Is [R,R] ever an ideal?

Yes for commutative rings, where it is zero, and in other special cases, but not in general. For M_n(k) with n ≥ 2 it is the trace-zero subspace, which is a Lie ideal but not an associative one.

The Commutator Subspace and Counting Simple Modules in Characteristic p

Q: Why is T(R) defined with the radical in it?

Because [R,R] alone cannot see the difference between R and R/rad R, while the number of simple modules depends only on the quotient. Adding rad R makes T(R) the exact preimage of [R,R], so the codimension computes an invariant of the semisimple quotient.

Q: Does the counting formula hold in characteristic p?

Yes. (7.17) makes no assumption on the characteristic; it needs finite dimensionality and splitting. Characteristic p enters only through (7.15), which is used to convert the formula into the group-theoretic count of p-regular classes.

Q: What goes wrong if k does not split R?

The simple components of R are M_n_i(D_i) with D_i ≠ k, and M_n_i(D_i)/[M_n_i(D_i), M_n_i(D_i)] can have dimension larger than 1 — for a commutative D_i it has dimension _k D_i. The codimension therefore overcounts, which is why (7.18) is an inequality.

Q: Why does (7.15) prove Fermat's Little Theorem?

Counting the words of length p in n symbols by cyclic orbits shows there are n fixed points and (n^p - n)/p orbits of size p, so p divides n^p - n. The ring-theoretic lemma and the number-theoretic one are the same orbit count.

Q: Can one recover the individual dimensions n_i from T(R)?

No. The codimension counts the simple components but discards their sizes. Recovering the n_i requires either the dimension identity (7.2) together with extra information, or an actual decomposition of the algebra.

Executive Summary

Counting simple modules normally requires finding them. For a finite-dimensional algebra that splits over its ground field, it does not: the number is the codimension of the subspace $T (R) = rad R + [R, R]$ , where $[R, R]$ is the span of all additive commutators $a b - b a$ .

The mechanism is a single computation — $[M_{n} (k), M_{n} (k)]$ is the space of trace-zero matrices, of codimension one — propagated across the simple components of $R / rad R$ . Over a non-splitting field the equality becomes an inequality, which is still enough to bound the number of simple modules from both sides.

$\dim_{k} R / T (R)$ Number of simple modules (split case)

$n^{2} - 1$

\dim_{k} [M_{n} (k), M_{n} (k)]

$r \leq r' \leq \dim_{k} R / T (R)$ General bounds

$p$ Where the power lemma bites

Overview

For a ring $R$ and $a, b \in R$ write $[a, b] = a b - b a$ , and let $[R, R]$ denote the additive subgroup generated by all such elements. If $R$ is an algebra over a commutative ring $k$ then $[R, R]$ is a $k$ -submodule, because $λ [a, b] = [λ a, b]$ . It is almost never a one-sided ideal — the trace-zero matrices are not closed under multiplication by arbitrary matrices.

That observation explains the shape of the main theorem. Over a splitting field the characters of the simple modules are the trace functions, one per simple module; adding the radical to the picture kills the functions that cannot distinguish anything. What is left has dimension exactly $r$ .

The characteristic $p$ lemma $(7.15)$ is not needed for the counting theorem itself. It is the tool that makes the same circle of ideas work for modular group algebras, where the relevant subspace is described by $p$ -th power conditions and the count comes out as the number of $p$ -regular conjugacy classes.

Learning Objectives

Define $[R, R]$ and show it is a $k$ -subspace but usually not an ideal.
Prove that $(a_{1} + \dots + a_{n})^{p} \equiv a_{1}^{p} + \dots + a_{n}^{p}$ modulo $[R, R]$ in characteristic $p$ .
Prove $[M_{n} (k), M_{n} (k)]$ is the trace-zero subspace, for $k$ any commutative ring.
Prove that $\dim_{k} R / T (R)$ counts the simple modules when $R$ splits over $k$ .
Show $T (R)$ contains every nilpotent element of a split algebra.
Establish the inequalities $r \leq r' \leq \dim_{k} R / T (R)$ for an arbitrary ground field.

Definitions

DefinitionCommutator subspace

For a ring $R$ and $a, b \in R$ , the additive commutator (or Lie product) is $[a, b] = a b - b a$ . The commutator subspace $[R, R]$ is the additive subgroup of $R$ generated by all $[a, b]$ . If $R$ is a $k$ -algebra, $[R, R]$ is a $k$ -subspace; in general it is neither a left nor a right ideal.

For a finite-dimensional $k$ -algebra put

T (R) = rad R + [R, R]

(7.17a)

A $k$ -subspace of $R$ , again not an ideal in general.

$[R, R]$: Spanned over $k$ by $[b_{i}, b_{j}]$ for any $k$ -basis ${b_{i}}$ of $R$ , since the bracket is $k$ -bilinear.
$tr$: The ordinary matrix trace on $M_{n} (k)$ ; it satisfies $tr (A B) = tr (B A)$ and so kills all commutators.
$T (R)$: $rad R + [R, R]$ . Contains all nilpotent elements when $R$ splits over $k$ .
$r$ , $r'$: The number of simple left modules over $R$ and over $R^{K}$ for a splitting field $K \supseteq k$ .
Trace function: A $k$ -linear map $f : R \to k$ with $f (a b) = f (b a)$ ; equivalently, a linear functional vanishing on $[R, R]$ .

The characteristic of $k$ is arbitrary except in $(7.15)$ , which is a statement about rings of prime characteristic and is used later for modular group algebras.

Core Concepts

Commutators, traces and characters

The dual space of $R / [R, R]$ is the space of trace functions on $R$ . Characters of modules are trace functions, so the dimension of $R / [R, R]$ bounds the number of linearly independent characters. Adding $rad R$ to the subspace discards the functionals that vanish on all semisimple subquotients — precisely those that carry no representation-theoretic information.

[R, R]

trace functions on

R

characters of modulescount of simple modules

Why $p$ -th powers behave additively

In characteristic $p$ the expansion of $(a + b)^{p}$ contains, besides $a^{p}$ and $b^{p}$ , the mixed words of length $p$ . Cyclic rotation permutes these words in orbits of size $p$ , and two words in the same orbit differ by a commutator, so each orbit contributes $p$ congruent terms — zero in characteristic $p$ . The Freshman's Dream survives noncommutativity, modulo $[R, R]$ .

Key Results

Lemma(7.15)

p

-th powers modulo commutators

Let $R$ be a ring of prime characteristic $p$ , that is $p \cdot 1_{R} = 0$ , and put $S = [R, R]$ . Then for all $a_{1}, \dots, a_{n} \in R$ and all integers $r \geq 0$ :

$(a_{1} + \dots + a_{n})^{p^{r}} \equiv a_{1}^{p^{r}} + \dots + a_{n}^{p^{r}} (\mod S)$ ;
if $s \in S$ then $s^{p^{r}} \in S$ .

Proof

It suffices to prove both statements for $r = 1$ ; the general case follows by induction. Indeed, granted the case $r = 1$ , if (1) and (2) hold for $r$ then $(\sum_{i} a_{i})^{p^{r + 1}} = ((\sum_{i} a_{i})^{p^{r}})^{p}$ , and $(\sum_{i} a_{i})^{p^{r}} = \sum_{i} a_{i}^{p^{r}} + s$ with $s \in S$ ; applying the case $r = 1$ to the $n + 1$ elements $a_{1}^{p^{r}}, \dots, a_{n}^{p^{r}}, s$ gives $\sum_{i} a_{i}^{p^{r + 1}} + s^{p} (\mod S)$ , and $s^{p} \in S$ . Similarly $s^{p^{r + 1}} = (s^{p})^{p^{r}} \in S$ .

**(1) for $r = 1$ .** Regard $a_{1}, \dots, a_{n}$ as noncommuting symbols. Expanding $(a_{1} + \dots + a_{n})^{p}$ gives the sum of all $n^{p}$ words of length $p$ in the $a_{i}$ . Let the cyclic group of order $p$ act on these words by cyclic rotation. If a word factors as $u v$ , then its rotation $v u$ satisfies $u v - v u = [u, v] \in S$ , so all words in one orbit are congruent modulo $S$ .

Because $p$ is prime, every orbit has size $1$ or $p$ . The singleton orbits are exactly the constant words $a_{1}^{p}, \dots, a_{n}^{p}$ . Each orbit of size $p$ contributes $p$ mutually congruent words, whose sum is congruent to $p \cdot w = 0$ modulo $S$ . Summing over orbits gives $(a_{1} + \dots + a_{n})^{p} \equiv a_{1}^{p} + \dots + a_{n}^{p} (\mod S)$ .

**(2) for $r = 1$ .** Write $s = \sum_{j} (a_{j} b_{j} - b_{j} a_{j})$ . By part (1), $s^{p} \equiv \sum_{j} (a_{j} b_{j} - b_{j} a_{j})^{p} (\mod S)$ . Applying (1) again to the two elements $a_{j} b_{j}$ and $- b_{j} a_{j}$ , and using $(- 1)^{p} = - 1$ (which also holds when $p = 2$ ), we get $(a_{j} b_{j} - b_{j} a_{j})^{p} \equiv (a_{j} b_{j})^{p} - (b_{j} a_{j})^{p} (\mod S)$ .

Finally set $x = a_{j}$ and $y = (b_{j} a_{j})^{p - 1} b_{j}$ . Then $x y = a_{j} (b_{j} a_{j})^{p - 1} b_{j} = (a_{j} b_{j})^{p}$ and $y x = (b_{j} a_{j})^{p - 1} b_{j} a_{j} = (b_{j} a_{j})^{p}$ , so $(a_{j} b_{j})^{p} - (b_{j} a_{j})^{p} = [x, y] \in S$ . Hence $s^{p} \in S$ .

Lemma(7.16)Commutators in a matrix ring

Let $k$ be a commutative ring and $R = M_{n} (k)$ . Then $[R, R] = {A \in M_{n} (k) : tr (A) = 0}$ . In particular, if $k$ is a field then $\dim_{k} R / [R, R] = 1$ .

Proof

Inclusion. $tr (A B) = tr (B A)$ gives $tr [A, B] = 0$ , and the trace-zero matrices form an additive subgroup, so $[R, R]$ is contained in it.

Reverse inclusion. Let $S = [R, R]$ and let $E_{i j}$ be the matrix units. For $i \neq j$ , $E_{i j} = E_{i i} E_{i j} - E_{i j} E_{i i} \in S$ , and $E_{i i} - E_{j j} = E_{i j} E_{j i} - E_{j i} E_{i j} \in S$ . Since $S$ is a $k$ -submodule, for any $A = (a_{i j})$ we may discard the off-diagonal terms and then collapse the diagonal:

A = \sum_{i, j} a_{i j} E_{i j} \equiv \sum_{i} a_{i i} E_{i i} \equiv (\sum_{i} a_{i i}) E_{11} = tr (A) E_{11} (\mod S)

So $tr (A) = 0$ forces $A \in S$ .

Theorem(7.17)Counting simple modules over a splitting field

Let $R$ be a $k$ -algebra with $\dim_{k} R < \infty$ which splits over $k$ , and set $T (R) = rad R + [R, R]$ . Then the number of isomorphism classes of simple left $R$ -modules equals $\dim_{k} R / T (R)$ . Moreover $T (R)$ contains every nilpotent element of $R$ .

Proof

Let $\bar{R} = R / rad R$ and let $π : R \to \bar{R}$ be the projection. Since $π$ is a surjective ring homomorphism, $π ([R, R]) = [\bar{R}, \bar{R}]$ , and since $rad R \subseteq T (R)$ we get $T (R) = π^{- 1} ([\bar{R}, \bar{R}])$ and therefore

\dim_{k} R / T (R) = \dim_{k} \bar{R} / [\bar{R}, \bar{R}]

Because $R$ splits over $k$ , the matrix criterion $(7.7)$ gives $\bar{R} ≅ A_{1} \times \dots \times A_{r}$ with each $A_{i} = M_{n_{i}} (k)$ and $r$ the number of simple left $R$ -modules. Commutators in a finite direct product are computed componentwise, and each $[A_{i}, A_{i}]$ is realised by elements supported in the $i$ -th factor, so $[\bar{R}, \bar{R}] = \prod_{i} [A_{i}, A_{i}]$ and

\bar{R} / [\bar{R}, \bar{R}] ≅ \prod_{i = 1}^{r} A_{i} / [A_{i}, A_{i}]

By $(7.16)$ each factor has $k$ -dimension $1$ , so the product has dimension $r$ . This proves the counting formula.

For the last claim, let $x \in R$ be nilpotent. Then $π (x)$ is nilpotent in $\bar{R}$ , so each of its components is a nilpotent matrix over $k$ and hence has trace zero; by $(7.16)$ each component lies in $[A_{i}, A_{i}]$ . Therefore $π (x) \in [\bar{R}, \bar{R}]$ , i.e. $x \in T (R)$ .

Corollary(7.18)Bounds over an arbitrary field

Let $R$ be a $k$ -algebra with $\dim_{k} R < \infty$ and let $K \supseteq k$ be a splitting field for $R$ . Write $r$ for the number of simple left $R$ -modules and $r'$ for the number of simple left $R^{K}$ -modules. Then

r \leq r' = \dim_{K} R^{K} / T (R^{K}) \leq \dim_{k} R / T (R)

(7.18)

Proof

The first inequality is the corollary of $(7.13) (2)$ : distinct simple $R$ -modules give disjoint nonempty families of composition factors over $K$ .

The middle equality is $(7.17)$ applied to the algebra $R^{K}$ over the field $K$ , which is legitimate because $K$ splits $R$ .

For the last inequality, $[R, R]^{K} \subseteq [R^{K}, R^{K}]$ and $(rad R)^{K} \subseteq rad (R^{K})$ , so $T (R)^{K} \subseteq T (R^{K})$ . Hence $\dim_{K} R^{K} / T (R^{K}) \leq \dim_{K} R^{K} / T (R)^{K} = \dim_{k} R / T (R)$ , the final equality because extension of scalars is exact and preserves dimension.

CounterexampleSplitting is needed in

(7.17)

Take $k = ℝ$ and $R = ℂ$ , viewed as a two-dimensional $ℝ$ -algebra. It is commutative, so $[R, R] = 0$ , and semisimple, so $rad R = 0$ ; hence $T (R) = 0$ and $\dim_{ℝ} R / T (R) = 2$ . But $R$ has only one simple module. The formula fails, exactly as it must: $ℝ$ does not split $ℂ$ . The inequality of $(7.18)$ survives: $r = r' = 1 \leq 2$ .

Proof Techniques and Method

How these proofs work, and which move to reuse.

Four techniques, each transferable.

Move 1

Group actions on words

When a noncommutative expansion has too many terms, let a finite group permute them. Orbits of size divisible by the characteristic vanish; fixed points are the answer. The same argument proves Fermat's Little Theorem.

Move 2

Matrix units settle everything

Any identity in $M_{n} (k)$ that is $k$ -linear can be checked on the $E_{i j}$ . Two products of matrix units generate all off-diagonal matrices and all differences of diagonal idempotents — that is the whole of $(7.16)$ .

Move 3

Push the question to $\bar{R}$

Commutator subspaces map onto commutator subspaces under surjections, so a statement about $R / T (R)$ becomes a statement about the semisimple quotient, where Wedderburn applies.

Move 4

Turn equality into an inequality

When a hypothesis fails, look for the containment that still holds. $T (R)^{K} \subseteq T (R^{K})$ converts the exact count over a splitting field into a usable upper bound over any field.

Move 1 is the one to remember. It is the standard device for proving that Frobenius-type maps are additive in noncommutative settings, and it is what makes the counting theory work for modular group algebras.

Worked Example

$R = 𝔽_{2} S_{3}$ : counting without finding the modules

Let $k = 𝔽_{2}$ and $R = k S_{3}$ , of dimension $6$ . Since $char k = 2$ divides $| S_{3} | = 6$ , Maschke does not apply and $R$ is not semisimple.

Write $S_{3} = C_{3} ⋊ C_{2}$ . As $2 ∤ 3$ , $k C_{3} ≅ 𝔽_{2} \times 𝔽_{4}$ , and the generator of $C_{2}$ acts on $C_{3}$ by inversion, hence trivially on the factor $𝔽_{2}$ and by the Frobenius on $𝔽_{4}$ . This gives

𝔽_{2} S_{3} ≅ 𝔽_{2} [C_{2}] \times (𝔽_{4} ⋊ Gal (𝔽_{4} / 𝔽_{2})) ≅ 𝔽_{2} [x] / (x^{2}) \times M_{2} (𝔽_{2})

Now compute the pieces. The first factor is commutative, so contributes nothing to $[R, R]$ and contributes its radical $(x)$ , of dimension $1$ . The second factor is $M_{2} (𝔽_{2})$ : semisimple, with $[M_{2}, M_{2}]$ the trace-zero matrices, of dimension $3$ .

Dimension audit for $𝔽_{2} S_{3}$
Quantity	First factor	Second factor	Total
$\dim_{k}$	$2$	$4$	$6$
$\dim_{k} rad$	$1$	$0$	$1$
$\dim_{k} [\cdot, \cdot]$	$0$	$3$	$3$
$\dim_{k} T$	$1$	$3$	$4$
Codimension of $T$	$1$	$1$	$2$

So $\dim_{k} R / T (R) = 6 - 4 = 2$ . Since $R / rad R ≅ 𝔽_{2} \times M_{2} (𝔽_{2})$ is a product of matrix algebras over $𝔽_{2}$ , the field $𝔽_{2}$ splits $R$ and $(7.17)$ applies: $R$ has exactly $2$ simple modules. They are the trivial module and a two-dimensional module, and the dimension test $(7.8)$ confirms $6 = 1 + (1^{2} + 2^{2})$ .

The answer agrees with Brauer's count: $S_{3}$ has two $2$ -regular conjugacy classes, namely ${1}$ and the three-cycles.

A one-line case

For $R = M_{n} (k)$ we get $rad R = 0$ , $[R, R]$ of dimension $n^{2} - 1$ , so $\dim_{k} R / T (R) = 1$ — the unique simple module $k^{n}$ . For $R = k [x] / (x^{n})$ the algebra is commutative, $[R, R] = 0$ , $rad R = (x)$ of dimension $n - 1$ , and again the codimension is $1$ .

Process and Workflow

Pick a basisTake a

k

-basis

b_{1}, \dots, b_{N}

R

; the products

[b_{i}, b_{j}]

span

[R, R]

by bilinearity.

Row reduceAssemble the

(\binom{N}{2})

vectors

[b_{i}, b_{j}]

into a matrix and compute its rank; that is

\dim_{k} [R, R]

Add the radicalAdjoin a basis of

rad R

and recompute the rank to obtain

\dim_{k} T (R)

Check splittingVerify

R / rad R

is a product of matrix algebras over

k

, or use the dimension test

(7.8)

Read off the countIf

R

splits, the number of simple modules is

N - \dim_{k} T (R)

. If not, that number is only an upper bound.

Does $k$ split $R$ ?

Yes

\dim_{k} R / T (R)

is exactly the number of simple left

R

-modules, and

T (R)

contains all nilpotent elements.

No, but a splitting field $K$ is knownCompute over

K

r' = \dim_{K} R^{K} / T (R^{K})

, and

r \leq r' \leq \dim_{k} R / T (R)

No, and no splitting field is available

\dim_{k} R / T (R)

is still an upper bound for the number of simple modules over any extension that splits

R

Comparison and Classification

The count in practice
$R$ over $k$	$\dim_{k} R$	$\dim_{k} [R, R]$	$\dim_{k} rad R$	$\dim_{k} R / T (R)$	$r$	Splits?
$M_{n} (k)$	$n^{2}$	$n^{2} - 1$	$0$	$1$	$1$	yes
$k [x] / (x^{n})$	$n$	$0$	$n - 1$	$1$	$1$	yes
$T_{2} (k)$	$3$	$1$	$1$	$2$	$2$	yes
$𝔽_{2} S_{3}$	$6$	$3$	$1$	$2$	$2$	yes
$ℂ$ over $ℝ$	$2$	$0$	$0$	$2$	$1$	no
$ℍ$ over $ℝ$	$4$	$3$	$0$	$1$	$1$	no

Which hypotheses each conclusion needs
	$\dim_{k} R < \infty$	$R$ splits over $k$	$char k = p$
$[R, R]$ is a $k$ -subspace	no	no	no
$[M_{n} (k), M_{n} (k)] =$ trace zero	no	no	no
$\dim_{k} R / T (R)$ counts simples	yes	yes	no
$T (R)$ contains all nilpotents	yes	yes	no
$p$ -th powers additive mod $[R, R]$	no	no	yes
$r \leq r' \leq \dim_{k} R / T (R)$	yes	no	no

Which hypotheses each conclusion needs

Relationship Map

The section is a chain of reductions, ending in a statement that can be evaluated by row reduction.

$(7.16)$ : matrix commutators
- feeds $(7.17)$ : the count for split algebras
  - $(7.18)$ : bounds over an arbitrary field
  - the modular counting theorem for $k G$ in characteristic $p$
  - linear independence of characters over a splitting field
- combines with $(7.15)$
  - the $p$ -power description of the relevant subspace
  - Brauer's theorem: simple $k G$ -modules correspond to $p$ -regular classes

[R, R]

T (R) = rad R + [R, R]

\dim_{k} R / T (R)

number of simple modules

Applications and Industry Use

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

Modular representation theory

Counting irreducibles

For $k G$ with $char k = p$ and $k$ a splitting field, the number of simple modules equals the number of $p$ -regular conjugacy classes. The commutator subspace is how that theorem is proved.

Computational algebra

Counting before constructing

Computing $\dim_{k} R / T (R)$ needs only linear algebra on structure constants, whereas constructing the simple modules needs a MeatAxe run. The count is used as a stopping criterion: once that many non-isomorphic simples have been found, the search terminates.

Lie theory

The trace form and $𝔰_{𝔩 n}$

$[M_{n} (k), M_{n} (k)] = 𝔰_{𝔩 n} (k)$ identifies the derived subalgebra of the general linear Lie algebra; the codimension-one statement is the reason $𝔤_{𝔩 n} = 𝔰_{𝔩 n} \oplus k \cdot 1$ when $p ∤ n$ .

Coding and symmetry

Independent invariants

Trace functions on an algebra of symmetries are exactly the functionals killing $[R, R]$ ; their number bounds how many independent numerical invariants a symmetric object can have.

Honestly stated, this is internal machinery: its purpose is to make representation-theoretic counting arguments effective, and its consumers are the algorithms and theorems built on top.

Computational Notes

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

With $N = \dim_{k} R$ , computing $[R, R]$ costs $O (N^{2})$ structure-constant multiplications followed by a rank computation on an $O (N^{2}) \times N$ matrix — dominated by the $O (N^{4})$ elimination, and usually far cheaper because the brackets are sparse.
In characteristic $p$ the subspace $T' (R)$ of elements with $a^{p^{n}} \in [R, R]$ is computed by repeated Frobenius: form the induced semilinear map on $R / [R, R]$ and take the kernel of a sufficiently high power.
The count is a cheap consistency check on any Wedderburn decomposition: the number of simple components returned must equal $\dim_{k} R / T (R)$ when the algebra splits.
None of this requires the simple modules themselves, which is the practical point — module construction over large finite fields is far more expensive than a rank computation.

Failure Modes and Common Mistakes

Do not confuse $[R, R]$ with the ideal it generates; the latter is usually all of $R$ for a noncommutative simple algebra.
Do not assume $tr$ is the reduced trace when $D_{i} \neq k$ ; $(7.16)$ is about the ordinary matrix trace over a commutative base.
Do not expect $\dim_{k} R / T (R)$ to be stable under field extension — it can drop, which is exactly the content of the inequality in $(7.18)$ .

Quick Reference

Commutator

[a, b] = a b - b a

Subspace

[R, R]

, spanned by

[b_{i}, b_{j}]

on any basis

$T (R)$

rad R + [R, R]

, a

k

-subspace

Matrix case

[M_{n} (k), M_{n} (k)] = {tr = 0}

, codimension

1

Split countnumber of simple modules

= \dim_{k} R / T (R)

Nilpotentsall lie in

T (R)

when

R

splits

General bounds

r \leq r' \leq \dim_{k} R / T (R)

Characteristic $p$

(\sum a_{i})^{p^{r}} \equiv \sum a_{i}^{p^{r}}

and

S^{p^{r}} \subseteq S

Trace functionsthe dual of

R / [R, R]

Statement finder
Result	Content	Reference
Power lemma	$p$ -th powers additive modulo $[R, R]$	(7.15)
Matrix commutators	trace-zero matrices	(7.16)
Counting theorem	$\dim_{k} R / T (R)$ counts simples for split $R$	(7.17)
Nilpotents in $T (R)$	consequence of trace zero	(7.17)
Bounds	$r \leq r' \leq \dim_{k} R / T (R)$	(7.18)
Failure without splitting	$ℂ$ over $ℝ$	counterexample

Frequently Asked Questions

Why is

T (R)

defined with the radical in it?

Because $[R, R]$ alone cannot see the difference between $R$ and $R / rad R$ , while the number of simple modules depends only on the quotient. Adding $rad R$ makes $T (R)$ the exact preimage of $[\bar{R}, \bar{R}]$ , so the codimension computes an invariant of the semisimple quotient.

[R, R]

ever an ideal?

Yes for commutative rings, where it is zero, and in other special cases, but not in general. For $M_{n} (k)$ with $n \geq 2$ it is the trace-zero subspace, which is a Lie ideal but not an associative one.

Does the counting formula hold in characteristic

p

Yes. $(7.17)$ makes no assumption on the characteristic; it needs finite dimensionality and splitting. Characteristic $p$ enters only through $(7.15)$ , which is used to convert the formula into the group-theoretic count of $p$ -regular classes.

What goes wrong if

k

does not split

R

The simple components of $\bar{R}$ are $M_{n_{i}} (D_{i})$ with $D_{i} \neq k$ , and $M_{n_{i}} (D_{i}) / [M_{n_{i}} (D_{i}), M_{n_{i}} (D_{i})]$ can have dimension larger than $1$ — for a commutative $D_{i}$ it has dimension $\dim_{k} D_{i}$ . The codimension therefore overcounts, which is why $(7.18)$ is an inequality.

Why does

(7.15)

prove Fermat's Little Theorem?

Counting the words of length $p$ in $n$ symbols by cyclic orbits shows there are $n$ fixed points and $(n^{p} - n) / p$ orbits of size $p$ , so $p$ divides $n^{p} - n$ . The ring-theoretic lemma and the number-theoretic one are the same orbit count.

Can one recover the individual dimensions

n_{i}

from

T (R)

No. The codimension counts the simple components but discards their sizes. Recovering the $n_{i}$ requires either the dimension identity $(7.2)$ together with extra information, or an actual decomposition of the algebra.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §7 (pp. 118–121).
C. W. Curtis and I. Reiner, Representation Theory of Finite Groups and Associative Algebras, Wiley-Interscience, 1962, §83.
W. Feit, The Representation Theory of Finite Groups, North-Holland, 1982, Chapter I, §17.
R. Brauer, “On the representation of a group of order $g$ in the field of the $g$ -th roots of unity”, American Journal of Mathematics 62 (1940), 565–584.
L. H. Rowen, Ring Theory, Volume I, Academic Press, 1988, Chapter 2.

AI Suggested Questions

Prove that the number of simple $k G$ -modules in characteristic $p$ equals the number of $p$ -regular conjugacy classes, for $k$ a splitting field.
Compute $\dim_{k} R / T (R)$ for $R = k A_{4}$ with $char k = 2$ and $k$ a splitting field.
For which $n$ and $p$ is the trace-zero subspace of $M_{n} (𝔽_{p})$ equal to its own derived subspace?
How does $\dim_{k} R / [R, R]$ relate to the zeroth Hochschild homology of the algebra?
Give an example of a finite-dimensional algebra where $\dim_{k} R / T (R)$ strictly exceeds the number of simple modules over every extension field.
Describe an efficient algorithm for computing the subspace of elements with $a^{p^{n}}$ in the commutator subspace.
What is the analogue of (7.17) for an artinian ring that is not an algebra over a field?

Commutator Subspace Methods

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Commutators, traces and characters

Why $p$ -th powers behave additively

Key Results

Proof Techniques and Method

Group actions on words

Matrix units settle everything

Push the question to $\bar{R}$

Turn equality into an inequality

Worked Example

$R = 𝔽_{2} S_{3}$ : counting without finding the modules

A one-line case

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Counting irreducibles

Counting before constructing

The trace form and $𝔰_{𝔩 n}$

Independent invariants

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

References

AI Suggested Questions

Continue learning

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Commutators, traces and characters

Why p-th powers behave additively

06Key Results

07Proof Techniques and Method

Group actions on words

Matrix units settle everything

Push the question to R¯

Turn equality into an inequality

08Worked Example

R=𝔽2S3: counting without finding the modules

A one-line case

09Process and Workflow

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

Counting irreducibles

Counting before constructing

The trace form and 𝔰𝔩n

Independent invariants

13Computational Notes

14Failure Modes and Common Mistakes

15Quick Reference

16Frequently Asked Questions

17Related KEVOS Topics

18References

19AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Why $p$ -th powers behave additively

Key Results

Proof Techniques and Method

Push the question to $\bar{R}$

Worked Example

$R = 𝔽_{2} S_{3}$ : counting without finding the modules

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

The trace form and $𝔰_{𝔩 n}$

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions