Matrix Rings over Division Rings: The Model Simple Artinian Ring

Q: Why is the column space a right D-vector space rather than a left one?

Because M_n(D) acts on the left, and the two actions must commute for V to be a bimodule: (Av)d = A(vd) holds when d multiplies on the right. If you insist on a left D-action you must move the matrices to the right, which replaces D by D^op throughout.

Q: How many minimal left ideals does M_n(D) have?

As many as there are lines in the n-dimensional left D-space D^1 × n, that is |P^n-1(D)|. For D = F_q this is (q^n-1)/(q-1); for infinite D it is infinite. They are all isomorphic to the unique simple module V.

Q: Is n the same as the dimension of the simple module?

Only over a field with D = k. In general n = _D V, the dimension over the endomorphism division ring, while _k V = n _k D. For M_2(H) over R, n = 2 but _R V = 8.

Q: Why is M_n(D) noetherian as well as artinian?

Both follow from the same n^2-dimensional bound on chains of left ideals. In general, for left artinian rings with identity, Hopkins–Levitzki shows left artinian implies left noetherian — but here no such theorem is needed.

Q: Can two non-isomorphic division rings give isomorphic matrix rings?

No. By (3.3a), D is recovered as the endomorphism ring of the unique simple module, so M_n(D) ≅ M_m(D') forces D ≅ D' and n = m. This is special to the simple artinian case: for general rings, M_2(R) ≅ M_2(S) does not force R ≅ S.

Executive Summary

Let $D$ be a division ring and $R = M_{n} (D)$ . This single family of rings carries the whole of Wedderburn–Artin theory: $R$ is simple, left and right artinian and noetherian, and left and right semisimple; it has exactly one simple left module up to isomorphism, the column space $V = D^{n}$ ; the regular module is $n$ copies of it; and $End (_{R} V) ≅ D$ .

The last statement is the important one. It says $D$ is recoverable from the ring $R$ alone, which is why the Wedderburn–Artin decomposition is unique and not merely available.

$n \cdot V$ Structure of

_{R} R

1Simple modules up to iso

$D$

End (_{R} V)

$n^{2}$

\dim_{D} R

Overview

A division ring is left and right semisimple for trivial reasons: its modules are vector spaces and every short exact sequence of vector spaces splits. The question is how to build more examples, and the answer is that matrices over a division ring, and finite products of those, exhaust the possibilities.

This page establishes the building block half of that claim. The classification half — that there is nothing else — is The Wedderburn–Artin Theorem.

R = M_{n} (D) ⟹_{R} R ≅ {\underset{⏟}{V \oplus \dots \oplus V}}_{n}, V = D^{n}, End (_{R} V) ≅ D

(3.3)

Read left to right this constructs semisimple rings; read right to left it recovers $n$ and $D$ from $R$ .

Note the asymmetry of sides that is not present here. $M_{n} (D)$ is simultaneously left and right artinian, noetherian and semisimple. The general theory has many one-sided notions; this family is where they all coincide, which is what makes it a safe model to reason with.

Learning Objectives

State $(3.3)$ in full, distinguishing the chain conditions from the module statements.
Prove the column space $V = D^{n}$ is a simple faithful left $M_{n} (D)$ -module.
Prove $_{R} R ≅ n \cdot V$ using the decomposition into column ideals.
Prove $End (_{R} V) ≅ D$ by evaluating an endomorphism at the first standard basis column.
Classify the left ideals of $M_{n} (D)$ and count the minimal ones over $𝔽_{q}$ .
Deduce that $M_{n} (D) ≅ M_{m} (D')$ forces $n = m$ and $D ≅ D'$ .

Definitions

Definition—The column space module

Let $D$ be a division ring and $n \geq 1$ . Write $V = D^{n}$ for the set of $n$ -tuple columns, regarded as a right $D$ -vector space by entrywise multiplication on the right. The ring $R = M_{n} (D)$ acts on $V$ on the left by matrix multiplication, and this action is $D$ -linear on the right, so $V$ is an $(R, D)$ -bimodule.

Identifying a matrix with the right- $D$ -linear map it induces gives a ring isomorphism $R ≅ End (V_{D})$ — the familiar matrix representation of linear maps, valid over a division ring exactly as over a field.

$C_{j}$: The $j$ -th column ideal: matrices whose columns other than the $j$ -th are zero. A left ideal of $R$ , isomorphic to $V$ .
$D^{1 \times n}$: The row space, a left $D$ -vector space of dimension $n$ . Its subspaces index the left ideals of $M_{n} (D)$ .
Faithful: $ann (V) = 0$ . For $V = D^{n}$ this is immediate: a matrix killing every column vector is zero.
Composition length: The common length of all composition series; for $_{R} R$ with $R = M_{n} (D)$ it equals $n$ .
$Z (D)$: The centre of $D$ , a field; it is also the centre of $M_{n} (D)$ up to the identification with scalar matrices.

Columns carry the right D-action and rows the left one. Swapping them transposes every statement on this page.

Core Concepts

Chain conditions come from a dimension count

Left multiplication by the scalar matrix $d \cdot 1_{n}$ makes $R$ a left $D$ -vector space with $\dim_{D} R = n^{2}$ . A left ideal is closed under left multiplication by every element of $R$ , in particular by scalar matrices, so every left ideal is a $D$ -subspace. A strictly monotone chain of subspaces of an $n^{2}$ -dimensional space has length at most $n^{2}$ , so both the descending and the ascending chain conditions hold on left ideals. The same argument on the right gives the right-handed versions.

One simple module, $n$ times over

Splitting a matrix into its columns splits $R$ as a left module:

_{R} R = C_{1} \oplus C_{2} \oplus \dots \oplus C_{n}, C_{j} ≅ V,

Left multiplication acts columnwise, so each $C_{j}$ is a left ideal, and $C_{j} \to V$ picking out the $j$ -th column is an $R$ -isomorphism.

Since $V$ is simple, this exhibits $_{R} R$ as a direct sum of simple submodules, which is the definition of left semisimple. It also pins the composition length of $_{R} R$ at exactly $n$ .

The left ideal lattice is a subspace lattice

For $A \in R$ , the rows of $B A$ are left $D$ -combinations of the rows of $A$ , so $R A$ consists of all matrices whose rows lie in the left row space of $A$ . Running this in both directions gives an inclusion-preserving bijection

{left ideals of M_{n} (D)} ⟷ {left D -subspaces of D^{1 \times n}},

A left ideal is ${M : every row of M lies in W}$ for a unique subspace $W$ .

Minimal left ideals therefore correspond to lines in $D^{1 \times n}$ , that is to the points of the projective space $ℙ^{n - 1} (D)$ . Over $𝔽_{q}$ there are $(q^{n} - 1) / (q - 1)$ of them; over an infinite $D$ there are infinitely many. All are isomorphic to $V$ — the contrast with the two two-sided ideals from Ideals of Matrix Rings and the Correspondence Theorem could not be sharper.

Recovering the division ring

$D$ sits inside $End (_{R} V)$ as right scalar multiplication, and $(3.3) (3)$ says that is everything. Combined with $R ≅ End (V_{D})$ this is a double centraliser statement: each of $R$ and $D$ is the full centraliser of the other in the endomorphism ring of the abelian group $V$ .

Key Results

Theorem(3.3)Structure of

M_{n} (D)

Let $D$ be a division ring, $n \geq 1$ , and $R = M_{n} (D)$ . Then:

$R$ is simple, left semisimple, left artinian and left noetherian (and, by the symmetric argument, right semisimple, right artinian and right noetherian).
$R$ has a unique simple left module $V$ up to isomorphism, namely the column space $D^{n}$ ; $R$ acts faithfully on $V$ ; and $_{R} R ≅ n \cdot V$ .
$End (_{R} V)$ , viewed as a ring of right operators on $V$ , is isomorphic to $D$ .

Proof

(1) Simplicity. A division ring has only the ideals $0$ and $D$ , so $M_{n} (D)$ is simple by the ideal correspondence $(3.1)$ .

(1) Chain conditions. $R$ is a left $D$ -vector space of dimension $n^{2}$ under left multiplication by scalar matrices, and every left ideal is a $D$ -subspace, so chains of left ideals have length at most $n^{2}$ . Both DCC and ACC follow.

**(2) $V$ is simple and faithful.** Let $0 \neq v \in V$ and let $w \in V$ be arbitrary. Extend $v$ to a basis of the right $D$ -space $V$ ; the right- $D$ -linear map sending $v \mapsto w$ and the other basis vectors to $0$ is given by a matrix $A$ , so $A v = w$ . Hence $R v = V$ and $V$ is simple. Faithfulness is immediate: if $A V = 0$ then $A$ kills every standard basis column, so every column of $A$ is zero.

**(2) $_{R} R ≅ n \cdot V$ .** Let $C_{j} \subseteq R$ be the set of matrices whose columns other than the $j$ -th vanish. Since the $j$ -th column of $B A$ depends only on $B$ and the $j$ -th column of $A$ , each $C_{j}$ is a left ideal and the map $C_{j} \to V$ extracting the $j$ -th column is an isomorphism of left $R$ -modules. Clearly $R = C_{1} \oplus \dots \oplus C_{n}$ as abelian groups, hence as left $R$ -modules. So $_{R} R$ is semisimple and $R$ is a left semisimple ring.

**(2) Uniqueness of $V$ .** Let $V'$ be any simple left $R$ -module. Then $V' ≅ R / 𝔪$ for some maximal left ideal $𝔪$ , so $V'$ is a quotient of $_{R} R$ and therefore a composition factor of it. By the Jordan–Hölder theorem the composition factors of $_{R} R ≅ n \cdot V$ are all isomorphic to $V$ , so $V' ≅ V$ .

**(3) $End (_{R} V) ≅ D$ .** Define $λ : D \to End (_{R} V)$ by $v \cdot λ (d) = v d$ , right scalar multiplication. This is $R$ -linear because matrix multiplication on the left commutes with scalar multiplication on the right, and it is a ring homomorphism in the right-operator convention: $v (λ (d) λ (d')) = (v d) d' = v (d d') = v λ (d d')$ . It is injective, since $λ (d) = 0$ applied to the first standard column $e_{1}$ gives $e_{1} d = 0$ , hence $d = 0$ .

For surjectivity, take $f \in End (_{R} V)$ and put $u = e_{1} f \in V$ , with coordinates $u_{1}, \dots, u_{n}$ . Given any $a \in V$ , let $A \in R$ be the matrix whose first column is $a$ and whose other columns are zero, so that $A e_{1} = a$ . Then $R$ -linearity gives

a f = (A e_{1}) f = A (e_{1} f) = A u = a u_{1},

the last step because only the first column of $A$ is nonzero, so $A u = a u_{1}$ . Thus $f = λ (u_{1})$ , and $λ$ is onto. (Taking $a = e_{1}$ also shows $u = e_{1} u_{1}$ , i.e. the remaining coordinates of $u$ vanish, as consistency demands.)

Corollary(3.3a)Recovering the data

Let $D, D'$ be division rings and $n, m \geq 1$ . If $M_{n} (D) ≅ M_{m} (D')$ as rings, then $n = m$ and $D ≅ D'$ .

Proof

Fix an isomorphism and transport modules along it. By $(3.3) (2)$ each side has a unique simple left module, so the simple modules correspond; call the common module $V$ . The integer $n$ is the composition length of the left regular module, an isomorphism invariant, and it equals $m$ by the same computation on the other side. By $(3.3) (3)$ , $D ≅ End (_{R} V) ≅ D'$ .

Corollary(3.3b)Elementary consequences

For $R = M_{n} (D)$ : $rad R = 0$ ; $R$ is prime and semiprime; $Z (R) = Z (D) \cdot 1_{n} ≅ Z (D)$ , a field; $R$ is Dedekind-finite, that is $a b = 1$ implies $b a = 1$ ; and every finitely generated left $R$ -module is isomorphic to $m \cdot V$ for a unique $m \geq 0$ , so $R$ has the invariant basis number property.

Proof

$rad R$ is a proper two-sided ideal, hence $0$ by simplicity. Primeness follows from simplicity. The centre is computed in Ideals of Matrix Rings and the Correspondence Theorem, and $Z (D)$ is a field because a commutative division ring is a field. Every left module over a semisimple ring is a direct sum of simple modules, and here there is only one simple module, so any f.g. module is $m \cdot V$ ; the integer $m$ is its composition length, hence unique. For Dedekind-finiteness, suppose $a b = 1$ . Right multiplication $ρ_{a} : x \mapsto x a$ is an endomorphism of $_{R} R$ and is injective, since $x a = 0$ gives $x = x a b = 0$ . A module of finite length admits no injective non-surjective endomorphism, so $ρ_{a}$ is bijective; since $ρ_{a} ρ_{b} = id$ , the map $ρ_{b}$ is its two-sided inverse, whence $x b a = x$ for all $x$ , and $x = 1$ gives $b a = 1$ .

Proof Techniques and Method

How these proofs work, and which moves to reuse.

Move 1

Turn a chain condition into a dimension count

If a ring is a finite-dimensional vector space over something that acts on all its one-sided ideals, DCC and ACC are free. This is the cheapest possible proof that finite-dimensional algebras are artinian.

Move 2

Slice the regular module

Decompose $_{R} R$ along an obvious geometric splitting — here, columns — and identify each piece. Semisimplicity of a ring is always proved by exhibiting such a splitting.

Move 3

Evaluate at a generator

To compute $End$ of a cyclic module, evaluate an endomorphism at a generator; $R$ -linearity then determines it everywhere. Here $e_{1}$ generates $V$ and the whole surjectivity proof is three lines.

Move 3 is the reusable one. It is the same technique that computes $End (_{R} R) ≅ R$ (evaluate at $1$ ), and it is why cyclic modules are easy and non-cyclic ones are not.

Worked Example

$M_{2} (𝔽_{2})$ counted completely

Take $D = 𝔽_{2}$ , $n = 2$ , $R = M_{2} (𝔽_{2})$ , a ring with $2^{4} = 16$ elements. The simple module is $V = 𝔽_{2}^{2}$ with four elements, and $_{R} R ≅ V \oplus V$ — consistent with $16 = 4^{2}$ .

$M_{2} (𝔽_{2})$ : everything at once
Invariant	Value	Why
Order	16	$\| D \|^{n^{2}} = 2^{4}$
Two-sided ideals	2	simple, by $(3.1)$
Left ideals	5	$0$ , three minimal, $R$ — the subspace lattice of $𝔽_{2}^{1 \times 2}$
Minimal left ideals	3	$(q^{n} - 1) / (q - 1) = 3$ points of $ℙ^{1} (𝔽_{2})$
Simple modules	1	$V = 𝔽_{2}^{2}$ , of order 4
Composition length of $_{R} R$	2	$n = 2$
$End (_{R} V)$	$𝔽_{2}$	$(3.3) (3)$
Unit group	$G L_{2} (𝔽_{2}) ≅ S_{3}$ , order 6	invertible matrices

The three minimal left ideals are the sets of matrices whose rows all lie in one of the three lines of $𝔽_{2}^{1 \times 2}$ , spanned by $(1, 0)$ , $(0, 1)$ and $(1, 1)$ . Each has four elements and each is isomorphic to $V$ ; any two distinct ones intersect in $0$ and sum to $R$ , which is why $R ≅ V \oplus V$ can be seen in several ways at once.

A noncommutative division ring: $M_{2} (ℍ)$

Let $D = ℍ$ , the real quaternions, and $R = M_{2} (ℍ)$ . As a real algebra $\dim_{ℝ} R = 4 \cdot 4 = 16$ . The simple module is $V = ℍ^{2}$ , of real dimension $8$ , with $End (_{R} V) ≅ ℍ$ ; and $_{R} R ≅ V \oplus V$ , matching $16 = 2 \cdot 8$ .

An infinite family of minimal left ideals

In $M_{2} (ℚ)$ the minimal left ideals are indexed by $ℙ^{1} (ℚ)$ , so there are infinitely many, all isomorphic to $V = ℚ^{2}$ . Semisimplicity does not mean 'few submodules'; it means every submodule is a direct summand.

Comparison and Classification

$M_{n} (D)$ against neighbouring classes of rings
Ring	Simple?	Left artinian?	Simple modules	$End$ of a simple module
$D$ a division ring	yes	yes	1	$D$
$M_{n} (D)$	yes	yes	1	$D$
$M_{n_{1}} (D_{1}) \times \dots \times M_{n_{r}} (D_{r})$	only if $r = 1$	yes	$r$	$D_{i}$
$ℤ$	no	no	one per prime	$𝔽_{p}$
Weyl algebra $A_{1} (ℂ)$	yes	no	infinitely many	$ℂ$
$End (V_{D})$ , $\dim_{D} V$ infinite	no	no	$V$ , plus those of $E / I$	$D$ for $V$
Upper triangular $T_{n} (k)$	no	yes	$n$	$k$

The row for the Weyl algebra is the warning: simple alone gives none of the structure on this page. It is simplicity plus a chain condition — equivalently, by Simple Artinian Rings and Minimal One-Sided Ideals, the existence of a minimal left ideal — that forces the matrix form.

Relationship Map

All ringsno structure assumed

Simple ringsonly

0

and

R

as ideals; includes

A_{1} (ℂ)

Simple with a minimal left idealequivalently simple artinian

$M_{n} (D)$ the content of

(3.3)

and of

(3.13)

$D$ itselfthe case

n = 1

D

division ring

M_{n} (D)

(3.3)

finite products by

(3.4)

all semisimple rings by

(3.5)

The reverse reading is the classification: The Wedderburn–Artin Theorem says the last arrow can be inverted, and Uniqueness in the Wedderburn–Artin Decomposition says the inversion is essentially unique — which relies on $(3.3 a)$ above.

Design Considerations

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

Columns or rows? Putting $R$ on the left forces $D$ to the right on $V$ . If your application naturally has row vectors, work with $M_{n} (D)^{op} ≅ M_{n} (D^{op})$ rather than silently transposing.
Coordinates or not? $M_{n} (D)$ and $End (V_{D})$ are isomorphic but not equal: the first has a chosen basis. Use $End (V_{D})$ when the statement should be basis-free — for instance in the density theorem — and matrices when you need to compute.
Which invariant is canonical? $D$ is canonical up to isomorphism (as $End$ of the simple module) but there is no canonical isomorphism; conjugation by any invertible matrix gives a different one. Never treat an identification $R ≅ M_{n} (D)$ as unique.
Field of scalars. If $R$ is an algebra over a field $k$ , then $k \subseteq Z (R) ≅ Z (D)$ , so $D$ is automatically a $k$ -algebra and $\dim_{k} R = n^{2} \dim_{k} D$ . Use that equation as an arithmetic constraint when guessing a decomposition.

Computational Notes

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

Detect simplicityFor a finite-dimensional algebra given by structure constants, compute the radical; if it is nonzero the algebra is not semisimple and

(3.3)

does not apply.

Find a minimal left idealFind a primitive idempotent

e

by splitting idempotents; then

R e

is a minimal left ideal and

V ≅ R e

Compute $D$

D ≅ End (_{R} V) ≅ e R e

for a primitive idempotent

e

. Computing

e R e

is a linear algebra problem of size

\dim_{k} R

Read off $n$

n = \dim_{k} R / \dim_{k} V

, equivalently the composition length of

_{R} R

, equivalently

\dim_{D} V

Arithmetic in $M_{n} (D)$ costs $O (n^{ω})$ operations in $D$ with fast matrix multiplication; the constant depends on how $D$ is represented, and for $D$ a field extension of degree $d$ multiply by the cost of arithmetic in that extension.
Over a finite field every finite division ring is a field, by Wedderburn's little theorem, so $M_{n} (𝔽_{q})$ is the only simple finite ring of matrix type — a substantial simplification exploited by every finite-field library.
Constructing an explicit isomorphism $R ≅ M_{n} (D)$ is harder than proving it exists; over $ℚ$ this is the explicit isomorphism problem, related to finding a zero divisor in a quaternion algebra and to factoring integers.
GAP, Magma and Sage provide WedderburnDecomposition or equivalent for group algebras and for finite-dimensional algebras, returning both the $n_{i}$ and the $D_{i}$ .

Failure Modes and Common Mistakes

$\dim_{k} M_{n} (D) = n^{2} \dim_{k} D$ , not $n \dim_{k} D$ ; getting this wrong breaks every Wedderburn dimension count.
$Z (M_{n} (D)) ≅ Z (D)$ , a field, even when $D$ is noncommutative. The centre never grows with $n$ .
For $n \geq 2$ the ring has zero divisors and nontrivial idempotents; do not carry over intuition from division rings about cancellation.
The isomorphism $End (_{R} V) ≅ D$ depends on writing endomorphisms opposite the scalars. With the other convention the answer is $D^{op}$ , which changes the statement of Wedderburn–Artin.

Quick Reference

Setting

D

a division ring,

R = M_{n} (D)

n \geq 1

Simplicity

R

is simple; only ideals

0

and

R

Chain conditionsartinian and noetherian on both sides;

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

Simple module

V = D^{n}

columns, unique up to isomorphism

Regular module

_{R} R ≅ n \cdot V

, composition length

n

Endomorphisms

End (_{R} V) ≅ D

;

End (V_{D}) ≅ R

Left idealssubspaces of

D^{1 \times n}

; minimal ones

\leftrightarrow ℙ^{n - 1} (D)

Uniqueness

M_{n} (D) ≅ M_{m} (D') \Rightarrow n = m

D ≅ D'

Centre

Z (R) ≅ Z (D)

, a field

Radical

rad R = 0

Numerical invariants for a $k$ -algebra $M_{n} (D)$
Quantity	Formula	$M_{2} (ℍ)$ over $ℝ$
$\dim_{k} R$	$n^{2} \dim_{k} D$	16
$\dim_{k} V$	$n \dim_{k} D$	8
Composition length of $_{R} R$	$n$	2
$\dim_{k} Z (R)$	$\dim_{k} Z (D)$	1
Number of simple modules	1	1

Frequently Asked Questions

Why is the column space a right D-vector space rather than a left one?

Because $M_{n} (D)$ acts on the left, and the two actions must commute for $V$ to be a bimodule: $(A v) d = A (v d)$ holds when $d$ multiplies on the right. If you insist on a left $D$ -action you must move the matrices to the right, which replaces $D$ by $D^{op}$ throughout.

How many minimal left ideals does

M_{n} (D)

have?

As many as there are lines in the $n$ -dimensional left $D$ -space $D^{1 \times n}$ , that is $| ℙ^{n - 1} (D) |$ . For $D = 𝔽_{q}$ this is $(q^{n} - 1) / (q - 1)$ ; for infinite $D$ it is infinite. They are all isomorphic to the unique simple module $V$ .

Does

(3.3)

prove that every simple artinian ring is a matrix ring?

No — it proves the converse direction, that matrix rings over division rings have all these properties. The classification statement is Lam $(3.10)$ and $(3.13)$ , proved either through the isotypic component construction or through Rieffel's double centraliser argument, and it is covered in Simple Artinian Rings and Minimal One-Sided Ideals.

n

the same as the dimension of the simple module?

Only over a field with $D = k$ . In general $n = \dim_{D} V$ , the dimension over the endomorphism division ring, while $\dim_{k} V = n \dim_{k} D$ . For $M_{2} (ℍ)$ over $ℝ$ , $n = 2$ but $\dim_{ℝ} V = 8$ .

Why is

M_{n} (D)

noetherian as well as artinian?

Both follow from the same $n^{2}$ -dimensional bound on chains of left ideals. In general, for left artinian rings with identity, Hopkins–Levitzki shows left artinian implies left noetherian — but here no such theorem is needed.

Can two non-isomorphic division rings give isomorphic matrix rings?

No. By $(3.3 a)$ , $D$ is recovered as the endomorphism ring of the unique simple module, so $M_{n} (D) ≅ M_{m} (D')$ forces $D ≅ D'$ and $n = m$ . This is special to the simple artinian case: for general rings, $M_{2} (R) ≅ M_{2} (S)$ does not force $R ≅ S$ .

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §3, (3.3) and (3.13) (pp. 33–34, 40).
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, revised edition, 1964, Chapter IV.
I. N. Herstein, Noncommutative Rings, Carus Mathematical Monographs 15, Mathematical Association of America, 1968, Chapter 1.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §13.
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988, §2.1–§2.3.

AI Suggested Questions

Prove that every finitely generated module over $M_{n} (D)$ is free of a well-defined rank over the simple module.
Describe the lattice of left ideals of $M_{3} (𝔽_{2})$ and count its elements.
How does Wedderburn's little theorem restrict the possible $D$ when $M_{n} (D)$ is finite?
Work out the double centraliser statement $R ≅ End (V_{D})$ and $D ≅ End (_{R} V)$ as an instance of Morita equivalence.
What is the explicit isomorphism problem for simple algebras over $ℚ$ , and why is it computationally hard?
Give a simple ring with no minimal left ideal and explain which part of $(3.3)$ fails for it.
Compute the automorphism group of $M_{n} (D)$ and relate it to the Skolem–Noether theorem.

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Chain conditions come from a dimension count

One simple module, n times over

The left ideal lattice is a subspace lattice

Recovering the division ring

06Key Results

07Proof Techniques and Method

Turn a chain condition into a dimension count

Slice the regular module

Evaluate at a generator

08Worked Example

M2(𝔽2) counted completely

A noncommutative division ring: M2(ℍ)

An infinite family of minimal left ideals

09Comparison and Classification

10Relationship Map

11Design Considerations

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

One simple module, $n$ times over

Key Results

Proof Techniques and Method

Worked Example

$M_{2} (𝔽_{2})$ counted completely

A noncommutative division ring: $M_{2} (ℍ)$

Comparison and Classification

Relationship Map

Design Considerations

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions