Burnside’s Theorem on Irreducible Subalgebras of Matrix Rings

Q: Is Burnside's theorem just the density theorem?

It is the density theorem plus two hypotheses that each remove one gap. Density holds for any simple module over any ring and says R approximates End_D V on finite sets. Finite D-dimension makes approximation exact; algebraic closure identifies D with k. Neither is automatic, and each has its own counterexample — the Weyl algebra on k[x], and C inside M_2(R).

Q: How do I check absolute irreducibility in practice?

Compute the centraliser. Solve the linear system f : f g_i = g_i f for all generators g_i; its solution space always contains the scalars, and the module is absolutely irreducible exactly when the space is one-dimensional. This is an n^2 × n^2 nullspace computation and is what the MeatAxe implementations report.

Q: Does the theorem hold for subalgebras without an identity element?

Yes, provided irreducibility is taken in the strong sense that V is a simple module, in particular RV ≠ 0. Then RV is a nonzero submodule, so RV = V, and the density argument runs unchanged. The pathological case excluded is R acting as zero on a one-dimensional space, which has no proper submodules but is not simple.

Q: What replaces the theorem over a non-closed field?

(7.3b): R = End_D(V) where D = End_R(V) is a division algebra, finite-dimensional over k. The possible D are governed by the Brauer group of k; over R only R, C and H occur, over a finite field only fields occur by Wedderburn's little theorem, so every irreducible module over a finite field has Span ≅ M_m(F_q^d).

Q: Why is this the right hypothesis for the Trace Lemma?

The Trace Lemma needs to pick n^2 elements *of the group or semigroup itself* forming a basis of M_n(k), so that the coordinates tr( g_i) are again traces of elements of G. That is only possible if G spans the full matrix algebra, which is precisely what Burnside's theorem delivers from irreducibility over k.

Q: Does irreducibility of a single matrix mean anything?

A single matrix generates a commutative algebra, and a commutative algebra acts irreducibly on a finite-dimensional space over k only if that space is one-dimensional — by Burnside, M_n(k) is commutative only for n = 1. So irreducibility is intrinsically a statement about a non-commuting family; the clock and shift pair is the minimal interesting example.

Executive Summary

Burnside's theorem. Let $k$ be algebraically closed, $V$ a nonzero finite-dimensional $k$ -vector space, and $R \subseteq {End}_{k} V$ a $k$ -subalgebra acting irreducibly. Then $R = {End}_{k} V ≅ M_{n} (k)$ , $n = \dim_{k} V$ .

There is no room between irreducibility and everything. The statement is a rigidity theorem: the only way to be a proper subalgebra is to preserve a subspace. It fails over $ℝ$ — the rotation algebra $ℂ \subseteq M_{2} (ℝ)$ is the standard witness — and it fails in infinite dimension, where irreducibility buys only density.

$n^{2}$

\dim_{k} R

forced

$k = \bar{k}$ Essential hypothesis

$\dim V < \infty$ Second essential hypothesis

1905Burnside

Overview

Two questions sit behind this theorem. First: given a set of matrices, how big is the algebra they generate? Second: when is a simple module still simple after enlarging the field? Burnside's theorem answers both at once, and the answers are as clean as possible over an algebraically closed field.

The route is the Jacobson density theorem. A simple module $V$ over any ring $R$ makes $R$ dense in ${End}_{D} V$ , where $D = {End}_{R} (V)$ is a division ring by Schur's lemma. Two extra hypotheses collapse density into equality: algebraic closure forces $D = k$ , and finite dimension turns dense into onto.

For §9 the theorem is used in one specific way: it converts irreducible into *spans all of $M_{n} (k)$ *, which is exactly the hypothesis the Trace Lemma needs. Everything Burnside proved about linear groups of bounded exponent runs through this conversion.

Learning Objectives

State Burnside's theorem $(7.3)$ with the hypotheses $k = \bar{k}$ , $\dim_{k} V < \infty$ , $V \neq 0$ .
Prove it from Schur's lemma and the Jacobson density theorem.
Prove that $D = {End}_{R} (V) = k$ when $k$ is algebraically closed and $\dim_{k} V < \infty$ .
State the equivalence: $M$ absolutely irreducible $\Leftrightarrow$ $R \to {End}_{k} M$ onto $\Leftrightarrow$ $M$ simple with ${End}_{R} (M) = k$ .
Give the arbitrary-field form $R = {End}_{D} (V)$ and identify $D$ in examples.
Apply the theorem to show that a matrix semigroup acting irreducibly over $\bar{k}$ spans $M_{n} (k)$ .

Definitions

Definition(7.2)Absolute irreducibility

Let $R$ be a $k$ -algebra and $M$ a finite-dimensional $R$ -module, $M \neq 0$ . Call $M$ absolutely irreducible if $M^{K} : = M \otimes_{k} K$ is a simple $R^{K} = R \otimes_{k} K$ -module for every field extension $K \supseteq k$ . It is enough to test $K = \bar{k}$ .

Acts irreducibly: $V \neq 0$ and the only $R$ -invariant subspaces of $V$ are $0$ and $V$ ; equivalently $V$ is a simple $R$ -module.
${End}_{R} (V)$: The centraliser: all $k$ -linear $f : V \to V$ with $f (r v) = r f (v)$ . A division ring when $V$ is simple.
${Span}_{k} (G)$: For $G$ a multiplicatively closed subset of ${End}_{k} V$ , the $k$ -span of $G$ — automatically a subalgebra.
Dense subring: $R \subseteq {End}_{D} (V)$ is dense if for every $D$ -independent $v_{1}, \dots, v_{m}$ and arbitrary $w_{1}, \dots, w_{m}$ there is $r \in R$ with $r v_{i} = w_{i}$ .
Split simple module: A simple module whose centraliser is exactly $k$ ; the same thing as absolutely irreducible in finite dimension.

Subalgebras are assumed to contain the identity of End(V). For a subalgebra without identity the same statements hold provided the action is irreducible, since then RV = V.

Core Concepts

Why algebraic closure kills the centraliser

Let $V$ be a simple $R$ -module with $\dim_{k} V = n < \infty$ , and $D = {End}_{R} (V)$ . Since $R$ is a $k$ -algebra, the scalars $k \cdot {id}_{V}$ lie in $D$ ; and $D \subseteq {End}_{k} V$ , so $\dim_{k} D \leq n^{2}$ .

Take $d \in D$ . Then $k [d]$ is a commutative subring of the division ring $D$ , finite-dimensional over $k$ and without zero divisors, hence a field, hence algebraic over $k$ . If $k$ is algebraically closed this forces $k [d] = k$ , so $d$ is a scalar and $D = k$ .

From density to equality

The Jacobson density theorem says: if $V$ is a semisimple left $R$ -module and $D = {End}_{R} (V)$ , then for any $D$ -independent $v_{1}, \dots, v_{m} \in V$ and any targets $w_{1}, \dots, w_{m} \in V$ there is $r \in R$ with $r v_{i} = w_{i}$ .

When $\dim_{D} V = m < \infty$ , take $v_{1}, \dots, v_{m}$ to be a full $D$ -basis: every $D$ -linear map is then realised by some $r \in R$ , so $R \to {End}_{D} V$ is onto. Density is an approximation statement; finite dimension removes the approximation.

The trace form as a certificate

Once $R = M_{n} (k)$ , the nondegeneracy of $(σ, τ) \mapsto tr (σ τ)$ becomes available on elements of the generating set: a matrix is determined by its traces against any basis, and the basis may be chosen inside a spanning multiplicative set. This is precisely the mechanism of the Trace Lemma $(9.3)$ .

Key Results

Theorem(7.3)Burnside

Let $k$ be an algebraically closed field, $V$ a $k$ -vector space with $0 < \dim_{k} V = n < \infty$ , and $R \subseteq {End}_{k} V$ a $k$ -subalgebra acting irreducibly on $V$ . Then

R = {End}_{k} V ≅ M_{n} (k), \dim_{k} R = n^{2} .

Proof

$V$ is a simple $R$ -module, so $D : = {End}_{R} (V)$ is a division ring by Schur's lemma. As shown above, every $d \in D$ generates a finite field extension $k [d]$ of the algebraically closed field $k$ inside $D$ , so $D = k \cdot {id}_{V}$ .

By the Jacobson density theorem $R$ is dense in ${End}_{D} V = {End}_{k} V$ . Since $\dim_{D} V = \dim_{k} V = n$ is finite, choose a $k$ -basis $v_{1}, \dots, v_{n}$ of $V$ ; density supplies, for each $f \in {End}_{k} V$ , an $r \in R$ with $r v_{i} = f (v_{i})$ for all $i$ , whence $r = f$ . Therefore $R = {End}_{k} V$ .

Corollary(7.3a)Semigroup form

Let $k$ be algebraically closed and let $G \subseteq M_{n} (k)$ be closed under multiplication and act irreducibly on $k^{n}$ . Then ${Span}_{k} (G) = M_{n} (k)$ ; in particular some $n^{2}$ elements of $G$ form a $k$ -basis of $M_{n} (k)$ .

Indeed ${Span}_{k} (G)$ is a subalgebra with the same invariant subspaces as $G$ , so Burnside's theorem applies to it. This is the form used throughout §9.

Theorem(7.5)Criterion for absolute irreducibility

Let $R$ be a $k$ -algebra and $M$ a nonzero $R$ -module with $\dim_{k} M = n < \infty$ . The following are equivalent:

$M$ is absolutely irreducible;
the structure map $ρ : R \to {End}_{k} M$ is surjective;
$M$ is a simple $R$ -module and ${End}_{R} (M) = k$ .

In particular, over an algebraically closed field simple and absolutely irreducible coincide for finite-dimensional modules.

Proof

**(2) $\Rightarrow$ (1).** If $ρ$ is onto then so is $ρ \otimes K : R^{K} \to {End}_{K} (M^{K}) ≅ M_{n} (K)$ for every extension $K / k$ , and $M^{K} ≅ K^{n}$ is simple over $M_{n} (K)$ . Hence $M^{K}$ is simple over $R^{K}$ .

**(1) $\Rightarrow$ (3).** Simplicity of $M$ is the case $K = k$ . Let $D = {End}_{R} (M)$ , a division algebra with $\dim_{k} D \leq n^{2}$ . If $\dim_{k} D > 1$ then $D \otimes_{k} \bar{k}$ is a nonzero finite-dimensional $\bar{k}$ -algebra of dimension $> 1$ which is not a division ring — it contains zero divisors, being either split or non-reduced. But $D \otimes_{k} \bar{k}$ embeds in ${End}_{R^{\bar{k}}} (M^{\bar{k}})$ , which is a division ring by Schur if $M^{\bar{k}}$ is simple. Contradiction; so $D = k$ .

**(3) $\Rightarrow$ (2).** $M$ is simple with centraliser $k$ , so by the density theorem $ρ (R)$ is dense in ${End}_{k} M$ , and finite dimension upgrades this to $ρ (R) = {End}_{k} M$ . (When $k = \bar{k}$ this step is exactly Burnside's theorem.)

Corollary(7.3b)Arbitrary base field

Let $k$ be any field, $V$ finite-dimensional over $k$ and nonzero, and $R \subseteq {End}_{k} V$ a subalgebra acting irreducibly. Put $D = {End}_{R} (V)$ , a division ring finite-dimensional over $k$ . Then

R = {End}_{D} (V),

a full matrix ring over a division ring once a $D$ -basis of $V$ is chosen; and $\dim_{k} V = (\dim_{D} V) (\dim_{k} D)$ . Burnside's theorem is the case $D = k$ .

Counterexample(7.3c)Algebraic closure cannot be dropped

Take $k = ℝ$ , $V = ℝ^{2}$ , and

R = {(\begin{matrix} a & - b \\ b & a \end{matrix}) : a, b \in ℝ} ≅ ℂ \subseteq M_{2} (ℝ) .

$R$ acts irreducibly — a nonzero $ℝ$ -line is never stable under rotation by $90^{\circ}$ — yet $\dim_{ℝ} R = 2 \neq 4$ . Here $D = {End}_{R} (V) = ℂ$ and $(7.3 b)$ reads $R = {End}_{ℂ} (ℂ) = ℂ$ , which is correct. The same phenomenon with $ℍ \subseteq M_{4} (ℝ)$ gives $\dim_{ℝ} R = 4 \neq 16$ .

Remark—Infinite dimension

If $\dim_{k} V = \infty$ , irreducibility gives only density. For $V = k [x]$ and $R$ the Weyl algebra $A_{1} (k)$ in characteristic zero, $V$ is a simple faithful $A_{1}$ -module with centraliser $k$ , and $A_{1} (k)$ is dense in ${End}_{k} (V)$ but very far from equal to it — $A_{1} (k)$ is countable-dimensional, ${End}_{k} (V)$ is not.

Proof Techniques and Method

How these proofs work, and which move to reuse.

1. Read the action as a module

R \subseteq {End}_{k} V

acting irreducibly means exactly that

V

is a simple

R

-module. Everything afterwards is module theory.

2. Apply Schur

D = {End}_{R} (V)

is a division ring. This is where simplicity is spent.

3. Compute DFinite dimension makes every element of

D

algebraic over

k

; algebraic closure then forces

D = k

4. Apply density

R

is dense in

{End}_{D} V

. Finite

D

-dimension turns density into surjectivity.

5. Count

\dim_{k} R = n^{2}

is then automatic, and a basis of

M_{n} (k)

can be extracted from any spanning set.

The reusable move is step 3: finite dimension makes centralisers algebraic. It is the same move that shows a finite-dimensional division algebra over an algebraically closed field is trivial, that a finite division ring is commutative in the Wedderburn argument, and that endomorphism rings of finite length modules are semiperfect.

Worked Example

The standard representation of $S_{3}$ spans all of $M_{2}$

Let $k$ be a field and $V = {(a, b, c) \in k^{3} : a + b + c = 0}$ with $S_{3}$ permuting coordinates. Take the basis $u = e_{1} - e_{2}$ , $v = e_{2} - e_{3}$ . Writing matrices in this basis, $σ = (1 2 3)$ and $τ = (1 2)$ act by

σ = (\begin{matrix} 0 & - 1 \\ 1 & - 1 \end{matrix}), τ = (\begin{matrix} - 1 & 1 \\ 0 & 1 \end{matrix}), σ τ = (\begin{matrix} 0 & - 1 \\ - 1 & 0 \end{matrix}) .

(E.1)

Test whether ${I, σ, τ, σ τ}$ is a basis of $M_{2} (k)$ by listing coordinates in the order $(a_{11}, a_{12}, a_{21}, a_{22})$ and taking the determinant:

\det (\begin{matrix} 1 & 0 & 0 & 1 \\ 0 & - 1 & 1 & - 1 \\ - 1 & 1 & 0 & 1 \\ 0 & - 1 & - 1 & 0 \end{matrix}) = - 3 .

(E.2)

Nonzero exactly when $char k \neq 3$ .

So for $char k \neq 3$ the four group elements already span $M_{2} (k)$ : the module is absolutely irreducible, and $(7.5)$ confirms ${End}_{k S_{3}} (V) = k$ — over $ℚ$ , over $𝔽_{5}$ , over $ℂ$ alike.

What goes wrong in characteristic 3

In characteristic $3$ the determinant $(E . 2)$ vanishes, and the span drops to dimension $3$ . The reason is structural, not accidental: over $𝔽_{3}$ the vector $(1, 1, 1)$ satisfies $1 + 1 + 1 = 0$ and therefore lies in $V$ , spanning a trivial submodule. $V$ is no longer simple, Burnside's theorem no longer applies, and ${Span}_{k} (S_{3})$ is the $3$ -dimensional algebra of matrices stabilising that line — a non-semisimple algebra with one-dimensional radical.

Clock and shift: two matrices that generate everything

Let $k$ be algebraically closed with $char k \neq m$ , let $ω \in k$ be a primitive $m$ th root of unity, and set $X = diag (1, ω, \dots, ω^{m - 1})$ and $Y$ the cyclic shift $e_{i} \mapsto e_{i + 1}$ (indices mod $m$ ). Then $Y X = ω X Y$ .

A subspace invariant under $X$ is a sum of eigenspaces, i.e. spanned by coordinate vectors; $Y$ permutes the coordinate vectors in a single cycle; so no proper nonzero subspace is invariant under both. The action is irreducible, and Burnside gives ${Span}_{k} ⟨ X, Y ⟩ = M_{m} (k)$ — indeed the $m^{2}$ monomials $X^{a} Y^{b}$ form a basis.

Comparison and Classification

Irreducible subalgebras of End(V) and what Burnside gives
Base field $k$	$R$	$D = {End}_{R} (V)$	$\dim_{k} R$
$ℂ$	any irreducible subalgebra of $M_{n} (ℂ)$	$ℂ$	$n^{2}$ — everything
$ℝ$	rotations $≅ ℂ$ in $M_{2} (ℝ)$	$ℂ$	$2$
$ℝ$	$ℍ$ in $M_{4} (ℝ)$	$ℍ$	$4$
$ℚ$	$ℚ (ζ_{5})$ in $M_{4} (ℚ)$	$ℚ (ζ_{5})$	$4$
$ℚ$	image of $ℚ S_{3}$ on the standard module	$ℚ$	$4$ — everything
$𝔽_{3}$	image of $𝔽_{3} S_{3}$ on the standard module	not applicable — reducible	$3$

Which hypothesis each conclusion needs
	$k$ algebraically closed	$\dim V < \infty$	$V$ simple
${End}_{R} (V)$ is a division ring	no	no	yes
${End}_{R} (V) = k$	yes	yes	yes
$R$ dense in ${End}_{D} V$	no	no	yes
$R = {End}_{D} V$	no	yes	yes
$R = M_{n} (k)$	yes	yes	yes

Which hypothesis each conclusion needs

Relationship Map

Burnside's theorem is a specialisation of the density theorem, and in turn specialises to the split case of Wedderburn–Artin.

Jacobson density theorem — $V$ semisimple over $R$ , $D = {End}_{R} (V)$ : $R$ is dense in ${End}_{D} V$
- add $\dim_{D} V < \infty$
  - $R ↠ {End}_{D} V$ : the arbitrary-field Burnside $(7.3 b)$
  - $R / ann (V) ≅ M_{m} (D')$ for a division ring $D'$
- add $k = \bar{k}$ as well
  - $D = k$ , so $R = M_{n} (k)$ : Burnside $(7.3)$
  - simple $\Leftrightarrow$ absolutely irreducible $(7.5)$
  - the Trace Lemma $(9.3)$ becomes applicable
- take V semisimple, R artinian
  - Wedderburn–Artin structure theorem
  - uniqueness of the simple factors

V

simple over

R

D = {End}_{R} (V)

division

k = \bar{k}, \dim V < \infty

R = M_{n} (k)

traces detect everything

Applications and Industry Use

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

Linear groups

The trace method

Every finiteness theorem in §9 begins by replacing irreducible with *spans $M_{n} (k)$ *. Without Burnside's theorem the Trace Lemma has no hypothesis it can use.

Computational algebra

The MeatAxe

Parker's MeatAxe and Norton's irreducibility criterion decide whether a matrix module over a finite field is irreducible by searching for a singular element of the generated algebra with small kernel. The correctness of the criterion rests on the Burnside dichotomy: either a proper invariant subspace exists, or the algebra is everything.

Quantum information

Universality and controllability

A set of Hamiltonians or gates acting irreducibly on the state space generates the full operator algebra; operator controllability criteria in quantum control are stated exactly as irreducibility conditions, with Burnside's theorem supplying the equivalence.

Operator theory

Invariant subspaces

The Lomonosov–Rosenthal proof extends the statement to algebras of operators on Banach spaces, where it becomes a tool in the invariant subspace problem for algebras of compact-perturbation type.

Inside algebra the theorem is what makes character theory work over $ℂ$ : the matrix coefficients of the irreducible representations of a finite group span the full matrix algebra of each block, and the Wedderburn decomposition $ℂ G ≅ \prod_{i} M_{n_{i}} (ℂ)$ is exactly Burnside's theorem applied blockwise.

Computational Notes

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

Burnside's theorem converts a hard question (is this module irreducible?) into a linear algebra question (does the generated algebra have dimension $n^{2}$ ?) over an algebraically closed field. Both directions are used in practice.

Spinning up the algebra. Given generators $g_{1}, \dots, g_{s} \in M_{n} (k)$ , maintain an echelonised basis of the algebra they generate: repeatedly multiply current basis elements by generators and reduce. The basis has at most $n^{2}$ elements; a straightforward implementation costs $O (s n^{6})$ field operations, dominated by the reductions.
Spinning up a submodule. Cheaper: from a vector $v$ , close ${v}$ under the generators. Cost $O (s n^{3})$ . If the result is proper and nonzero, the module is reducible and no algebra computation is needed.
Norton's criterion / MeatAxe. Pick a pseudo-random element $a$ of the algebra, factor its characteristic polynomial, and spin up kernel vectors of $f (a)$ for irreducible factors $f$ . This finds a proper submodule with high probability, or certifies irreducibility. Implemented in GAP (SMTX package), Magma and Sage.
Over non-closed fields. Irreducibility over $k$ does not imply absolute irreducibility; the MeatAxe computes the centraliser ${End}_{R} (M)$ as well, and the module is absolutely irreducible precisely when that centraliser has dimension $1$ .

Failure Modes and Common Mistakes

Do not conclude $\dim_{k} R = n^{2}$ from irreducibility of a set of matrices without first checking closure under multiplication — the span of an arbitrary irreducible set need not be an algebra.
Do not forget that $V \neq 0$ is part of the statement; the zero module is vacuously without proper submodules and is not simple by convention.
Do not assume that the composition factors of a non-simple module can be read off from the dimension of the span; the span only reveals the flag it preserves.
Do not confuse Burnside's theorem on matrix algebras with Burnside's $p^{a} q^{b}$ theorem or with the Burnside problems on torsion groups. All three are his.

Quick Reference

Statement

k = \bar{k}

\dim_{k} V = n < \infty

R

irreducible

\Rightarrow

R = M_{n} (k)

IngredientsSchur's lemma

+

Jacobson density

+

finite dimension

General field

R = {End}_{D} V

with

D = {End}_{R} (V)

Absolute irreducibility

ρ : R \to {End}_{k} M

onto

\Leftrightarrow

M

simple and

{End}_{R} (M) = k

Semigroup form

G

multiplicatively closed and irreducible over

\bar{k}

\Rightarrow

{Span}_{k} G = M_{n} (k)

Failure over $ℝ$

ℂ \subseteq M_{2} (ℝ)

: irreducible,

\dim = 2

Failure in $\dim \infty$

A_{1} (k)

k [x]

: dense, not onto

Used forTrace Lemma

(9.3)

, Wedderburn blocks, MeatAxe

Reference numbers
Reference	Statement
$(7.3)$	Burnside: irreducible subalgebra over $\bar{k}$ is ${End}_{k} V$
$(7.3 a)$	Semigroup form: irreducible multiplicative set spans $M_{n} (k)$
$(7.3 b)$	Arbitrary field: $R = {End}_{D} V$
$(7.5)$	Absolute irreducibility $\Leftrightarrow$ surjective structure map
$(9.3)$	Trace Lemma — the main consumer

Frequently Asked Questions

Is Burnside's theorem just the density theorem?

It is the density theorem plus two hypotheses that each remove one gap. Density holds for any simple module over any ring and says $R$ approximates ${End}_{D} V$ on finite sets. Finite $D$ -dimension makes approximation exact; algebraic closure identifies $D$ with $k$ . Neither is automatic, and each has its own counterexample — the Weyl algebra on $k [x]$ , and $ℂ$ inside $M_{2} (ℝ)$ .

How do I check absolute irreducibility in practice?

Compute the centraliser. Solve the linear system ${f : f g_{i} = g_{i} f for all generators g_{i}}$ ; its solution space always contains the scalars, and the module is absolutely irreducible exactly when the space is one-dimensional. This is an $n^{2} \times n^{2}$ nullspace computation and is what the MeatAxe implementations report.

Does the theorem hold for subalgebras without an identity element?

Yes, provided irreducibility is taken in the strong sense that $V$ is a simple module, in particular $R V \neq 0$ . Then $R V$ is a nonzero submodule, so $R V = V$ , and the density argument runs unchanged. The pathological case excluded is $R$ acting as zero on a one-dimensional space, which has no proper submodules but is not simple.

What replaces the theorem over a non-closed field?

$(7.3 b)$ : $R = {End}_{D} (V)$ where $D = {End}_{R} (V)$ is a division algebra, finite-dimensional over $k$ . The possible $D$ are governed by the Brauer group of $k$ ; over $ℝ$ only $ℝ$ , $ℂ$ and $ℍ$ occur, over a finite field only fields occur by Wedderburn's little theorem, so every irreducible module over a finite field has $Span ≅ M_{m} (𝔽_{q^{d}})$ .

Why is this the right hypothesis for the Trace Lemma?

The Trace Lemma needs to pick $n^{2}$ elements of the group or semigroup itself forming a basis of $M_{n} (k)$ , so that the coordinates $tr (σ g_{i})$ are again traces of elements of $G$ . That is only possible if $G$ spans the full matrix algebra, which is precisely what Burnside's theorem delivers from irreducibility over $\bar{k}$ .

Does irreducibility of a single matrix mean anything?

A single matrix generates a commutative algebra, and a commutative algebra acts irreducibly on a finite-dimensional space over $\bar{k}$ only if that space is one-dimensional — by Burnside, $M_{n} (k)$ is commutative only for $n = 1$ . So irreducibility is intrinsically a statement about a non-commuting family; the clock and shift pair is the minimal interesting example.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §7 and §9 (pp. 100–110, 149–162).
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapter II.
W. Burnside, “On the condition of reducibility of any group of linear substitutions”, Proceedings of the London Mathematical Society (2) 3 (1905).
V. Lomonosov and P. Rosenthal, “The simplest proof of Burnside's theorem on matrix algebras”, Linear Algebra and its Applications 383 (2004), 45–47.
C. W. Curtis and I. Reiner, Representation Theory of Finite Groups and Associative Algebras, Wiley-Interscience, 1962, §27.
D. F. Holt, B. Eick and E. A. O'Brien, Handbook of Computational Group Theory, Chapman and Hall/CRC, 2005, Chapter 7.

AI Suggested Questions

Write out the Lomonosov-Rosenthal proof of Burnside's theorem and compare its hypotheses with the density-based proof.
How does the Brauer group of a field constrain which division algebras can arise as centralisers of irreducible modules?
Give the precise correctness argument for Norton's irreducibility criterion in the MeatAxe.
What is the analogue of Burnside's theorem for Lie algebras of operators, and how does Lie's theorem relate to it?
For which infinite-dimensional Banach space operator algebras does an irreducible algebra have to be dense in the strong operator topology?
Show that the m squared monomials in the clock and shift matrices form a basis of the matrix algebra, and identify the resulting algebra structure.
How is Burnside's theorem used to prove that the number of irreducible complex representations of a finite group equals the number of conjugacy classes?

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Why algebraic closure kills the centraliser

From density to equality

The trace form as a certificate

06Key Results

07Proof Techniques and Method

08Worked Example

The standard representation of S3 spans all of M2

What goes wrong in characteristic 3

Clock and shift: two matrices that generate everything

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

The trace method

The MeatAxe

Universality and controllability

Invariant subspaces

12Computational Notes

13Failure Modes and Common Mistakes

14Quick Reference

15Frequently Asked Questions

16Related KEVOS Topics

17References

18AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Key Results

Proof Techniques and Method

Worked Example

The standard representation of $S_{3}$ spans all of $M_{2}$

Comparison and Classification

Relationship Map

Applications and Industry Use

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions