Group Rings in Prime Characteristic

Q: Why must k be reduced in (6.13), when the conclusion is about kG and not about k?

Because k sits inside kG. If a ∈ k is a nonzero nilpotent then a kG is a nonzero nil ideal of kG, so the conclusion fails before the group is even consulted. Reducedness is also used at the end of the proof, where _1^,p^n ≠ 0 is needed for every n.

Q: Does (6.13) say that rad kG = 0?

No, and the gap is the substance of the subject. It says there is no nonzero nil left ideal. The radical of a group ring need not be nil, so this does not bound it directly. The conclusion becomes J-semisimplicity only after tensoring up to a rational function field, where Amitsur's theorem says the radical is controlled by a nil ideal of the original ring.

Q: Is the p'-hypothesis really necessary for J-semisimplicity?

Not as stated. What is necessary is the weaker condition that every finite normal subgroup be a p'-group. The infinite dihedral group in characteristic 2 satisfies the weaker condition and fails the stronger one, and Wallace proved its group algebra is J-semisimple. No characterisation is known.

Q: Why is the prime field the hard case?

Every step in the proof consumes transcendence. Amitsur's rational-extension theorem needs a nonempty transcendence basis to have anything to adjoin; over F_p there is nothing to adjoin and the argument has no purchase. The same phenomenon occurs in characteristic zero, where the algebraic number fields are the unresolved case.

Q: How does this relate to Maschke's theorem?

Maschke settles finite groups completely: kG is semisimple exactly when |G| is invertible in k. For finite G in characteristic p, being a p'-group is precisely p |G|, so (6.13) recovers half of Maschke. The content of this page is everything Maschke cannot see, namely infinite G, where semisimplicity is impossible and J-semisimplicity is the right question.

Q: Where does the commutativity of k actually get used?

In the collapse of free orbits. The p tuples in a rotation orbit give the products _g_1…_g_p, _g_2…_g_p_g_1 and so on. These agree only if the coefficients commute, and only then does the orbit contribute p times a single element, hence zero.

Executive Summary

Over a field of characteristic $0$ , Rickart and then Amitsur showed that $k G$ is J-semisimple for essentially every base field. In characteristic $p$ the picture changes: an element of order $p$ manufactures a nonzero nilpotent ideal, so the hypothesis ** $G$ is a $p'$ -group** — no element of order $p$ — enters at once. Under that hypothesis Passman and Connell proved that $k G$ has no nonzero nil left ideals, and Passman's theorem converts this into J-semisimplicity for every field $K$ that is not algebraic over $𝔽_{p}$ .

What remains open is exactly the case of the prime field: nobody knows whether $𝔽_{p} G$ is J-semisimple for every $p'$ -group $G$ . As in characteristic zero, the smallest base field is the hardest.

$p'$ Required torsion condition

(6.13)No nil left ideals

1962Passman's theorem

$𝔽_{p}$ Case still open

Overview

Let $k$ be a commutative ring and $G$ a group. Maschke's theorem answers the semisimplicity question for finite $G$ : $k G$ is semisimple exactly when $| G |$ is invertible in $k$ . For infinite $G$ the group ring is never semisimple, so the sharp question becomes J-semisimplicity, $rad k G = 0$ . The characteristic-zero half of that programme is carried by Rickart's analytic theorem and Amitsur's algebraic strengthening; this page is the characteristic- $p$ half.

One obstruction is immediate. If $x \in G$ has order $p$ and $char k = p$ , then inside the commutative subring $k ⟨ x ⟩$ we have

(x - 1)^{p} = x^{p} - 1 = 0,

(6.13a)

The Frobenius identity in characteristic $p$ : an element of order $p$ produces a nilpotent in $k G$ .

so $(x - 1) k G$ is a nonzero one-sided ideal all of whose elements are nilpotent when $x$ is central, and in general a nonzero nilpotent ideal appears as soon as $G$ has a finite normal subgroup of order divisible by $p$ . Ruling out $p$ -torsion is therefore the first move, and the definition of a $p'$ -group is designed for it.

Learning Objectives

Define a $p'$ -group and explain why it is forced on us in characteristic $p$ .
Prove the trace identity $tr (β^{p}) = tr (β)^{p}$ by orbit counting.
Deduce $(6.13)$ : $k G$ has no nonzero nil left ideals for $k$ commutative reduced of characteristic $p$ .
Use $(6.14)$ to move J-semisimplicity up an algebraic extension of fields of characteristic $p$ .
State Passman's theorem $(6.15)$ with its exact hypothesis on $K / 𝔽_{p}$ .
Show that the condition on finite normal subgroups is necessary but not sufficient.

Definitions

Definition(6.13d)

p'

-group

For a prime $p$ , a group $G$ is a ** $p'$ -group** if it has no element of order $p$ . By Cauchy's theorem, a finite group is a $p'$ -group precisely when $p ∤ | G |$ . Torsion-free groups are $p'$ -groups for every $p$ simultaneously.

$tr (α)$: For $α = \sum_{g \in G} a_{g} g \in k G$ , the coefficient $a_{1}$ of the identity element. It is $k$ -linear and satisfies $tr (α β) = tr (β α)$ .
Reduced: A ring $k$ is reduced if $a^{2} = 0$ forces $a = 0$ ; equivalently $a^{n} = 0$ forces $a = 0$ . Fields, domains and products of domains are reduced.
Nil left ideal: A left ideal every element of which is nilpotent. Every nil one-sided ideal lies inside $rad R$ , so ruling these out is a strong form of J-semisimplicity.
$K / F$ nonalgebraic: There exists $t \in K$ transcendental over $F$ ; equivalently the transcendence degree of $K / F$ is positive.
Modular group algebra: $k G$ with $char k = p$ dividing the order of some element of $G$ — the case Maschke's theorem excludes.

Throughout, $k$ is commutative with identity unless stated otherwise, and $K$ , $F$ denote fields.

Core Concepts

Why the trace is the right invariant

The map $tr : k G \to k$ picking off the coefficient of $1$ is the group-ring analogue of a matrix trace: it is $k$ -linear, it is invariant under cyclic permutation of products, and it is conjugation invariant. Crucially it detects nonzero elements after a shift — if $β \neq 0$ has $β_{g} \neq 0$ , then $g^{- 1} β$ lies in the same left ideal and has nonzero trace. That is the only place where one-sidedness is used, and it is why $(6.13)$ is stated for left ideals with no loss.

The $p$ -th power identity

Expanding a $p$ -th power in $k G$ gives a sum over $p$ -tuples of group elements whose product is $1$ :

tr (β^{p}) = \sum_{(g_{1}, \dots, g_{p}) \in S} β_{g_{1}} β_{g_{2}} \dots β_{g_{p}}, S = {(g_{1}, \dots, g_{p}) : g_{1} g_{2} \dots g_{p} = 1} .

(6.13b)

The set $S$ is stable under cyclic rotation, because $g_{1} \dots g_{p} = 1$ implies $g_{2} \dots g_{p} g_{1} = g_{1}^{- 1} (g_{1} \dots g_{p}) g_{1} = 1$ . So the cyclic group of order $p$ acts on $S$ , and since $p$ is prime every orbit has size $1$ or $p$ .

p

-tuples with product

1

cyclic rotation actionorbits of size

p

die (char

p

)orbits of size

1

need

g^{p} = 1

p'

-group

\Rightarrow

only

(1, \dots, 1)

Commutativity of $k$ is what makes all $p$ tuples in a free orbit contribute the same product, so their total contribution is $p$ times one term, hence zero. Fixed points force $g_{1} = \dots = g_{p}$ with $g_{1}^{p} = 1$ , and in a $p'$ -group that means $g_{1} = 1$ . Only the tuple $(1, \dots, 1)$ survives:

tr (β^{p}) = β_{1}^{p} = tr (β)^{p} .

(6.13c)

A Frobenius-like identity for the trace of a group ring element, valid whenever $char k = p$ and $G$ is a $p'$ -group.

From nil ideals to the radical

Statement $(6.13)$ is about nil left ideals, not about $rad k G$ directly. The bridge is Amitsur's theorem on rational scalar extensions: for a $k$ -algebra $R$ and a set of commuting indeterminates $T$ , $rad R (T) = N (T)$ where $N = R \cap rad R (T)$ is a nil ideal of $R$ . Hence a ring with no nonzero nil ideals has J-semisimple rational extensions — which is precisely how transcendence degree enters Passman's theorem.

Key Results

Proposition(6.13)Passman, Connell

Let $k$ be a commutative reduced ring of prime characteristic $p > 0$ and let $G$ be a ** $p'$ -group**. Then $R = k G$ has no nonzero nil left ideals.

Proof

Suppose $𝔅 \neq 0$ is a nil left ideal and pick $0 \neq β \in 𝔅$ . Choose $g$ with $β_{g} \neq 0$ and replace $β$ by $g^{- 1} β$ , which still lies in $𝔅$ ; so we may assume $tr (β) = β_{1} \neq 0$ .

Expand $tr (β^{p})$ as in $(6.13 b)$ and let the cyclic group of order $p$ act on the index set $S$ by rotation. An orbit of size $p$ consists of $p$ tuples with the same product of coefficients — here commutativity of $k$ is used — so it contributes $p$ times a single element of $k$ , which is $0$ because $char k = p$ . A fixed point satisfies $g_{1} = \dots = g_{p}$ and $g_{1}^{p} = 1$ ; since $G$ has no element of order $p$ , $g_{1} = 1$ . Thus the only fixed point is $(1, \dots, 1)$ and $tr (β^{p}) = β_{1}^{p}$ .

Iterating, $tr (β^{p^{n}}) = β_{1}^{p^{n}}$ for all $n \geq 1$ . Because $k$ is reduced and $β_{1} \neq 0$ , every power $β_{1}^{p^{n}}$ is nonzero, so $β^{p^{n}} \neq 0$ for all $n$ . This contradicts the nilpotence of $β$ . Hence no such $𝔅$ exists. $■$

Remark

Both hypotheses are needed. If $k$ has a nonzero nilpotent $a$ then $a \cdot k G$ is a nonzero nil ideal for any $G$ ; if $G$ has an element $x$ of order $p$ then $(x - 1) k G$ is a nonzero nil left ideal by $(6.13 a)$ . The characteristic-zero counterpart is Lam's $(6.11)$ , which instead uses an involution and a positivity hypothesis on $k$ .

Proposition(6.14)Descent along algebraic extensions

Let $K / F$ be an algebraic extension of fields of characteristic $p$ and let $G$ be a ** $p'$ -group**. If $F G$ is J-semisimple, then so is $K G$ .

Proof

First assume $[K : F] = n < \infty$ . The scalar-extension theorem for radicals gives $(rad K G)^{n} \subseteq (rad F G) \cdot K = 0$ , so $rad K G$ is a nilpotent — in particular nil — ideal of $K G$ . Since $K$ is a field, hence commutative and reduced of characteristic $p$ , and $G$ is a $p'$ -group, $(6.13)$ forces $rad K G = 0$ .

For general algebraic $K / F$ , take $a \in rad K G$ . Only finitely many elements of $K$ occur as coefficients of $a$ , so $a \in K_{0} G$ for some intermediate field $K_{0}$ with $[K_{0} : F] < \infty$ . Contraction of the radical along a scalar extension gives $K_{0} G \cap rad K G \subseteq rad K_{0} G$ , and the finite case already proved shows $rad K_{0} G = 0$ . Hence $a = 0$ . $■$

Theorem(6.15)Passman's Theorem (1962)

Let $K$ be a field extension of $𝔽_{p}$ which is not algebraic over $𝔽_{p}$ . Then for every ** $p'$ -group** $G$ , the group ring $K G$ is J-semisimple.

Proof

Let $T = {t_{i}}$ be a transcendence basis for $K / 𝔽_{p}$ ; by hypothesis $T \neq \emptyset$ . Put $F = 𝔽_{p} (T)$ , so that $F G$ is the rational scalar extension $(𝔽_{p} G) (T)$ .

By Amitsur's theorem on rational extensions, $rad F G = N (T)$ where $N = 𝔽_{p} G \cap rad F G$ is a nil ideal of $𝔽_{p} G$ . But $𝔽_{p}$ is a field of characteristic $p$ and $G$ is a $p'$ -group, so $(6.13)$ gives $N = 0$ ; hence $rad F G = 0$ .

Finally $K / F$ is algebraic — every element of $K$ is algebraic over the field generated by a transcendence basis — so $(6.14)$ upgrades J-semisimplicity from $F G$ to $K G$ . $■$

Proposition(6.15a)A necessary condition

Let $K$ be a field with $char K = p$ and suppose $K G$ is J-semisimple. Then every finite normal subgroup $H ⊴ G$ is a $p'$ -group.

Proof

Let $H ⊴ G$ be finite with $p ∣ | H |$ and set $a = \sum_{h \in H} h \in K G$ . Normality of $H$ makes $a$ invariant under conjugation by $G$ , so $a$ is central in $K G$ ; and $a^{2} = | H | a = 0$ because $p ∣ | H |$ . Then $K G \cdot a$ is a nonzero ideal with $(K G \cdot a)^{2} = K G \cdot a^{2} \cdot K G = 0$ , so it is a nonzero nilpotent ideal and lies in $rad K G$ . Hence $K G$ is not J-semisimple. $■$

Remark(6.15b)Necessary but far from sufficient

The converse fails, and fails for reasons that are still not understood. Wallace showed that for the infinite dihedral group $D_{\infty}$ and any field $K$ with $char K = 2$ , the ring $K G$ is J-semisimple — even though $D_{\infty}$ contains elements of order $2$ ; the point is that no finite normal subgroup of even order exists. Formanek showed that if $G$ is the group of permutations of an infinite set moving only finitely many points, then $K G$ is J-semisimple for every field $K$ , although $G$ has elements of order $p$ for every prime $p$ .

Proof Techniques and Method

How these proofs work, and which move to reuse.

Move 1

Normalise by a group element

A nonzero element of a left ideal can be translated by $g^{- 1}$ to have nonzero trace. Left ideals absorb left multiplication by $G$ , so nothing is lost — this is the standard first line of every group-ring trace argument.

Move 2

Count orbits, not terms

A sum over a set carrying a $ℤ / p$ -action collapses modulo $p$ to its fixed points. This is the same device that proves Cauchy's theorem and Fermat's little theorem, transplanted into $k G$ .

Move 3

Nil ideals as the strong form

Proving no nonzero nil ideals is stronger and more robust than proving $rad = 0$ : it survives scalar extension by Amitsur's theorem, and it converts nilpotence of $rad K G$ into its vanishing.

The three-step architecture — prime field, transcendental extension, algebraic extension — is worth remembering because the same skeleton carries Amitsur's characteristic-zero theorem. The only difference is the last step. In characteristic $0$ a prime field is perfect and the algebraic extension can be taken separable, so the separable descent lemma applies directly; in characteristic $p$ inseparability blocks that route, and $(6.14)$ substitutes an argument that consumes the $p'$ -hypothesis a second time.

Worked Example

A $2'$ -group over $𝔽_{2}$

Take $p = 2$ , $k = 𝔽_{2}$ and $G = C_{3} = ⟨ x ⟩$ , a $2'$ -group. Then

𝔽_{2} C_{3} ≅ 𝔽_{2} [t] / (t^{3} - 1) = 𝔽_{2} [t] / ((t + 1) (t^{2} + t + 1)) ≅ 𝔽_{2} \times 𝔽_{4},

(E.1)

$t^{2} + t + 1$ is irreducible over $𝔽_{2}$ , and the two factors are coprime, so CRT applies.

The result is a product of fields: semisimple, radical zero, no nonzero nil ideals — exactly what $(6.13)$ predicts.

Verifying the trace identity by hand

With $β = β_{1} \cdot 1 + β_{x} x + β_{x^{2}} x^{2}$ in $𝔽_{2} C_{3}$ , the pairs $(g_{1}, g_{2})$ with $g_{1} g_{2} = 1$ are $(1, 1)$ , $(x, x^{2})$ and $(x^{2}, x)$ . Hence

tr (β^{2}) = β_{1}^{2} + 2 β_{x} β_{x^{2}} = β_{1}^{2} = tr (β)^{2} in 𝔽_{2},

(E.2)

The two-element orbit ${(x, x^{2}), (x^{2}, x)}$ contributes $2 β_{x} β_{x^{2}} = 0$ ; only the fixed point survives.

What goes wrong without the $p'$ -hypothesis

Replace $C_{3}$ by $C_{2} = ⟨ y ⟩$ , still over $𝔽_{2}$ . Now $𝔽_{2} C_{2} ≅ 𝔽_{2} [t] / (t^{2} - 1) = 𝔽_{2} [t] / (t + 1)^{2}$ , a local ring of dimension $2$ with

rad (𝔽_{2} C_{2}) = {0, 1 + y}, (1 + y)^{2} = 1 + y^{2} = 0 .

(E.3)

The radical is a nonzero nilpotent ideal, $U (𝔽_{2} C_{2}) = {1, y}$ , and the orbit count in $(6.13 b)$ now has the extra fixed point $(y, y)$ contributing $β_{y}^{2}$ — precisely the term the $p'$ -hypothesis was there to exclude.

Passman's theorem in action

Let $K = 𝔽_{2} (t)$ , which is not algebraic over $𝔽_{2}$ , and let $G$ be any group without elements of order $2$ — a free group, a torsion-free nilpotent group, $C_{3}^{(\infty)}$ , or the additive group $ℚ$ . Then $(6.15)$ gives $rad K G = 0$ with no further work. Replace $K$ by $𝔽_{2}$ itself and the theorem says nothing: that case is open.

Process and Workflow

How to decide J-semisimplicity of a modular group algebra in practice.

Check the characteristicIf

char K = 0

, use Rickart-Amitsur territory instead; only positive characteristic is treated here.

Look for $p$ -torsion in finite normal subgroupsIf some finite

H ⊴ G

has

p ∣ | H |

, stop:

rad K G \neq 0

(6.15 a)

Test the $p'$ -condition on $G$ No element of order

p

at all is the hypothesis under which the general theorems apply.

Measure the base fieldIf

K

has positive transcendence degree over

𝔽_{p}

, Passman's theorem

(6.15)

finishes the job.

Otherwise look for structureFor

K

algebraic over

𝔽_{p}

, fall back on special classes: abelian, locally finite, or ordered groups, each of which has its own theorem.

$char K = p$ — is $K G$ J-semisimple?

Finite normal subgroup of order divisible by $p$ No. A central square-zero ideal is exhibited explicitly in

(6.15 a)

$G$ a $p'$ -group, $K$ nonalgebraic over $𝔽_{p}$ Yes, by Passman's theorem

(6.15)

$G$ a $p'$ -group, $K$ algebraic over $𝔽_{p}$ Open in general. Known for abelian

G

, for locally finite

G

, and for ordered groups.

$G$ has $p$ -torsion but no bad finite normal subgroupNo general answer.

D_{\infty}

in characteristic

2

(Wallace) and the finitary symmetric group (Formanek) are J-semisimple; other such groups are not classified.

Comparison and Classification

The two characteristics side by side
Ingredient	Characteristic $0$	Characteristic $p$
Group hypothesis	none	$G$ a $p'$ -group
Nil-ideal input	$(6.11)$ : involution and formal reality	$(6.13)$ : trace and orbit counting
Base-field hypothesis	$K$ nonalgebraic over $ℚ$	$K$ nonalgebraic over $𝔽_{p}$
Final descent step	separable descent, since $ℚ$ is perfect	$(6.14)$ , using $(6.13)$ a second time
Main theorem	Amitsur $(6.12)$	Passman $(6.15)$
Open case	$K$ algebraic over $ℚ$	$K$ algebraic over $𝔽_{p}$

Which conclusions hold for which data
	$k G$ semisimple	$k G$ J-semisimple	No nonzero nil left ideals
$G$ finite, $p ∤ \| G \|$ , $char k = p$	yes	yes	yes
$G$ finite, $p ∣ \| G \|$ , $char k = p$	no	no	no
$G$ infinite $p'$ -group, $K = 𝔽_{p} (t)$	no	yes	yes
$G$ infinite $p'$ -group, $K = 𝔽_{p}$	no	open	yes
$G = D_{\infty}$ , $char K = 2$	no	yes	no
$G$ abelian $p'$ -group, $K$ any field of char $p$	no	yes	yes

Which conclusions hold for which data

The row for $𝔽_{p} G$ shows the asymmetry that makes the problem hard: the strong conclusion $(6.13)$ is known, yet J-semisimplicity is not.

Relationship Map

The logical dependencies among the characteristic- $p$ results are linear, with $(6.13)$ feeding everything downstream.

$(6.13)$ no nonzero nil left ideals — $k$ commutative reduced, $char k = p$ , $G$ a $p'$ -group
- with Amitsur on $R (T)$
  - $rad 𝔽_{p} (T) G = 0$
- $(6.14)$ algebraic descent
  - $F G$ J-semisimple $\Rightarrow$ $K G$ J-semisimple for $K / F$ algebraic
- $(6.15)$ Passman
  - $K G$ J-semisimple for every $K$ nonalgebraic over $𝔽_{p}$
  - specialises to abelian $G$ via $(6.30)$

p

-torsion in a finite normal subgroupcentral square-zero ideal

rad K G \neq 0

In the other direction, the results of this page sit under Maschke's theorem, which handles finite groups outright, and above the study of normal $p$ -subgroups: for a finite group $G$ with normal $p$ -subgroup $H$ and $char k = p$ , the ideal generated by the augmentation ideal of $k H$ lies inside $rad k G$ — the topic of Normal p-Subgroups and the Radical of kG.

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

The defect of Maschke

When $p$ divides $| G |$ , $k G$ is not semisimple and $rad k G$ measures the failure. Brauer theory, blocks and defect groups are the systematic study of this radical; the $p'$ -condition here is the boundary of the classical theory.

Coding theory

Group codes in characteristic $p$

Cyclic and abelian group codes are ideals in $𝔽_{q} G$ . When $p ∤ | G |$ the algebra splits into fields and the code decomposes cleanly; when $p$ divides $| G |$ the radical produces codes with no complement, which is why coding theory almost always assumes $\gcd (| G |, q) = 1$ .

Computational algebra

Meataxe and module splitting

Algorithms for chopping modules over $𝔽_{q} G$ compute the radical first. GAP and Magma expose the radical of a modular group algebra as a primitive; its dimension is the standard quick diagnostic for whether Maschke applies.

Infinite group theory

A testing ground

The J-semisimplicity problem is a benchmark: a class of infinite groups is well understood only once one can settle $rad k G$ for it. Progress has come class by class — abelian, locally finite, ordered, solvable, linear.

The honest summary is that this material is infrastructure inside algebra. Its downstream consumers are representation theory and the parts of coding theory and symbolic computation that inherit representation-theoretic machinery.

Computational Notes

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

For finite $G$ and finite $k$ the whole question is decidable and cheap; for infinite $G$ nothing is.

For finite $G$ and $char k = p$ , $\dim_{k} rad k G$ is computed by linear algebra in $\dim_{k} k G = | G |$ variables; the Friedl-Rónyai algorithm handles characteristic $p$ , where the trace form alone is inadequate.
$rad k G = 0$ for finite $G$ is decided by a divisibility test, $p ∤ | G |$ — no computation needed once Maschke is invoked.
For infinite $G$ given by a presentation there is no algorithm: the word problem is already undecidable, so membership in $k G$ of a specified element cannot be tested in general.
The reduction in $(6.14)$ is effective in principle: an element of $rad K G$ has finitely many coefficients, hence lives over a finitely generated subfield, and finite-degree instances can be handled by linear algebra.
For abelian $G$ the answer is a decision procedure on the group rather than the ring: test whether $G$ has $p$ -torsion, by $(6.30)$ .

Failure Modes and Common Mistakes

Do not conclude from $(6.15 a)$ that $G$ itself must be a $p'$ -group. $D_{\infty}$ over a field of characteristic $2$ is a J-semisimple counterexample.
Do not read $(6.13)$ as $rad k G = 0$ . It says something formally stronger about nil ideals but does not by itself bound the radical, which is why Amitsur's rational-extension theorem is needed.
Do not assume $rad K G$ is nil in general. That is known here only because $(6.14)$ first makes it nilpotent by a degree bound.
Do not use Maschke's theorem for infinite $G$ : no group ring of an infinite group over a nonzero ring is semisimple.

Historical Notes and Lessons Learned

1898MaschkeSemisimplicity of $k G$ for finite $G$ with $| G |$ invertible in $k$ , the origin of the whole subject and of the $p'$ -condition.
1945Jacobson's radicalThe definition of $rad R$ for arbitrary rings makes J-semisimplicity a meaningful question for infinite groups.
1950RickartBanach-algebra methods show $ℂ G$ and $ℝ G$ are J-semisimple for every group $G$ , setting the agenda.
1959Amitsur $K G$ is J-semisimple for every field $K$ of characteristic zero that is not algebraic over $ℚ$ .
1962-63Passman and ConnellThe trace and orbit-counting argument $(6.13)$ appears independently in work of Passman and of Connell; Passman deduces the characteristic- $p$ analogue of Amitsur's theorem.
1977Passman's treatiseThe Algebraic Structure of Group Rings consolidates the $Δ$ -methods and the semiprimitivity results into the standard reference.

The methodological lesson is that the analytic proof came first and the algebraic proof came second, and the algebraic proof is the one that generalised. Rickart's argument is tied to $ℂ$ ; the trace identity $(6.13 c)$ needs only a commutative reduced coefficient ring and works uniformly in every positive characteristic.

Quick Reference

$p'$ -groupno element of order

p

; for finite

G

p ∤ | G |

$(6.13)$

k

commutative reduced, char

p

G

p'

-group

\Rightarrow

no nonzero nil left ideals in

k G

Trace identity

tr (β^{p}) = tr (β)^{p}

$(6.14)$

K / F

algebraic,

G

p'

-group,

F G

J-semisimple

\Rightarrow

K G

J-semisimple

$(6.15)$

K

nonalgebraic over

𝔽_{p}

G

p'

-group

\Rightarrow

rad K G = 0

Necessary conditionevery finite normal subgroup of

G

is a

p'

-group

Open case

K

algebraic over

𝔽_{p}

, e.g.

K = 𝔽_{p}

Standard failure

rad (𝔽_{2} C_{2}) = {0, 1 + y}

Hypotheses at a glance
Result	On $k$ or $K$	On $G$
$(6.13)$	commutative, reduced, char $p$	$p'$ -group
$(6.14)$	$K / F$ algebraic, char $p$	$p'$ -group
$(6.15)$	$K$ nonalgebraic over $𝔽_{p}$	$p'$ -group
$(6.15 a)$	any field of char $p$	some finite $H ⊴ G$ with $p ∣ \| H \|$
Maschke	$\| G \|$ invertible in $k$	finite

Frequently Asked Questions

Why must

k

be reduced in

(6.13)

, when the conclusion is about

k G

and not about

k

Because $k$ sits inside $k G$ . If $a \in k$ is a nonzero nilpotent then $a \cdot k G$ is a nonzero nil ideal of $k G$ , so the conclusion fails before the group is even consulted. Reducedness is also used at the end of the proof, where $β_{1}^{p^{n}} \neq 0$ is needed for every $n$ .

Does

(6.13)

say that

rad k G = 0

No, and the gap is the substance of the subject. It says there is no nonzero nil left ideal. The radical of a group ring need not be nil, so this does not bound it directly. The conclusion becomes J-semisimplicity only after tensoring up to a rational function field, where Amitsur's theorem says the radical is controlled by a nil ideal of the original ring.

Is the

p'

-hypothesis really necessary for J-semisimplicity?

Not as stated. What is necessary is the weaker condition that every finite normal subgroup be a $p'$ -group. The infinite dihedral group in characteristic $2$ satisfies the weaker condition and fails the stronger one, and Wallace proved its group algebra is J-semisimple. No characterisation is known.

Why is the prime field the hard case?

Every step in the proof consumes transcendence. Amitsur's rational-extension theorem needs a nonempty transcendence basis to have anything to adjoin; over $𝔽_{p}$ there is nothing to adjoin and the argument has no purchase. The same phenomenon occurs in characteristic zero, where the algebraic number fields are the unresolved case.

How does this relate to Maschke's theorem?

Maschke settles finite groups completely: $k G$ is semisimple exactly when $| G |$ is invertible in $k$ . For finite $G$ in characteristic $p$ , being a $p'$ -group is precisely $p ∤ | G |$ , so $(6.13)$ recovers half of Maschke. The content of this page is everything Maschke cannot see, namely infinite $G$ , where semisimplicity is impossible and J-semisimplicity is the right question.

Where does the commutativity of

k

actually get used?

In the collapse of free orbits. The $p$ tuples in a rotation orbit give the products $β_{g_{1}} \dots β_{g_{p}}$ , $β_{g_{2}} \dots β_{g_{p}} β_{g_{1}}$ and so on. These agree only if the coefficients commute, and only then does the orbit contribute $p$ times a single element, hence zero.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §6, results (6.13)-(6.15) (pp. 92-95).
D. S. Passman, The Algebraic Structure of Group Rings, Wiley-Interscience, New York, 1977.
D. S. Passman, “Nil ideals in group rings”, Michigan Mathematical Journal 9 (1962).
I. G. Connell, “On the group ring”, Canadian Journal of Mathematics 15 (1963).
S. A. Amitsur, “On the semi-simplicity of group algebras”, Michigan Mathematical Journal 6 (1959).
D. S. Passman, Infinite Group Rings, Pure and Applied Mathematics 6, Marcel Dekker, New York, 1971.

AI Suggested Questions

Write out the orbit-counting proof of $(6.13)$ for $p = 3$ and a concrete group of exponent coprime to $3$ .
What is known about $rad 𝔽_{p} G$ for $G$ a finitely generated torsion-free nilpotent group?
How does Formanek's theorem on the finitary symmetric group avoid the $p'$ -hypothesis entirely?
Compare $(6.13)$ with Lam's $(6.11)$ : what replaces the involution and formal reality in characteristic $p$ ?
Explain why an algebraic extension in characteristic $p$ can fail to be separable and what that costs in $(6.14)$ .
Give the modular representation theory reading of $rad k G$ for $G$ finite with a normal Sylow $p$ -subgroup.
Which classes of infinite groups are currently known to have $𝔽_{p} G$ J-semisimple?

Group Rings in Characteristic p

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Why the trace is the right invariant

The $p$ -th power identity

From nil ideals to the radical

Key Results

Proof Techniques and Method

Normalise by a group element

Count orbits, not terms

Nil ideals as the strong form

Worked Example

A $2'$ -group over $𝔽_{2}$

Verifying the trace identity by hand

What goes wrong without the $p'$ -hypothesis

Passman's theorem in action

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

The defect of Maschke

Group codes in characteristic $p$

Meataxe and module splitting

A testing ground

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

References

AI Suggested Questions

Continue learning

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

Why the trace is the right invariant

The p-th power identity

From nil ideals to the radical

06Key Results

07Proof Techniques and Method

Normalise by a group element

Count orbits, not terms

Nil ideals as the strong form

08Worked Example

A 2′-group over 𝔽2

Verifying the trace identity by hand

What goes wrong without the p′-hypothesis

Passman's theorem in action

09Process and Workflow

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

The defect of Maschke

Group codes in characteristic p

Meataxe and module splitting

A testing ground

13Computational Notes

14Failure Modes and Common Mistakes

15Historical Notes and Lessons Learned

16Quick Reference

17Frequently Asked Questions

18Related KEVOS Topics

19References

20AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

The $p$ -th power identity

Key Results

Proof Techniques and Method

Worked Example

A $2'$ -group over $𝔽_{2}$

What goes wrong without the $p'$ -hypothesis

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Group codes in characteristic $p$

Computational Notes

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions