Structure of Finite-Dimensional Algebras and Their Simple Modules

Q: Why is R/rad R semisimple here without further hypotheses?

Because _k R < ∈fty forces R to be left artinian, and for a left artinian ring the radical is nilpotent and the quotient is semisimple. Finite dimensionality is doing the work; over a general ring the quotient by the radical need not be semisimple, only semiprimitive.

Q: Does R itself decompose as a product like R does?

No. The decomposition of R into simple components uses central idempotents of R, and those need not lift to central idempotents of R in general. What does lift, for R semiperfect (in particular finite-dimensional), is a decomposition of 1 into primitive orthogonal idempotents, which gives the principal indecomposable modules rather than a product decomposition of the ring.

Q: Is the surjection R → End(M_i)_D_i ever an isomorphism?

Exactly when rad R = 0 and r = 1, i.e. when R is simple artinian. In general the kernel is rad R together with all simple components other than the i-th.

Q: Where is finite dimensionality actually used in Burnside's Lemma?

Only in the last step. Density gives an element of A agreeing with a prescribed f on any finite tuple of vectors; finite dimensionality lets that tuple be a whole basis, so the element equals f. For infinite-dimensional M one gets density, not equality.

Q: Can two different simple modules share the same division algebra D?

Yes, and it is the normal situation. In RS_3 all three simple modules have D_i = R while n_i takes the values 1, 1, 2. The D_i record arithmetic, the n_i record size; they vary independently.

Q: What replaces this analysis when the algebra is not semisimple?

The simple modules are no longer enough: one studies the principal indecomposable modules, the Cartan matrix recording composition multiplicities, and the indecomposable modules generally. The simple modules and (7.2) remain the starting data for all of it.

Executive Summary

A finite-dimensional $k$ -algebra $R$ is left and right artinian, so $\bar{R} : = R / rad R$ is semisimple and Wedderburn–Artin gives $\bar{R} ≅ M_{n_{1}} (D_{1}) \times \dots \times M_{n_{r}} (D_{r})$ . What finite dimensionality adds is bookkeeping: every object in sight — $R$ , $rad R$ , each simple module $M_{i}$ , each division algebra $D_{i}$ — is a finite-dimensional $k$ -vector space, and the decomposition becomes an identity between integers.

That identity, $\dim_{k} R = \dim_{k} rad R + \sum_{i} n_{i}^{2} \dim_{k} D_{i}$ , is the workhorse of the whole chapter. It is how one checks a proposed list of simple modules is complete, how splitting fields are recognised, and how character theory gets its numerical grip.

$r$ Simple modules, up to isomorphism

$n_{i} \dim_{k} D_{i}$

\dim_{k} M_{i}

$\sum n_{i}^{2} \dim_{k} D_{i}$

\dim_{k} \bar{R}

1905Burnside's theorem

Overview

Throughout, $k$ is a field of arbitrary characteristic and $R$ is a $k$ -algebra with $\dim_{k} R < \infty$ . Subalgebras and quotients of $R$ are again finite-dimensional, hence left and right artinian, so the artinian theory of Chapter 1 applies to all of them at once. In particular $rad R$ is nilpotent and $\bar{R} = R / rad R$ is semisimple.

The semisimple theory alone gives a decomposition; it does not tell you how big the pieces are. Over a field, dimension counting closes that gap. Two facts do the work: a module over a division algebra $D$ has a well-defined $D$ -dimension, and $\dim_{k}$ of a $D$ -space of dimension $n$ is $n \dim_{k} D$ . Chaining these through Wedderburn–Artin yields the formulas below.

The special case $D_{i} = k$ for all $i$ is where classical representation theory lives: then $\dim_{k} M_{i} = n_{i}$ and the sum becomes $\sum n_{i}^{2}$ , exactly the familiar identity from the character theory of finite groups. Deciding when that case occurs is the subject of Absolutely Irreducible Modules and Splitting Fields for Algebras.

Learning Objectives

Fix the notation $B_{i}$ , $M_{i}$ , $D_{i}$ , $n_{i}$ used throughout the chapter and know what each symbol denotes.
Prove $\dim_{k} M_{i} = n_{i} \dim_{k} D_{i}$ and the global dimension count for $\dim_{k} R$ .
Justify the surjectivity of $R \to End (M_{i})_{D_{i}}$ from Wedderburn–Artin.
State Burnside's Lemma correctly, including the hypothesis $End (_{A} M) = k$ , and prove it from density.
Show $D_{i} = k$ whenever $k$ is algebraically closed, and produce a counterexample over $ℚ$ .
Carry out the dimension audit on a concrete algebra with a nonzero radical.

Definitions

Definition(7.1)Standing notation for a finite-dimensional algebra

Let $R$ be a $k$ -algebra with $\dim_{k} R < \infty$ and put $\bar{R} = R / rad R$ . Write $\bar{R} ≅ B_{1} \times \dots \times B_{r}$ for the decomposition of the semisimple ring $\bar{R}$ into its simple components, and let $M_{i}$ be the unique simple left $B_{i}$ -module. Then $M_{1}, \dots, M_{r}$ is a complete, irredundant list of the simple left $\bar{R}$ -modules and — because every simple left $R$ -module is killed by $rad R$ — a complete, irredundant list of the simple left $R$ -modules.

Set $D_{i} = End (_{B_{i}} M_{i}) = End (_{R} M_{i})$ , a division ring by Schur's Lemma, and $n_{i} = \dim_{D_{i}} M_{i}$ where $M_{i}$ is regarded as a right $D_{i}$ -vector space. Wedderburn–Artin then reads

\bar{R} ≅ M_{n_{1}} (D_{1}) \times \dots \times M_{n_{r}} (D_{r}),_{R} \bar{R} ≅ n_{1} M_{1} \oplus \dots \oplus n_{r} M_{r}

(7.1)

$B_{i} ≅ End (M_{i})_{D_{i}} ≅ M_{n_{i}} (D_{i})$ ; the multiplicity of $M_{i}$ in the regular module equals $n_{i}$ .

$\bar{R}$: The semisimple quotient $R / rad R$ . Since $R$ is artinian, $rad R$ is nilpotent and $\bar{R}$ is semisimple.
$B_{i}$: The $i$ -th simple component of $\bar{R}$ : a simple artinian ring, and a two-sided ideal of $\bar{R}$ generated by a central idempotent.
$M_{i}$: The unique simple left $B_{i}$ -module, viewed as a simple left $R$ -module via $R ↠ \bar{R} ↠ B_{i}$ .
$D_{i}$: $End (_{R} M_{i})$ , written on the right of $M_{i}$ so that $M_{i}$ is a right $D_{i}$ -vector space and no opposite ring is needed.
$n_{i}$: $\dim_{D_{i}} M_{i}$ , equivalently the multiplicity of $M_{i}$ in the left regular module $_{R} R$ .

Writing endomorphisms as right operators is a convention, not a theorem. It is chosen precisely so that $End (M_{i})_{D_{i}} ≅ M_{n_{i}} (D_{i})$ rather than $M_{n_{i}} (D_{i}^{op})$ .

Core Concepts

Why the radical is the only obstruction

Simple left $R$ -modules and simple left $\bar{R}$ -modules are the same objects: $rad R$ annihilates every simple module, so the $R$ -action factors through $\bar{R}$ . Whatever is genuinely simple about $R$ is therefore already visible in $\bar{R}$ , and $rad R$ contributes only its dimension to the count.

R

\bar{R} = R / rad R

B_{i}

M_{n_{i}} (D_{i})

Each arrow is a surjection of $k$ -algebras, and the composite is exactly the representation $R \to End (M_{i})_{D_{i}}$ afforded by $M_{i}$ . Its surjectivity — part (3) of the proposition below — is the abstract form of Burnside's theorem, and it is what lets one manufacture elements of $R$ acting on the simple modules in any prescribed way.

Two dimensions, two indices

The single commonest source of error in this material is confusing $n_{i} = \dim_{D_{i}} M_{i}$ with $\dim_{k} M_{i}$ . They agree only when $D_{i} = k$ . For $R = ℍ$ over $k = ℝ$ , the unique simple module is $ℍ$ itself: $n_{1} = 1$ but $\dim_{ℝ} M_{1} = 4$ .

Key Results

Proposition(7.2)Dimension formulas

Let $R$ be a $k$ -algebra with $\dim_{k} R < \infty$ , in the notation of $(7.1)$ . Then:

$\dim_{k} M_{i} = n_{i} \dim_{k} D_{i}$ for each $i$ ;
$\dim_{k} R = \dim_{k} rad R + \sum_{i = 1}^{r} n_{i}^{2} \dim_{k} D_{i}$ ;
the natural map $R \to End (M_{i})_{D_{i}}$ giving the action of $R$ on $M_{i}$ is surjective, for each $i$ .

Proof

(1) As a right $D_{i}$ -vector space $M_{i}$ has dimension $n_{i}$ , so $M_{i} ≅ D_{i}^{n_{i}}$ as right $D_{i}$ -modules. The field $k$ acts centrally on $M_{i}$ and its action agrees with the one it induces on $D_{i}$ , so this is also an isomorphism of $k$ -vector spaces; taking $k$ -dimensions gives $\dim_{k} M_{i} = n_{i} \dim_{k} D_{i}$ .

(2) The exact sequence $0 \to rad R \to R \to \bar{R} \to 0$ of $k$ -spaces gives $\dim_{k} R = \dim_{k} rad R + \dim_{k} \bar{R}$ . By $(7.1)$ , $\dim_{k} \bar{R} = \sum_{i} \dim_{k} M_{n_{i}} (D_{i}) = \sum_{i} n_{i}^{2} \dim_{k} D_{i}$ , since $M_{n_{i}} (D_{i})$ is a free $D_{i}$ -module on $n_{i}^{2}$ matrix units.

(3) The action of $R$ on $M_{i}$ kills $rad R$ and kills $B_{j}$ for $j \neq i$ , so it factors as $R ↠ \bar{R} ↠ B_{i} \to End (M_{i})_{D_{i}}$ . Both projections are onto, and the last map is an isomorphism by Wedderburn–Artin. Hence the composite is onto.

Lemma(7.3)Burnside's Lemma

Let $M$ be a finite-dimensional right $k$ -vector space and let $A \subseteq End (M_{k})$ be a $k$ -subalgebra such that $M$ is simple as a left $A$ -module. If $End (_{A} M) = k$ , then $A = End (M_{k})$ .

The hypothesis $End (_{A} M) = k$ means the only $A$ -module endomorphisms of $M$ are the scalars; it is not automatic and cannot be dropped.

Proof

Because $A$ sits inside $End (M_{k})$ , the module $_{A} M$ is faithful; it is simple by hypothesis, and $D : = End (_{A} M) = k$ . The Jacobson Density Theorem therefore says $A$ is dense in $End (M_{D}) = End (M_{k})$ : for any finite set $m_{1}, \dots, m_{s} \in M$ that is $D$ -independent and any $y_{1}, \dots, y_{s} \in M$ , some $a \in A$ satisfies $a m_{j} = y_{j}$ for all $j$ .

Now take $m_{1}, \dots, m_{n}$ to be a $k$ -basis of $M$ , where $n = \dim_{k} M < \infty$ , and let $f \in End (M_{k})$ be arbitrary. Density supplies $a \in A$ with $a m_{j} = f (m_{j})$ for every $j$ . Two $k$ -linear maps agreeing on a basis are equal, so $a = f$ and $A = End (M_{k})$ .

PropositionThe algebraically closed case

Let $k$ be algebraically closed and $R$ a $k$ -algebra with $\dim_{k} R < \infty$ . Then $D_{i} = k$ for every $i$ , so

\bar{R} ≅ M_{n_{1}} (k) \times \dots \times M_{n_{r}} (k), \dim_{k} M_{i} = n_{i}, \dim_{k} R = \dim_{k} rad R + \sum_{i = 1}^{r} n_{i}^{2}

(7.2′)

Proof

Each $D_{i}$ is a division ring, finite-dimensional over $k$ , with $k$ central. Given $d \in D_{i}$ , the subring $k [d]$ is a commutative finite-dimensional $k$ -algebra with no zero divisors, hence a field, hence a finite field extension of $k$ . As $k$ is algebraically closed, $k [d] = k$ and $d \in k$ . Thus $D_{i} = k$ , and the displayed formulas are $(7.2)$ specialised to $\dim_{k} D_{i} = 1$ .

CorollaryBurnside's Theorem, classical form

Let $k$ be algebraically closed, $M$ a finite-dimensional $k$ -vector space and $A \subseteq End (M_{k})$ a $k$ -subalgebra acting irreducibly on $M$ . Then $A = End (M_{k})$ .

Indeed $End (_{A} M)$ is a division algebra, finite-dimensional over the algebraically closed field $k$ , hence equal to $k$ by the argument just given; Burnside's Lemma applies with no extra hypothesis.

Counterexample

D

need not be

k

Take $k = ℚ$ and $R = ℚ G$ where $G = ⟨ g ⟩$ is cyclic of order $3$ . The surjection $ℚ G \to ℚ (ω)$ , $g \mapsto ω$ (a primitive cube root of unity), makes $M = ℚ (ω)$ a simple left $R$ -module. Its $R$ -endomorphisms are the right multiplications by elements of $ℚ (ω)$ , so $D = End (_{R} M) = ℚ (ω) \neq ℚ$ , with $n = 1$ and $\dim_{ℚ} M = 2 = n \dim_{ℚ} D$ .

Proof Techniques and Method

How these proofs work, and which move to reuse.

Three moves recur throughout the section and are worth naming.

Move 1

Count in two stages

Compute $\dim_{D} M$ first, then multiply by $\dim_{k} D$ . Almost every numerical statement in this chapter is this one step, applied to a different module.

Move 2

Manufacture elements by surjectivity

Because $R \to End (M_{i})_{D_{i}}$ is onto and $\bar{R}$ is a product, one can choose $a \in R$ acting as any prescribed map on $M_{i}$ and as zero on every $M_{j}$ , $j \neq i$ . This is the engine of the character arguments in $(7.19)$ and $(7.20)$ .

Move 3

Density, then finite dimension

Density gives agreement on any finite tuple; finite dimensionality upgrades that to equality of maps. Burnside's Lemma is exactly this two-line upgrade.

Move 2 deserves emphasis. The idempotents ${\bar{e}}_{i} \in \bar{R}$ that are the identities of the simple components are central in $\bar{R}$ ; lifting them to arbitrary preimages $a_{i} \in R$ is enough for character computations, because only the action on semisimple subquotients matters. No idempotent lifting theorem is required for that.

Worked Example

A semisimple case: $R = ℝ S_{3}$

Here $char ℝ = 0$ does not divide $| S_{3} | = 6$ , so $R$ is semisimple by Maschke's Theorem and $rad R = 0$ . The three irreducible representations — trivial, sign, and the two-dimensional standard representation — are all realisable over $ℝ$ , so

ℝ S_{3} ≅ ℝ \times ℝ \times M_{2} (ℝ), r = 3, (n_{1}, n_{2}, n_{3}) = (1, 1, 2), D_{1} = D_{2} = D_{3} = ℝ

The audit: $\dim_{ℝ} R = 6$ and $0 + 1^{2} \cdot 1 + 1^{2} \cdot 1 + 2^{2} \cdot 1 = 6$ . Consistent. The module dimensions are $1, 1, 2$ , matching $n_{i} \dim_{ℝ} D_{i}$ .

A case with a nonzero radical and a noncommutative $D$

Let $k = ℝ$ and $R = ℍ [x] / (x^{2})$ , the algebra of dual quaternions $ℍ \oplus ℍ x$ with $x$ central and $x^{2} = 0$ . Then $\dim_{ℝ} R = 8$ .

The ideal $ℍ x$ is nilpotent of index $2$ , and the quotient $R / ℍ x ≅ ℍ$ is a division ring, hence semisimple with zero radical. Therefore $rad R = ℍ x$ , of $ℝ$ -dimension $4$ .

So $\bar{R} ≅ ℍ$ , giving $r = 1$ , $M_{1} = ℍ$ with $D_{1} = End (_{R} M_{1}) = ℍ$ acting on the right, and $n_{1} = \dim_{ℍ} ℍ = 1$ .

\dim_{ℝ} M_{1} = 4 = n_{1} \dim_{ℝ} D_{1}, \dim_{ℝ} R = 8 = {\underset{⏟}{4}}_{\dim rad R} + 1^{2} \cdot 4

Both parts of $(7.2)$ check out, and part (3) says $R \to End (M_{1})_{ℍ} = ℍ$ is onto — visibly true, since it is the quotient by $ℍ x$ .

Comparison and Classification

Decomposition data for small algebras
Algebra $R$ over $k$	$\dim_{k} R$	$rad R$	$\bar{R}$	$(n_{i}; \dim_{k} D_{i})$
$M_{n} (k)$	$n^{2}$	$0$	$M_{n} (k)$	$(n; 1)$
$ℍ$ over $ℝ$	$4$	$0$	$ℍ$	$(1; 4)$
$ℂ$ over $ℝ$	$2$	$0$	$ℂ$	$(1; 2)$
$ℚ C_{3}$	$3$	$0$	$ℚ \times ℚ (ω)$	$(1; 1), (1; 2)$
$ℝ S_{3}$	$6$	$0$	$ℝ \times ℝ \times M_{2} (ℝ)$	$(1; 1), (1; 1), (2; 1)$
$T_{2} (k)$ , upper triangular	$3$	strictly upper, $\dim 1$	$k \times k$	$(1; 1), (1; 1)$
$k [x] / (x^{n})$	$n$	$(x)$ , $\dim n - 1$	$k$	$(1; 1)$
$𝔽_{2} S_{3}$	$6$	$\dim 1$	$𝔽_{2} \times M_{2} (𝔽_{2})$	$(1; 1), (2; 1)$

Which conclusions need which hypotheses
	$\dim_{k} R < \infty$	$R$ semisimple	$k$ algebraically closed
$\bar{R}$ is a finite product of $M_{n_{i}} (D_{i})$	yes	yes	yes
$rad R$ is nilpotent	yes	yes	yes
$\dim_{k} R = \dim_{k} rad R + \sum n_{i}^{2} \dim_{k} D_{i}$	yes	yes	yes
$D_{i} = k$ for all $i$	no	no	yes
$\dim_{k} M_{i} = n_{i}$	no	no	yes
Every simple module is a direct summand of $_{R} R$	no	yes	partial

Which conclusions need which hypotheses

Relationship Map

The section sits between the general artinian theory and the field-theoretic refinements that follow.

RingsJacobson radical defined; simple modules kill

rad R

Left artinian rings

rad R

nilpotent,

\bar{R}

semisimple, finitely many simple modules

Finite-dimensional $k$ -algebraseverything acquires a

k

-dimension;

(7.2)

available

Split algebrasall

D_{i} = k

;

\dim_{k} R = \dim_{k} rad R + \sum_{i} (\dim_{k} M_{i})^{2}

Semisimple split algebras

R ≅ \prod M_{n_{i}} (k)

$(7.2)$ feeds
- directly
  - Burnside's Lemma $(7.3)$ , via part (3)
  - the splitting criterion $(7.8)$ , via part (2)
  - the character arguments $(7.19)$ – $(7.20)$ , via part (3)
- indirectly
  - counting simple modules through $rad R + [R, R]$
  - Cartan matrices and principal indecomposable modules
  - the dimension bookkeeping in modular representation theory

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

Dimension audits

The formula $\sum n_{i}^{2} = | G |$ for $ℂ G$ is $(7.2) (2)$ with $rad = 0$ and $D_{i} = ℂ$ . In characteristic $p$ the same equation, with the radical term restored, is how the modular irreducibles are pinned down.

Symbolic computation

Algebra recognition

Computer algebra systems present a finite-dimensional algebra by structure constants, compute the radical, then split the semisimple quotient. The output is exactly the tuple $(r; n_{i}; D_{i})$ described here.

Coding theory

Codes as ideals

Cyclic and abelian codes are ideals in $𝔽_{q} G$ ; the Wedderburn decomposition of that algebra lists the minimal ideals, hence the minimal codes, and the $D_{i}$ are the field extensions over which the idempotents live.

Quantum information

Symmetry-adapted bases

Decomposing an operator algebra generated by a symmetry group into $\prod M_{n_{i}} (D_{i})$ block-diagonalises every operator commuting with the symmetry — the numerical payoff of the abstract decomposition.

The honest summary: this material is infrastructure. Its value is that it converts structural questions about an algebra into integer arithmetic that a person or a machine can check.

Design Considerations

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

Which side. Left modules are used throughout; the right-handed theory is obtained by applying everything to $R^{op}$ . The integers $r$ and $n_{i}$ are the same on both sides, because ${\bar{R}}^{op} ≅ \prod M_{n_{i}} (D_{i}^{op})$ .
**Where to put $D_{i}$ .** Composing module endomorphisms as right operators keeps $End (M_{i})_{D_{i}} ≅ M_{n_{i}} (D_{i})$ ; the left-operator convention forces $D_{i}^{op}$ into every formula.
Which ground field. Enlarging $k$ shrinks the $D_{i}$ but can enlarge the radical when the extension is inseparable. Choose the ground field before, not after, computing the radical.
Whether to quotient early. For questions about simple modules, replace $R$ by $\bar{R}$ immediately. For questions about projectives, indecomposables or extensions, do not — that information lives entirely in $rad R$ .

Standards and Notation

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

Simple components

B_{1}, \dots, B_{r}

(Lam);

A_{i}

e_{i} R

elsewhere

Endomorphism ring

End (_{R} M)

, right operators (Lam);

{End}_{R} (M)

acting on the left is equally common

Matrix ring

M_{n} (D)

; some texts write

D_{n}

{Mat}_{n} (D)

Radical

rad R

;

J (R)

in module-theoretic sources

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

GAPRadicalOfAlgebra, WedderburnDecomposition (Wedderga)

Magma / SageWedderburnDecomposition, A.radical()

Computational Notes

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

Let $A$ be given by structure constants with $\dim_{k} A = n$ .

In characteristic $0$ , $rad A$ is the radical of the trace form $(a, b) \mapsto tr (L_{ab})$ — one nullspace computation, $O (n^{3})$ field operations.
In characteristic $p$ the trace form is insufficient; the Friedl–Rónyai algorithm iterates a chain of higher trace conditions and remains polynomial time.
Splitting the semisimple quotient over a finite field is randomised polynomial time (Rónyai); the MeatAxe does the module-theoretic version by searching for singular elements with nontrivial kernel.
Over $ℚ$ the picture changes: deciding whether a simple $ℚ$ -algebra is a matrix algebra, and exhibiting the isomorphism, is at least as hard as factoring integers (Rónyai). The $D_{i}$ are the hard part, not the $n_{i}$ .

Failure Modes and Common Mistakes

Do not assume the multiplicity of $M_{i}$ in $_{R} R$ equals $\dim_{k} M_{i}$ ; it equals $n_{i}$ .
Do not conclude from $r = 1$ that $R$ is simple: $k [x] / (x^{n})$ has one simple module and a large radical.
Do not transport the decomposition of $\bar{R}$ back to $R$ as a direct product; $R$ need not decompose at all.

Quick Reference

Setting

k

a field,

\dim_{k} R < \infty

\bar{R} = R / rad R

Wedderburn

\bar{R} ≅ M_{n_{1}} (D_{1}) \times \dots \times M_{n_{r}} (D_{r})

Simple modules

M_{1}, \dots, M_{r}

; the same for

R

and

\bar{R}

$D_{i}$

End (_{R} M_{i})

, a division ring, right operators

$n_{i}$

\dim_{D_{i}} M_{i}

= multiplicity of

M_{i}

_{R} R

Module dimension

\dim_{k} M_{i} = n_{i} \dim_{k} D_{i}

Global count

\dim_{k} R = \dim_{k} rad R + \sum n_{i}^{2} \dim_{k} D_{i}

Surjectivity

R ↠ End (M_{i})_{D_{i}}

$k$ algebraically closed

D_{i} = k

\dim_{k} M_{i} = n_{i}

Statement finder
Result	Content	Reference
Notation	$B_{i}$ , $M_{i}$ , $D_{i}$ , $n_{i}$ and the two decompositions	(7.1)
Dimension formulas	$\dim_{k} M_{i}$ , $\dim_{k} R$ , surjectivity onto $End (M_{i})_{D_{i}}$	(7.2)
Burnside's Lemma	irreducible action plus $End (_{A} M) = k$ forces $A = End (M_{k})$	(7.3)
Algebraically closed case	$D_{i} = k$ ; $\dim_{k} R = \dim_{k} rad R + \sum n_{i}^{2}$	after (7.3)
Simple modules of $R$ and $\bar{R}$	identical, since $rad R$ annihilates simples	(4.8)

Frequently Asked Questions

Why is

R / rad R

semisimple here without further hypotheses?

Because $\dim_{k} R < \infty$ forces $R$ to be left artinian, and for a left artinian ring the radical is nilpotent and the quotient is semisimple. Finite dimensionality is doing the work; over a general ring the quotient by the radical need not be semisimple, only semiprimitive.

Does

R

itself decompose as a product like

\bar{R}

does?

No. The decomposition of $\bar{R}$ into simple components uses central idempotents of $\bar{R}$ , and those need not lift to central idempotents of $R$ in general. What does lift, for $R$ semiperfect (in particular finite-dimensional), is a decomposition of $1$ into primitive orthogonal idempotents, which gives the principal indecomposable modules rather than a product decomposition of the ring.

Is the surjection

R \to End (M_{i})_{D_{i}}

ever an isomorphism?

Exactly when $rad R = 0$ and $r = 1$ , i.e. when $R$ is simple artinian. In general the kernel is $rad R$ together with all simple components other than the $i$ -th.

Where is finite dimensionality actually used in Burnside's Lemma?

Only in the last step. Density gives an element of $A$ agreeing with a prescribed $f$ on any finite tuple of vectors; finite dimensionality lets that tuple be a whole basis, so the element equals $f$ . For infinite-dimensional $M$ one gets density, not equality.

Can two different simple modules share the same division algebra

D

Yes, and it is the normal situation. In $ℝ S_{3}$ all three simple modules have $D_{i} = ℝ$ while $n_{i}$ takes the values $1, 1, 2$ . The $D_{i}$ record arithmetic, the $n_{i}$ record size; they vary independently.

What replaces this analysis when the algebra is not semisimple?

The simple modules are no longer enough: one studies the principal indecomposable modules, the Cartan matrix recording composition multiplicities, and the indecomposable modules generally. The simple modules and $(7.2)$ remain the starting data for all of it.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §7 (pp. 107–110).
C. W. Curtis and I. Reiner, Representation Theory of Finite Groups and Associative Algebras, Wiley-Interscience, 1962, §§25–26.
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapter IV.
R. S. Pierce, Associative Algebras, Graduate Texts in Mathematics 88, Springer-Verlag, 1982, Chapters 3–4.
L. Rónyai, “Computing the structure of finite algebras”, Journal of Symbolic Computation 9 (1990), 355–373.

AI Suggested Questions

Work out the full decomposition data $(r; n_{i}; D_{i})$ for $ℚ Q_{8}$ and for $ℝ Q_{8}$ , and compare.
How does the Cartan matrix of a finite-dimensional algebra refine the dimension formula (7.2)(2)?
Give an example of a finite-dimensional algebra over $ℚ$ whose endomorphism division algebra is noncommutative.
What is the analogue of (7.2) for an artinian ring that is not an algebra over a field?
Explain how the MeatAxe algorithm finds the simple constituents of a module over a finite-dimensional algebra over a finite field.
Which parts of (7.2) survive if $k$ is replaced by a commutative artinian base ring?
How do the integers $n_{i}$ change when the ground field is extended, and when do they stabilise?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Why the radical is the only obstruction

Two dimensions, two indices

06Key Results

07Proof Techniques and Method

Count in two stages

Manufacture elements by surjectivity

Density, then finite dimension

08Worked Example

A semisimple case: R=ℝS3

A case with a nonzero radical and a noncommutative D

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

Dimension audits

Algebra recognition

Codes as ideals

Symmetry-adapted bases

12Design Considerations

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

A semisimple case: $R = ℝ S_{3}$

A case with a nonzero radical and a noncommutative $D$

Comparison and Classification

Relationship Map

Applications and Industry Use

Design Considerations

Standards and Notation

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions