The Multiplicative Group of a Division Ring

Q: Why is proving G_2 = G_1 enough for the whole series?

Because G_2/G_1 = Z(G/G_1), so G_2 = G_1 says precisely that G/G_1 has trivial centre. The next term satisfies G_3/G_2 = Z(G/G_2) = Z(G/G_1) = 1, giving G_3 = G_2, and the same computation repeats. The series cannot resume climbing once it has stalled.

Q: Does (13.20) say D^* has trivial centre?

No — the centre is Z(D)^* and is usually large. What it says is that the *quotient* D^*/Z(D)^* has trivial centre. Equivalently, the hypercentre of D^* is no bigger than its centre.

Q: Where exactly does the proof use that D is a division ring rather than a group?

Twice, and both times through subtraction. Forming b = a - 1 requires addition, and forming - inside the field Z(D) requires that the centre be closed under subtraction. Strip away the ring structure and the statement becomes false, as any nilpotent non-abelian group shows.

Q: How does this relate to the Cartan–Brauer–Hua theorem?

They are the same argument with a different target. (13.17) feeds identity (13.13) a division subring K and concludes a ∈ K; (13.20) feeds identity (13.14) the centre and concludes a ∈ Z(D). In both cases what makes the conclusion possible is that the target set, together with 0, is closed under subtraction and inversion.

Executive Summary

For a division ring $D$ , the group $G = D^{*}$ has centre $G_{1} = Z (D)^{*}$ . The theorem of this page says the ascending central chain stops there: $G_{1} = G_{2} = G_{3} = \dots$ . There is no element outside the centre whose commutators are all central.

The consequence is decisive. A nilpotent group is one whose upper central series reaches the top; since the series never moves at all, $D^{*}$ is nilpotent precisely when $D^{*} = Z (D)^{*}$ , i.e. when $D$ is a field. Hua later proved the same with solvable in place of nilpotent, a considerably deeper result.

$G_{1} = G_{2}$ The theorem

$Z (D)^{*}$ The hypercentre of

D^{*}

$(13.14)$ Identity reused

HuaSolvable case

Overview

A division ring carries two structures, and $D^{*}$ discards one of them. The natural question is how much of $D$ survives in the group: can the group $D^{*}$ be commutative-like without $D$ being commutative? The answer given here is no, and in the strongest sense available — not merely is $D^{*}$ non-abelian when $D$ is noncommutative, its entire nilpotency hierarchy collapses to the centre.

Recall the upper central series of a group $G$ :

{1} \subseteq G_{1} \subseteq G_{2} \subseteq \dots, G_{1} = Z (G), G_{k + 1} / G_{k} = Z (G / G_{k}) .

(13.20a)

$G$ is nilpotent if $G_{n} = G$ for some $n$ ; the union of the $G_{k}$ is the hypercentre.

Membership in $G_{2}$ has a concrete meaning: $c \in G_{2}$ if and only if $x^{- 1} c x c^{- 1} \in G_{1}$ for every $x \in G$ — every multiplicative commutator involving $c$ is central. Stated that way, the theorem is visibly a relative of $(13.15)$ , and it is proved with the same identity.

This is the pivot on which the argument turns: $G_{1}$ is not merely a central subgroup, it is the unit group of a subfield, and subtraction is available. A general group has no such closure and no such theorem.

Learning Objectives

Write down the upper central series and characterise $G_{2}$ by the condition $x^{- 1} c x c^{- 1} \in G_{1}$ .
Prove that $Z (D^{*}) = Z (D)^{*}$ and that $Z (D^{*}) \cup {0}$ is a field.
Prove $(13.20)$ : $G_{2} = G_{1}$ , hence the whole series is constant.
Deduce $(13.21)$ : $D^{*}$ nilpotent $iff$ $D$ commutative.
Explain why identity $(13.14)$ is the right tool and where the field structure of $Z (D)$ enters.
State Hua's theorem on solvable multiplicative groups and locate it relative to $(13.21)$ .

Definitions

Definition(13.20a)Upper central series and nilpotency

For a group $G$ set $G_{0} = {1}$ and, inductively, let $G_{k + 1}$ be the preimage in $G$ of $Z (G / G_{k})$ under the quotient map. The resulting chain $G_{0} \subseteq G_{1} \subseteq G_{2} \subseteq \dots$ is the upper central series; $G_{1} = Z (G)$ . The group $G$ is nilpotent if $G_{n} = G$ for some finite $n$ , and the least such $n$ is its nilpotency class.

$G_{2}$ , the second centre: ${c \in G : x^{- 1} c x c^{- 1} \in Z (G) for all x \in G}$ . Elements that are central modulo the centre.
$Z (D^{*})$: Equal to $Z (D)^{*} = Z (D) ∖ {0}$ : an element commuting with every nonzero element of $D$ commutes with $0$ as well.
Hypercentre: The union of the terms of the upper central series, continued transfinitely. For $D^{*}$ it equals $Z (D)^{*}$ , by $(13.20)$ .
Derived series: $G^{(0)} = G$ , $G^{(k + 1)} = [G^{(k)}, G^{(k)}]$ . $G$ is solvable if $G^{(n)} = {1}$ for some $n$ . Nilpotent implies solvable, never the converse.
$b = a - 1$: The substitution from $(13.13)$ – $(13.14)$ , reused verbatim here. Requires $a \neq 1$ , which holds because $a$ is chosen not to commute with $c$ .

Nilpotency and solvability are properties of the abstract group D star; the theorem says both detect commutativity of the ring D exactly.

Core Concepts

Why the second centre is the only thing to check

Central series arguments usually proceed one layer at a time, but here a single layer suffices. If $G_{2} = G_{1}$ then $G / G_{1}$ has trivial centre, so $G_{3} / G_{2} = Z (G / G_{2}) = Z (G / G_{1})$ is trivial and $G_{3} = G_{2}$ ; the same step repeats indefinitely. Hence the whole content is the equality of the first two terms.

G_{2} = G_{1}

Z (G / G_{1})

trivial

G_{3} = G_{2}

G_{k} = G_{1}

for all

k \geq 1

The field structure of the centre

The proof of $(13.15)$ used that a difference of two commutators is invertible when nonzero. Here more is needed: the two commutators lie in $G_{1}$ , and their difference must lie in a place where it can be inverted and where $a$ is forbidden to live. Both are supplied by

G_{1} \cup {0} = Z (D^{*}) \cup {0} = Z (D),

(13.20b)

a field, hence closed under subtraction and under inversion of nonzero elements.

If $α, β \in G_{1}$ then $α - β \in Z (D)$ and $1 - β \in Z (D)$ . Identity $(13.14)$ says $a (α - β) = 1 - β \neq 0$ , so $α - β \neq 0$ and $a = (1 - β) (α - β)^{- 1}$ is a quotient of two elements of the field $Z (D)$ — hence central. That is the contradiction.

What fails for abstract groups

There is no group-theoretic theorem of this shape: plenty of nilpotent groups of class $2$ exist, the quaternion group $Q_{8}$ among them. Indeed $Q_{8} \subseteq ℍ^{*}$ is nilpotent of class $2$ while $ℍ^{*}$ is not nilpotent at all. The theorem is about the whole multiplicative group of a division ring, and its proof uses subtraction at every step.

Key Results

Theorem(13.20)The upper central series of

D^{*}

is constant

Let $D$ be a division ring and let ${1} \subseteq G_{1} \subseteq G_{2} \subseteq \dots$ be the upper central series of $G = D^{*}$ . Then $G_{1} = G_{2} = G_{3} = \dots$ ; equivalently, the hypercentre of $D^{*}$ is $Z (D)^{*}$ .

Proof

If $D$ is commutative then $G = G_{1}$ and every term equals $G$ , so assume $D$ is noncommutative. It suffices to prove $G_{2} = G_{1}$ ; the rest follows by the induction sketched above.

Suppose $c \in G_{2} ∖ G_{1}$ . Then $c \notin Z (D)$ , so there is $a \in D$ with $c a \neq a c$ ; necessarily $a \neq 0$ and $a \neq 1$ , so $b : = a - 1 \in D^{*}$ , and $b$ likewise fails to commute with $c$ .

Because $c \in G_{2}$ , its image in $G / G_{1}$ is central, which means $x^{- 1} c x c^{- 1} \in G_{1}$ for every $x \in D^{*}$ . Apply this with $x = a$ and with $x = b$ and set

α = a^{- 1} c a c^{- 1} \in G_{1}, β = b^{- 1} c b c^{- 1} \in G_{1} .

(13.20c)

Identity $(13.14)$ now reads

a (α - β) = 1 - β \neq 0 .

(13.20d)

By $(13.20 b)$ , $Z (D) = G_{1} \cup {0}$ is a field, so $α - β$ and $1 - β$ both lie in $Z (D)$ . Since $a (α - β) \neq 0$ we have $α - β \neq 0$ , hence it is invertible in $Z (D)$ and

a = (1 - β) (α - β)^{- 1} \in Z (D) .

(13.20e)

But then $a$ commutes with $c$ , contradicting the choice of $a$ . Therefore no such $c$ exists and $G_{2} = G_{1}$ .

Corollary(13.21)Nilpotency detects commutativity

For a division ring $D$ , the group $D^{*}$ is nilpotent if and only if $D$ is a field.

Proof

If $D$ is a field then $D^{*}$ is abelian, hence nilpotent of class $1$ . Conversely, if $D^{*}$ is nilpotent then $G_{n} = G$ for some $n$ ; by $(13.20)$ , $G_{n} = G_{1} = Z (D)^{*}$ , so $D^{*} = Z (D)^{*}$ and therefore $D = Z (D)$ is commutative.

Corollary—Nothing is central modulo the centre

Let $D$ be a division ring and $c \in D^{*}$ . If every multiplicative commutator $x^{- 1} c x c^{- 1}$ , $x \in D^{*}$ , lies in $Z (D)$ , then $c \in Z (D)$ . This is $(13.20)$ restated, and it strengthens $(13.15)$ : there the hypothesis was that $c$ commutes with all commutators, here it is that the commutators *involving $c$ * are central.

Remark—Hua's solvable analogue

$(13.21)$ remains true with nilpotent replaced by solvable: if $D^{*}$ is solvable then $D$ is a field. This stronger statement is due to L. K. Hua and its proof is substantially harder — the derived series does not admit the one-layer reduction that makes $(13.20)$ short, so no single application of $(13.14)$ suffices.

Proof Techniques and Method

How this proof works, and which move to reuse.

The argument is short because three separate reductions are available, and it is worth separating them.

Reduce an infinite chain to one step

G_{2} = G_{1}

makes

G / G_{1}

centreless, which propagates up the series automatically. Never verify more layers than you must.

Translate the layer condition into commutators

c \in G_{2}

means every

x^{- 1} c x c^{- 1}

is central — a statement about a set of commutators, which is the form the §13 identities consume.

Feed two values into the identityApply the condition at

x = a

and at

x = b = a - 1

. The identity

(13.14)

relates exactly those two.

Exploit that the centre is a fieldSubtract, invert, and read off

a \in Z (D)

. This is the step that uses more than group theory.

Three theorems, one identity
	Hypothesis on commutators	Set $S$ used	Conclusion
(13.15)	$c$ centralises them	centralizer of $c$	$c \in Z (D)$
(13.17) Cartan–Brauer–Hua	conjugates of $c$ lie in $K$	$K$	$a \in K$ , then $K \subseteq Z (D)$
(13.20)	commutators with $c$ are central	$Z (D)$	$a \in Z (D)$ , so $G_{2} = G_{1}$

Three theorems, one identity

Worked Example

The identity checked numerically in $ℍ$

Take $D = ℍ$ , $Z (D) = ℝ$ , and test whether $c = i$ could lie in $G_{2}$ . Choose $a = j$ , which does not commute with $i$ , so $b = a - 1 = j - 1$ , with $b^{- 1} = \frac{1}{2} (- 1 - j)$ since $‖ b ‖^{2} = 2$ . Compute the two commutators of $(13.20 c)$ :

α = j^{- 1} i j i^{- 1} = (- j) (i) (j) (- i) = (j i) (j i) = (- k) (- k) = - 1,

(E.1)

β = b^{- 1} i b i^{- 1} = \frac{1}{2} (- 1 - j) (k - i) (- i) = (- k) (- i) = j .

(E.2)

Intermediate steps: $i (j - 1) = k - i$ , and $\frac{1}{2} (- 1 - j) (k - i) = \frac{1}{2} (- k + i - i - k) = - k$ .

Now verify $(13.14)$ directly. The left-hand side is

a (α - β) = j (- 1 - j) = - j - j^{2} = 1 - j,

(E.3)

and the right-hand side is $1 - β = 1 - j$ . The identity holds exactly, and it is nonzero as promised.

Reading off the conclusion

Here $α = - 1$ is central, but $β = j$ is not. So the hypothesis " $c = i$ lies in $G_{2}$ " fails at $x = b$ , and no contradiction is needed — the theorem is confirmed rather than tested. Had both been central, $(13.20 e)$ would have given $j = (1 - β) (α - β)^{- 1} \in ℝ$ , which is false.

The corollary in the same example

$ℍ^{*}$ is therefore not nilpotent, and its hypercentre is $ℝ^{*}$ . Note this coexists with $Q_{8} = {\pm 1, \pm i, \pm j, \pm k} \subseteq ℍ^{*}$ being nilpotent of class $2$ — subgroups inherit nothing from $(13.20)$ .

Comparison and Classification

Group-theoretic conditions on $D^{*}$ and what they force
Condition on $D^{*}$	Forces $D$ commutative?	Source
Abelian	yes, by definition	trivial
Nilpotent	yes	(13.21)
Solvable	yes	Hua
Some $G_{k} \neq G_{1}$	impossible	(13.20)
Has a nilpotent subgroup	no	$Q_{8} \subseteq ℍ^{*}$
Has a finite subgroup	no	$(13.3)$ and $ℍ^{*}$
Finite	yes	$(13.1)$ Wedderburn

The pattern is that global hypotheses on $D^{*}$ collapse the ring, while local ones — about subgroups — do not. Wedderburn's Little Theorem is the extreme case of a global hypothesis; $(13.21)$ is the extreme case of a purely group-theoretic one.

$D^{*}$ for $D$ noncommutative — what is and is not true
- is not
  - nilpotent
  - solvable (Hua)
  - of nontrivial hypercentre
- may well be
  - generated by its commutators $(13.19)$
  - possessed of finite nilpotent subgroups
  - possessed of infinite abelian subgroups — every maximal subfield gives one

Relationship Map

(13.13)

–

(13.14)

(13.15)

(13.17)

(13.20)

(13.21)

Hua's solvable theorem

Everything on this page descends from the same two lines of algebra that prove Multiplicative Commutators in Division Rings. The novelty in $(13.20)$ is not the identity but the choice of target set: taking $S = Z (D)^{*}$ and using that $S \cup {0}$ is a field is what converts a statement about commutators into a statement about a central series.

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

Solvability tests

In the theory of linear groups over division rings, $(13.21)$ and Hua's theorem are the base cases: a solvable subgroup of $G L_{n} (D)$ cannot be all of $D^{*}$ embedded diagonally unless $D$ is a field. This underpins Lie–Kolchin style triangularisation results in the noncommutative setting.

Algebraic K-theory

$K_{1}$ of a division ring

$K_{1} (D) = D^{*} / [D^{*}, D^{*}]$ is the abelianisation of exactly this group. $(13.20)$ says the commutator subgroup is not confined near the centre, which is the qualitative input behind the study of $S K_{1}$ and the Dieudonné determinant.

Group theory

A source of centreless quotients

$D^{*} / Z (D)^{*}$ is a group with trivial centre for every noncommutative $D$ — a supply of centreless groups arising naturally rather than by construction, used as test objects in infinite group theory.

Internal to algebra

Honest summary

This result is a structural constraint rather than a computational tool. Its practical use is negative: it rules out attacking division rings by nilpotency or solvability arguments on their unit groups.

Failure Modes and Common Mistakes

Do not assume $G_{2} = G_{1}$ is a formality — in a general group the second centre is usually strictly larger, and every finite $p$ -group of class $2$ is an example.
Do not omit the reduction to noncommutative $D$ : for a field, $G_{1} = G$ already and the statement reads differently.
Do not evaluate the $G_{2}$ condition at a single $x$ . The example above shows $x = j$ alone gives a central commutator while $x = j - 1$ does not.

Historical Notes and Lessons Learned

1905WedderburnFiniteness of $D$ forces commutativity — the first theorem asserting that a global condition on $D^{*}$ collapses the ring.
1949Hua on sfieldsHua's work on the multiplicative structure of division rings, including his proof of the Cartan–Brauer–Hua theorem, establishes the identities used throughout §13.
1950Hua's solvability theoremHua proves that a division ring whose multiplicative group is solvable is commutative, strictly strengthening the nilpotent statement.
1955AmitsurThe classification of finite subgroups of division rings in characteristic zero shows how much freedom remains at the level of subgroups, in sharp contrast with the rigidity of $D^{*}$ itself.
1983Draxl's synthesisDraxl's Skew Fields collects the multiplicative theory of division rings, including the Dieudonné determinant and $S K_{1}$ , placing $(13.20)$ in its K-theoretic context.

The lesson is about the direction of information flow. One might expect the ring to constrain the group; here the group constrains the ring, and it does so because the group is not merely abstract — the proof secretly uses that its elements can be subtracted.

Quick Reference

Upper central series

G_{1} = Z (G)

G_{k + 1} / G_{k} = Z (G / G_{k})

Second centre

c \in G_{2} iff x^{- 1} c x c^{- 1} \in G_{1}

for all

x

Centre of $D^{*}$

G_{1} = Z (D)^{*}

, and

G_{1} \cup {0} = Z (D)

is a field

(13.20)

G_{1} = G_{2} = G_{3} = \dots

Key step

a (α - β) = 1 - β \neq 0

with

α, β \in G_{1}

(13.21)

D^{*}

nilpotent

iff

D

a field

Hua

D^{*}

solvable

iff

D

a field

Not impliedNothing about subgroups of

D^{*}

Worked values in $ℍ$ with $c = i$ , $a = j$ , $b = j - 1$
Quantity	Value	Central?
$α = a^{- 1} c a c^{- 1}$	$- 1$	yes
$β = b^{- 1} c b c^{- 1}$	$j$	no
$a (α - β)$	$1 - j$	no
$1 - β$	$1 - j$	no

Frequently Asked Questions

Why is proving

G_{2} = G_{1}

enough for the whole series?

Because $G_{2} / G_{1} = Z (G / G_{1})$ , so $G_{2} = G_{1}$ says precisely that $G / G_{1}$ has trivial centre. The next term satisfies $G_{3} / G_{2} = Z (G / G_{2}) = Z (G / G_{1}) = {1}$ , giving $G_{3} = G_{2}$ , and the same computation repeats. The series cannot resume climbing once it has stalled.

Does

(13.20)

say

D^{*}

has trivial centre?

No — the centre is $Z (D)^{*}$ and is usually large. What it says is that the quotient $D^{*} / Z (D)^{*}$ has trivial centre. Equivalently, the hypercentre of $D^{*}$ is no bigger than its centre.

Where exactly does the proof use that

D

is a division ring rather than a group?

Twice, and both times through subtraction. Forming $b = a - 1$ requires addition, and forming $α - β$ inside the field $Z (D)$ requires that the centre be closed under subtraction. Strip away the ring structure and the statement becomes false, as any nilpotent non-abelian group shows.

Why is Hua's solvable version harder?

The nilpotent case reduces to a single layer of the upper central series, and that layer is described by a condition on commutators that identity $(13.14)$ can consume directly. The derived series has no such one-step reduction: solvability of $D^{*}$ bounds the length of the derived series but gives no immediate statement about which commutators are central, so a genuinely different argument is required.

How does this relate to the Cartan–Brauer–Hua theorem?

They are the same argument with a different target. $(13.17)$ feeds identity $(13.13)$ a division subring $K$ and concludes $a \in K$ ; $(13.20)$ feeds identity $(13.14)$ the centre and concludes $a \in Z (D)$ . In both cases what makes the conclusion possible is that the target set, together with $0$ , is closed under subtraction and inversion.

Can

D^{*}

be finitely generated for noncommutative

D

That is a separate question and is not settled by anything on this page. $(13.20)$ constrains only the central series. What it does supply is that $D^{*}$ has no nontrivial hypercentre, so any finite generation would have to be by elements far from the centre.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §13, (13.20)–(13.21), pp. 223–224.
L. K. Hua, “Some properties of a sfield”, Proceedings of the National Academy of Sciences of the USA 35 (1949), 533–537.
P. Draxl, Skew Fields, London Mathematical Society Lecture Note Series 81, Cambridge University Press, 1983.
I. N. Herstein, Noncommutative Rings, Carus Mathematical Monographs 15, Mathematical Association of America, 1968, Chapter 3.
D. J. S. Robinson, A Course in the Theory of Groups, 2nd edition, Graduate Texts in Mathematics 80, Springer-Verlag, 1996, Chapter 5 (nilpotent and solvable groups).

AI Suggested Questions

Give a full proof of Hua's theorem that a division ring with solvable multiplicative group is commutative.
Compute the abelianisation $D^{*} / [D^{*}, D^{*}]$ for a quaternion division algebra over a number field.
Explain the Dieudonné determinant and its relationship to $K_{1}$ of a division ring.
Show that $D^{*} / Z (D)^{*}$ has trivial centre and decide whether it can be simple.
Which nilpotent groups embed as subgroups of the multiplicative group of a division ring?
How does $(13.20)$ interact with the classification of maximal subfields of a centrally finite division ring?
Is there an analogue of $(13.20)$ for the unit group of a simple artinian ring $M_{n} (D)$ ?

The Group D*

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

Why the second centre is the only thing to check

The field structure of the centre

What fails for abstract groups

Key Results

Proof Techniques and Method

Worked Example

The identity checked numerically in $ℍ$

Reading off the conclusion

The corollary in the same example

Comparison and Classification

Relationship Map

Applications and Industry Use

Solvability tests

$K_{1}$ of a division ring

A source of centreless quotients

Honest summary

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 second centre is the only thing to check

The field structure of the centre

What fails for abstract groups

06Key Results

07Proof Techniques and Method

08Worked Example

The identity checked numerically in ℍ

Reading off the conclusion

The corollary in the same example

09Comparison and Classification

10Relationship Map

11Applications and Industry Use

Solvability tests

K1 of a division ring

A source of centreless quotients

Honest summary

12Failure Modes and Common Mistakes

13Historical Notes and Lessons Learned

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 identity checked numerically in $ℍ$

Comparison and Classification

Relationship Map

Applications and Industry Use

$K_{1}$ of a division ring

Failure Modes and Common Mistakes

Historical Notes and Lessons Learned

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions