclusion $soc (M) \subseteq {ann}_{M} (rad R)$ holds.

Proof

The radical annihilates every simple left module, hence annihilates their sum, giving $soc (M) \subseteq {ann}_{M} (rad R)$ for any $R$ . Conversely put $N = {ann}_{M} (rad R)$ . Then $N$ is a module over $\bar{R} = R / rad R$ , and its $\bar{R}$ -submodules are exactly its $R$ -submodules. Since $\bar{R}$ is semisimple, $N$ is semisimple as an $\bar{R}$ -module by $(2.5) (2)$ , hence semisimple as an $R$ -module, hence contained in $soc (M)$ .

Corollary—Socle and semisimplicity of a ring

$R$ is left semisimple iff $soc (_{R} R) = R$ . Moreover a domain with nonzero left socle is a division ring: if $𝔞 = R a$ is a minimal left ideal with $a \neq 0$ , then $a^{2} \neq 0$ forces $R a^{2} = R a$ , so $a = r a^{2}$ for some $r$ , whence $(1 - r a) a = 0$ and $r a = 1$ ; then $(a r - 1) a = 0$ gives $a r = 1$ , so $a$ is a unit and $R a = R$ is simple as a left module.

Proof Techniques and Method

How these proofs work, and which move to reuse.

1. Use simplicity as a dichotomyA map out of a simple module has image

0

or a simple module; an intersection with a simple module is

0

or everything. Both the functoriality of the socle and the two-sidedness of

soc (_{R} R)

are this observation applied once.

2. Manufacture essential submodules with complementsFor any

N

, a Zorn-maximal

C

with

C \cap N = 0

makes

N \oplus C

essential. This is the standard way to produce essential submodules on demand.

3. Maximise subject to omitting an elementTo separate a specific element from a submodule, take a submodule maximal with respect to not containing it. The same move drives Lam's

(2.3)

; here it produces an essential submodule that misses

x

4. Convert to an annihilator to computeOver a semilocal ring

soc (M) = {ann}_{M} (rad R)

, so the socle becomes the kernel of a finite set of linear maps — the only form in which it is ever computed in practice.

Step 3 is worth isolating because it is the one that needs care. The submodule it produces is essential only after one knows there are no simple submodules around to obstruct it — which is why the proof first strips off $soc (M)$ and works inside a complement with zero socle.

Worked Example

Abelian groups

Over $ℤ$ the simple modules are the $ℤ / p$ . For $M = ℤ / 12$ the simple submodules are $⟨ 6 ⟩ ≅ ℤ / 2$ and $⟨ 4 ⟩ ≅ ℤ / 3$ , and

soc (ℤ / 12) = ⟨ 6 ⟩ + ⟨ 4 ⟩ = ⟨ 2 ⟩ ≅ ℤ / 2 \oplus ℤ / 3 ≅ ℤ / 6 .

(E.1)

Cross-check with the annihilator formula: $rad (ℤ / 12) = ⟨ 6 ⟩$ , and ${m : 6 m \equiv 0 (\mod 12)}$ is the set of even residues, which is $⟨ 2 ⟩$ . The two computations agree, as they must since $ℤ / 12$ is artinian.

Socles of standard $ℤ$ -modules
$M$	$soc (M)$	Essential in $M$ ?	Comment
$ℤ / p^{n}$	$⟨ p^{n - 1} ⟩ ≅ ℤ / p$	yes	uniserial, Loewy length $n$
$ℤ / 12$	$⟨ 2 ⟩ ≅ ℤ / 6$	yes	Loewy length $2$
$ℤ$	$0$	no	no minimal nonzero subgroup
$ℚ$	$0$	no	divisible and torsion-free
$ℤ (p^{\infty})$	$ℤ / p$	yes	the injective hull of $ℤ / p$
$⨁_{p} ℤ / p$	the whole module	yes	semisimple, infinite length
$\prod_{p} ℤ / p$	$⨁_{p} ℤ / p$	yes	socle does not commute with products

The last row is worth checking by hand. An element $x$ of $\prod_{p} ℤ / p$ of prime order $q$ must satisfy $q x = 0$ ; in a coordinate $p \neq q$ the integer $q$ is invertible modulo $p$ , so $x_{p} = 0$ . Hence $x$ is supported in the single coordinate $q$ , and the socle is the direct sum, strictly smaller than the product.

A ring whose two socles differ

Let $R = T_{2} (k)$ , the upper triangular $2 \times 2$ matrices over a field. Its radical is $rad R = (\begin{matrix} 0 & k \\ 0 & 0 \end{matrix})$ , and $R$ is artinian, so both socles can be computed as annihilators of the radical.

(\begin{matrix} 0 & y \\ 0 & 0 \end{matrix}) (\begin{matrix} a & b \\ 0 & c \end{matrix}) = (\begin{matrix} 0 & y c \\ 0 & 0 \end{matrix}), (\begin{matrix} a & b \\ 0 & c \end{matrix}) (\begin{matrix} 0 & y \\ 0 & 0 \end{matrix}) = (\begin{matrix} 0 & a y \\ 0 & 0 \end{matrix}) .

(E.2)

The first identity vanishes for all $y$ exactly when $c = 0$ ; the second exactly when $a = 0$ . Hence

soc (_{R} R) = (\begin{matrix} k & k \\ 0 & 0 \end{matrix}), soc (R_{R}) = (\begin{matrix} 0 & k \\ 0 & k \end{matrix}) .

(E.3)

Both are two-sided ideals of $R$ , both are two-dimensional, and they are different ideals.

Direct verification confirms the left socle: $R e_{11} = (\begin{matrix} k & 0 \\ 0 & 0 \end{matrix})$ and $(\begin{matrix} 0 & k \\ 0 & 0 \end{matrix})$ are minimal left ideals summing to it, while any element with $c \neq 0$ generates a left ideal properly containing $(\begin{matrix} 0 & k \\ 0 & 0 \end{matrix})$ and so lies in no minimal left ideal. Note $soc (_{R} R) \cap soc (R_{R}) = rad R$ and $soc (_{R} R) + soc (R_{R}) = R$ .

A socle series

For $M = ℤ / p^{n}$ the socle series is ${soc}_{i} (M) = ⟨ p^{n - i} ⟩$ , each layer isomorphic to $ℤ / p$ , terminating at $i = n$ . The Loewy length is $n$ , equal to the composition length because $M$ is uniserial.

Frameworks and Models

The socle series stratifies an arbitrary module into semisimple layers, from the bottom up. Each band below is contained in the next.

$M$ arbitrary module; the outermost layer need not be reached in finitely many steps

${soc}_{3} (M)$ third Loewy layer; quotient by

{soc}_{2}

is semisimple

${soc}_{2} (M)$ preimage of

soc (M / soc (M))

${soc}_{1} (M) = soc (M)$ largest semisimple submodule; contains every simple submodule

${soc}_{0} (M) = 0$ the base of the filtration

Three regimes are worth distinguishing. If $M$ has finite length the series terminates and its length is the Loewy length. If $R$ is left artinian and $M$ is arbitrary, the series is strictly increasing until it reaches $M$ , possibly transfinitely; modules for which it always reaches $M$ are the semiartinian ones. If $soc (M) = 0$ — as for $ℤ$ over $ℤ$ — the series never leaves $0$ and carries no information at all.

Comparison and Classification

Socle versus radical of a module
	Socle	Radical of a module
Built from	sum of minimal submodules	intersection of maximal submodules
Extremal property	largest semisimple submodule	smallest submodule with semisimple quotient, when one exists
Can be $0$	yes, e.g. $ℤ$ over $ℤ$	yes, e.g. $ℤ / 6$ over $ℤ$
Can be all of $M$	yes, exactly when $M$ is semisimple	yes, e.g. $ℚ$ over $ℤ$
For $M =_{R} R$	two-sided ideal, not side-symmetric	$rad R$ , two-sided and side-symmetric
Commutes with $⨁$	yes	yes
Commutes with $\prod$	no	yes for finite products

Socle versus radical of a module

Socles of some rings
Ring $R$	$soc (_{R} R)$	$soc (R_{R})$	Note
Division ring $D$	$D$	$D$	semisimple
$M_{n} (D)$	$M_{n} (D)$	$M_{n} (D)$	semisimple, simple artinian
$ℤ$	$0$	$0$	a domain that is not a division ring
$k [x]$	$0$	$0$	same reason
$ℤ / p^{n}$ , $n \geq 2$	$⟨ p^{n - 1} ⟩$	same	commutative, so both sides agree
$T_{2} (k)$	$(\begin{matrix} k & k \\ 0 & 0 \end{matrix})$	$(\begin{matrix} 0 & k \\ 0 & k \end{matrix})$	the two socles differ
${End}_{k} (V)$ , $\dim V$ infinite	finite-rank endomorphisms	same	the unique minimal two-sided ideal

Relationship Map

$soc (M)$ — largest semisimple submodule
- contains
  - every simple submodule of $M$
  - every semisimple submodule of $M$
  - $soc (N)$ for every submodule $N$
- is contained in
  - every essential submodule of $M$
  - ${ann}_{M} (rad R)$
  - $M$ , with equality iff $M$ is semisimple
- equals
  - the intersection of the essential submodules
  - ${ann}_{M} (rad R)$ when $R$ is semilocal
  - $⨁_{[S]} {soc}_{S} (M)$ , the isotypic decomposition

For rings, the left socle is the meeting point of two different theories: it is the sum of the minimal left ideals, which is where the structure theory of primitive rings begins, and it is the annihilator of the radical, which is where the module theory of artinian rings begins. Minimal Ideals and the Socle of a Primitive Ring develops the first; the artinian side is developed alongside the Wedderburn–Artin theorem.

Applications and Industry Use

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

Coding over rings

Frobenius rings and MacWilliams identities

A finite ring is Frobenius exactly when its socle is isomorphic to $R / rad R$ as a one-sided module. Wood's theorem shows this is precisely the condition under which the MacWilliams extension theorem holds for codes over the ring, which is why $ℤ / 4$ codes behave well and general finite rings do not.

Modular representation theory

Socle series of projectives

For $k G$ in characteristic dividing $| G |$ , the Loewy structure of the principal indecomposable modules is the standard invariant used to distinguish blocks; socle and radical series are computed routinely for this purpose.

Injective hulls

Essential extensions

The socle is the intersection of the essential submodules, so it controls essential extensions and hence injective hulls. Over a noetherian ring the indecomposable injectives are the hulls of the modules $R / 𝔭$ , and their socles identify them.

Operator theory

Finite-rank ideals

In ${End}_{k} (V)$ for infinite-dimensional $V$ , the socle is the ideal of finite-rank maps — the algebraic ancestor of the compact operators, and the minimal nonzero two-sided ideal of the ring.

Honest summary: the socle is a measuring instrument. It is rarely interesting for its own sake, but it is the standard way to extract a semisimple invariant from a module that is not semisimple.

Standards and Notation

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

Preferred notation

soc (M)

, lower case, as in Lam

Common variant

Soc (M)

in Anderson–Fuller and much of the module-theory literature

Sides

soc (_{R} R)

and

soc (R_{R})

must be distinguished; they can differ

Essential

N \subseteq_{e} M

, or the words essential or large

Socle seriesindexed from

{soc}_{0} = 0

here; some authors start at

{soc}_{1}

MarkupPresentation MathML per ISO/IEC 40314; symbol conventions per ISO 80000-2

ImplementationsGAP MTX for modules over finite fields; Magma Socle; Sage via the MeatAxe interface

Computational Notes

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

Finite-dimensional algebras. Compute $rad A$ , take a basis $r_{1}, \dots, r_{m}$ , and form $soc (M) = ⋂_{i} \ker (r_{i} : M \to M)$ . With $d = \dim_{k} M$ this is $m$ nullspace computations, each $O (d^{3})$ field operations, and the intersection is another linear-algebra step.
Socle series. Iterate on $M / {soc}_{n} (M)$ . For a module of length $ℓ$ the series has at most $ℓ$ terms, so the total cost stays polynomial in $\dim_{k} M$ .
Modules over finite fields. The MeatAxe computes composition factors and exposes radical and socle bases; GAP's MTX interface and Magma's module machinery both provide the socle directly.
Rings. The left socle of a finite-dimensional algebra is the left annihilator of $rad A$ , a single kernel computation; the right socle is a different kernel and must be computed separately.
Limits. For infinite-dimensional modules there is no general algorithm: even deciding whether a nonzero simple submodule exists is out of reach without strong finiteness hypotheses.

Failure Modes and Common Mistakes

Do not use $soc (M) = {ann}_{M} (rad R)$ without checking that $R / rad R$ is semisimple. Over $ℤ$ the right-hand side is all of $M$ because $rad ℤ = 0$ , while the socle is usually much smaller.
Do not confuse $soc (M)$ with the set of elements of finite order, or with any torsion condition; the correspondence is an accident of the $ℤ$ case.
Do not assume the socle series reaches $M$ . It does for modules of finite length, and never gets off the ground when $soc (M) = 0$ .
Do not conflate the socle of a ring with its minimal two-sided ideals. The left socle is a sum of minimal left ideals; it happens to be two-sided, but its summands are not.

Quick Reference

Definition

soc (M) = \sum {S \subseteq M : S simple}

Extremallargest semisimple submodule of

M

Essential form

soc (M) = ⋂ {E : E essential in M}

Annihilator form

soc (M) = {ann}_{M} (rad R)

R

is semilocal

Semisimplicity

M

semisimple

⟺

soc (M) = M

Ring socle

soc (_{R} R)

is a two-sided ideal

Asymmetryleft socle

\neq

right socle in general

Artinian case

soc (M)

is essential in

M

, and nonzero if

M \neq 0

Facts and their hypotheses
Statement	Hypothesis needed	Fails without it
$soc (M) \neq 0$ for $M \neq 0$	$R$ left artinian, or $M$ of finite length	$soc (ℤ) = 0$
$soc (M)$ essential in $M$	$R$ left artinian, or $M$ semiartinian	$soc (ℚ) = 0$ is not essential
$soc (M) = {ann}_{M} (rad R)$	$R / rad R$ semisimple	false over $ℤ$
left socle $=$ right socle	$R$ commutative, or semisimple	false for $T_{2} (k)$
socle commutes with the construction	direct sums	false for infinite products
socle series reaches $M$	$M$ of finite length, or semiartinian	stalls at $0$ for $ℤ$

Frequently Asked Questions

Can the socle of a nonzero module be zero?

Yes. $soc (ℤ) = 0$ because no subgroup $n ℤ$ is minimal, and $soc (ℚ) = 0$ for the same reason. It cannot happen over a left artinian ring: there every nonzero module has a simple submodule, since every cyclic submodule is artinian.

Is the left socle of a ring equal to its right socle?

Not in general. For $R = T_{2} (k)$ the left socle is the top row and the right socle is the second column, two different two-sided ideals of the same dimension. They agree when $R$ is commutative and when $R$ is semisimple, since then both are $R$ . This is a real difference from the Jacobson radical, which is always side-neutral.

How is the socle related to the radical of a module?

They are dual constructions: the socle sums the minimal submodules, the radical intersects the maximal ones. The socle is the largest semisimple submodule, whereas $M / rad (M)$ is the largest semisimple quotient when it exists. Over a semilocal ring the socle can be written as the annihilator of $rad R$ acting on $M$ , which is how it is computed.

Why is the left socle a two-sided ideal when minimal left ideals are not?

Because right multiplication by a fixed element is a homomorphism of left modules. It carries a minimal left ideal to zero or to another minimal left ideal, so it preserves the sum of all of them even though it moves the individual summands around.

What does the socle series tell me that the composition series does not?

A composition series records the multiset of composition factors but not how they are glued. The socle series records the gluing depth: the Loewy length is the number of semisimple layers needed to build the module, so a uniserial module of length $n$ has Loewy length $n$ while a semisimple module of length $n$ has Loewy length $1$ .

Is the socle preserved by ring homomorphisms or base change?

Not in general. Restricting scalars along $S \to R$ can change which submodules are simple, and extending the base field can split a simple module into several. The socle is functorial in the module argument only, for a fixed ring.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §2, result (2.3), pp. 26–27; the socle of a ring reappears in §11.
T. Y. Lam, Lectures on Modules and Rings, Graduate Texts in Mathematics 189, Springer-Verlag, 1999, §8 (essential submodules, socle and singular submodule).
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §9 (the socle and the radical of a module).
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapter IV (rings with minimal one-sided ideals).
J. A. Wood, “Duality for modules over finite rings and applications to coding theory”, American Journal of Mathematics 121 (1999).

AI Suggested Questions

Prove that a module is semiartinian if and only if its transfinite socle series exhausts it.
Construct a ring whose left socle is zero and whose right socle is nonzero.
Prove that a finite ring is Frobenius if and only if its right socle is isomorphic to the quotient by its radical.
Compute the socle series of the projective indecomposable modules of the group algebra of the symmetric group on three letters in characteristic three.
Show that the socle of an injective module over a noetherian ring determines the module up to isomorphism when the socle is essential and finitely generated.
Explain how the socle of the endomorphism ring of an infinite-dimensional vector space relates to the ideal of compact operators on a Hilbert space.
Give an example where the socle series and the radical series of a module have different lengths.

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

The socle is a functor

Behaviour under sums and products

Socle and radical as a dual pair

The socle series

06Key Results

07Proof Techniques and Method

08Worked Example

Abelian groups

A ring whose two socles differ

A socle series

09Frameworks and Models

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

Frobenius rings and MacWilliams identities

Socle series of projectives

Essential extensions

Finite-rank ideals

13Standards and Notation

14Computational Notes

15Failure Modes and Common Mistakes

16Quick Reference

17Frequently Asked Questions

18Related KEVOS Topics

19References

20AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Worked Example

Frameworks and Models

Comparison and Classification

Relationship Map

Applications and Industry Use

Standards and Notation

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions