The Opposite Ring and Left–Right Duality

Q: Is R^op ever equal to R, rather than merely isomorphic?

Yes, and exactly when R is commutative. Then the identity map on the underlying set is a ring isomorphism, so the two rings are the same ring with the same multiplication. For noncommutative R the identity map is never a homomorphism, though some other bijection may be.

Q: If R ≅ R^op, does that mean every left theorem holds on the right?

For that particular ring, yes — any left-handed property R enjoys is enjoyed on the right, because the isomorphism transports it. It says nothing about other rings, and it does not make the property side-neutral as a general notion.

Q: How does the opposite ring interact with tensor products of algebras?

For F-algebras, (A ⊗_F B)^op ≅ A^op ⊗_F B^op. The important consequence is in Brauer theory: for a central simple F-algebra A of dimension n^2, A ⊗_F A^op ≅ M_n(F) ≅ End_F(A), which is why [A^op] inverts [A] in the Brauer group.

Q: Does a group ring always equal its opposite?

For a commutative coefficient ring k and any group G, the k-linear extension of g g^-1 is an involution of kG, so kG ≅ (kG)^op. The computation needs k commutative: the coefficients must be allowed to swap past one another when the product is reversed.

Executive Summary

The opposite ring $R^{op}$ has the same elements and the same addition as $R$ , with multiplication written backwards. It costs nothing to define and it buys a precise formulation of the duality principle: a theorem about left ideals of $R$ is literally a theorem about right ideals of $R^{op}$ .

The catch is that $R^{op}$ need not be isomorphic to $R$ . When it is — for commutative rings, matrix rings over commutative rings, group rings, and any ring with an involution — left and right theory coincide for that ring. When it is not, one-sided phenomena become possible, and they occur.

$a b ⇝ b a$ The whole construction

$(R^{op})^{op} = R$ Involutive

$≅$ iff anti-aut.Self-opposite test

$M_{n} (R)^{op}$ Transpose handles it

Overview

Fix a ring $R$ . Write $a^{op}$ for the element $a$ regarded as a member of a second copy of the additive group of $R$ , and define

a^{op} \cdot b^{op} = (b a)^{op} (a, b \in R) .

(1.0c)

Addition is unchanged; the identity of $R^{op}$ is $1^{op}$ .

Associativity holds because $(a^{op} b^{op}) c^{op} = (c b a)^{op} = a^{op} (b^{op} c^{op})$ , and distributivity is inherited. The construction is involutive: $(R^{op})^{op} = R$ on the nose.

This distinction is the difference between a labour-saving device and a false step. Lam is explicit that results proved on one side may be used freely on the other provided the same argument works there — and the reason it usually does is that the argument, transported to $R^{op}$ , is the same argument.

Learning Objectives

Define $R^{op}$ and check the ring axioms, including $(R^{op})^{op} = R$ .
State the isomorphism of categories between left $R$ -modules and right $R^{op}$ -modules.
Prove $M_{n} (R)^{op} ≅ M_{n} (R^{op})$ using the transpose.
Show that $R ≅ R^{op}$ if and only if $R$ carries an anti-automorphism, and give examples on both sides of the dichotomy.
List which ring properties are side-neutral and which are not, with a witness for each failure.
Use the opposite ring to convert a left-noetherian-only example into a right-noetherian-only example.

Definitions

Definition§1The opposite ring

For a ring $R$ , the opposite ring $R^{op}$ has underlying abelian group $R$ and multiplication $a^{op} b^{op} = (b a)^{op}$ . A map $f : R \to S$ is an anti-homomorphism if it is additive, sends $1$ to $1$ , and satisfies $f (a b) = f (b) f (a)$ ; equivalently $f$ is a ring homomorphism $R \to S^{op}$ , equivalently $R^{op} \to S$ .

$R^{op}$: Opposite ring. Some authors write $R^{\circ}$ or $R^{*}$ ; the superscript should be typeset upright.
Involution: An anti-automorphism $σ$ with $σ^{2} = id$ . Complex conjugation on $ℂ$ , quaternion conjugation on $ℍ$ and the transpose on $M_{n} (k)$ for commutative $k$ are the standard examples.
Self-opposite: A ring with $R ≅ R^{op}$ . The isomorphism is not required to be canonical, and is not required to be an involution.
Left ideal of $R$: The same subset is a right ideal of $R^{op}$ , and conversely.
$Z (R)$: The centre is unchanged: $Z (R^{op}) = Z (R)$ as subsets, and the identity map is a ring isomorphism between them because central elements commute.

Nothing in the construction requires $R$ to be interesting: $R^{op} = R$ as rings precisely when $R$ is commutative.

Core Concepts

What the opposite ring is for

There are three distinct uses, and conflating them causes errors.

As a translation device. Every left-handed definition becomes a right-handed one by passing to $R^{op}$ . This is bookkeeping and is always valid.
As a test for symmetry. A property is side-neutral exactly when it is preserved by $R \mapsto R^{op}$ and the left version implies the right version for each individual ring. The first condition is automatic for well-posed definitions; the second is a theorem or a falsehood, case by case.
As a genuine new ring. For central simple algebras, $D^{op}$ carries the inverse Brauer class, and $D \otimes_{F} D^{op} ≅ {End}_{F} (D)$ . Here the opposite is not a bookkeeping shadow but an object of study.

Modules change sides, functors do not

The category of left $R$ -modules and the category of right $R^{op}$ -modules are equal, not merely equivalent: same objects with a relabelled action, same morphisms. This is why $End (_{R} R) ≅ R^{op}$ — the endomorphism ring of the regular left module is computed by right multiplications, whose composition law is reversed.

_{R} Mod

{Mod}_{R^{op}}

_{R^{op}} Mod

{Mod}_{R}

When is $R$ its own opposite?

Exactly when an anti-automorphism exists. The supply is generous: every commutative ring; every matrix ring over a commutative ring, via the transpose; every group ring $k G$ over a commutative $k$ , via $g \mapsto g^{- 1}$ ; the real quaternions, via conjugation; every algebra with an involution, by definition.

Key Results

Proposition§1Modules over the opposite ring

Let $R$ be a ring and $M$ an abelian group. The assignments $m \cdot a^{op} : = a m$ and $a m : = m \cdot a^{op}$ are mutually inverse bijections between left $R$ -module structures on $M$ and right $R^{op}$ -module structures on $M$ . They preserve submodules and module homomorphisms, hence give an isomorphism of categories $_{R} Mod = {Mod}_{R^{op}}$ .

Proof

Given a left $R$ -action, biadditivity of $(m, a^{op}) \mapsto a m$ is immediate. For the associativity axiom of a right module,

(m \cdot a^{op}) \cdot b^{op} = b (a m) = (b a) m = m \cdot (b a)^{op} = m \cdot (a^{op} b^{op}),

which is exactly what is required. Unitality is $m \cdot 1^{op} = 1 m = m$ . The reverse assignment is the same computation read backwards, and both leave the underlying abelian group and the collection of additive maps untouched, so submodules and homomorphisms correspond.

Theorem§1Transpose identifies the opposite of a matrix ring

Let $R$ be any ring and $n \geq 1$ . The map sending a matrix to its transpose, with entries read in $R^{op}$ , is a ring isomorphism

Φ : M_{n} (R)^{op} \tilde{⟶} M_{n} (R^{op}), Φ (A)_{i j} = (a_{j i})^{op} .

(1.0d)

In particular $M_{n} (R)$ is self-opposite whenever $R$ is; for commutative $R$ this is the familiar statement that the transpose is an anti-automorphism of $M_{n} (R)$ .

Proof

$Φ$ is additive and bijective, being the transpose on underlying sets. It sends the identity matrix to the identity matrix. It remains to check multiplicativity, where the multiplication on the source is $A * B = B A$ computed in $M_{n} (R)$ .

Compute the $(i, j)$ entry of $Φ (A * B) = Φ (B A)$ :

Φ (B A)_{i j} = ((B A)_{j i})^{op} = (textstyle \sum_{k} b_{j k} a_{k i})^{op} = \sum_{k} (b_{j k} a_{k i})^{op} = \sum_{k} (a_{k i})^{op} (b_{j k})^{op},

the last step by the definition of multiplication in $R^{op}$ . On the other side,

(Φ (A) Φ (B))_{i j} = \sum_{k} Φ (A)_{i k} Φ (B)_{k j} = \sum_{k} (a_{k i})^{op} (b_{j k})^{op} .

The two agree term by term, so $Φ$ is a ring isomorphism. Note that the transpose alone is not an anti-automorphism of $M_{n} (R)$ for noncommutative $R$ — the entries must be reinterpreted in $R^{op}$ , which is precisely what the theorem records.

Proposition§1Self-oppositeness criterion

For a ring $R$ the following are equivalent: (1) $R ≅ R^{op}$ as rings; (2) $R$ admits an anti-automorphism. If $R$ is commutative both hold, with the identity map serving.

Proof

An isomorphism $f : R \to R^{op}$ composed with the identification of underlying sets is an additive bijection $R \to R$ with $f (a b) = f (a) \cdot_{op} f (b) = f (b) f (a)$ , i.e. an anti-automorphism, and conversely. For commutative $R$ , $id (a b) = a b = b a = id (b) id (a)$ .

Example(1.1)Quaternion conjugation

On $ℍ$ , the map $α = a + b i + c j + d k \mapsto \bar{α} = a - b i - c j - d k$ is additive, fixes $1$ , satisfies $\bar{\bar{α}} = α$ , and reverses products: $\bar{α β} = \bar{β} \bar{α}$ . Hence it is an involution and $ℍ ≅ ℍ^{op}$ . Consistently, $[ℍ]$ is the unique element of order $2$ in $Br (ℝ) ≅ ℤ / 2 ℤ$ .

Corollary—Transporting one-sided examples

If $R$ is left noetherian and not right noetherian, then $R^{op}$ is right noetherian and not left noetherian. The same holds with artinian, primitive, perfect, hereditary or Goldie in place of noetherian. One-sided counterexamples therefore always come in pairs, and only one of each pair needs to be constructed.

Worked Example

Opposites of triangular rings

Let $R, S$ be rings and $M$ an $(R, S)$ -bimodule, and set $A = (\begin{matrix} R & M \\ 0 & S \end{matrix})$ . Regard $M$ as an $(S^{op}, R^{op})$ -bimodule via $s^{op} \cdot m : = m s$ and $m \cdot r^{op} : = r m$ . Then

A^{op} ≅ (\begin{matrix} S^{op} & M \\ 0 & R^{op} \end{matrix}), (\begin{matrix} r & m \\ 0 & s \end{matrix}) ⟼ (\begin{matrix} s^{op} & m \\ 0 & r^{op} \end{matrix}) .

(E.1)

Verification is one line of matrix arithmetic: the product in $A$ of $(\begin{matrix} r' & m' \\ 0 & s' \end{matrix})$ then $(\begin{matrix} r & m \\ 0 & s \end{matrix})$ has corner entries $r' r$ , $s' s$ and middle entry $r' m + m' s$ , and the image matrices multiply to give exactly $(s' s)^{op}$ , $(r' r)^{op}$ and $s^{op} \cdot m' + m \cdot r'^{op} = m' s + r' m$ .

Small's example and its mirror

Take $R = ℚ$ , $S = ℤ$ , $M = ℚ$ as a $(ℚ, ℤ)$ -bimodule:

A = (\begin{matrix} ℚ & ℚ \\ 0 & ℤ \end{matrix}), A^{op} ≅ (\begin{matrix} ℤ & ℚ \\ 0 & ℚ \end{matrix}) .

(E.2)

By the triangular-ring criterion, $A$ is left noetherian: $ℚ$ and $ℤ$ are noetherian and $M = ℚ$ is noetherian as a left $ℚ$ -module, being one-dimensional. It is not right noetherian, because $M = ℚ$ is not noetherian as a right $ℤ$ -module — the chain $ℤ ⊊ \frac{1}{2} ℤ ⊊ \frac{1}{4} ℤ ⊊ \dots$ never stops.

The mirror ring $(\begin{matrix} ℤ & ℚ \\ 0 & ℚ \end{matrix})$ — Small's standard example — is therefore right noetherian and not left noetherian, and the conclusion required no new argument. Neither ring is artinian on either side, since $ℤ$ has the infinite descending chain $(2) ⊋ (4) ⊋ \dots$ .

A self-opposite triangular ring

Let $k$ be commutative and $T_{n} (k)$ the upper triangular $n \times n$ matrices. Let $J$ be the permutation matrix with $1$ s on the anti-diagonal. Then $A \mapsto J A^{t} J$ is an anti-automorphism of $M_{n} (k)$ that carries upper triangular matrices to upper triangular matrices, because conjugation by $J$ reverses the index order. Hence $T_{n} (k) ≅ T_{n} (k)^{op}$ , which is consistent with $T_{n} (k)$ being both left and right artinian.

Process and Workflow

Write the statement with all sides labelledReplace every occurrence of ideal, module and chain condition by its explicitly one-sided form.

Apply the theorem to $R^{op}$ Every left-handed hypothesis about

R^{op}

is a right-handed hypothesis about

R

Translate the conclusion backA conclusion about left ideals of

R^{op}

is a conclusion about right ideals of

R

Ask whether $R ≅ R^{op}$ Only if yes may the mirrored conclusion be combined with the original as two facts about the same ring.

Is the property you care about side-neutral?

Yes — two-sided by definitionSimplicity, primeness, semiprimeness and every statement about two-sided ideals transfer without comment.

Yes — by a theoremThe Jacobson radical, semisimplicity and von Neumann regularity are symmetric, but each needs its own proof. Cite the theorem, do not appeal to the opposite ring alone.

NoNoetherian, artinian, primitive, perfect, hereditary and Goldie all fail. State the side, and expect the mirror example to exist.

UnknownTreat as one-sided until proved otherwise, and record which side your hypotheses live on.

Comparison and Classification

Side-neutral or not
Property	Side-neutral?	Reason or witness
Simple	yes	defined by two-sided ideals
Prime, semiprime	yes	defined by two-sided ideals
Jacobson radical	yes	characterised by invertibility of $1 - x y z$ , a two-sided condition
Semisimple	yes	left semisimple $iff$ right semisimple; Wedderburn–Artin form is self-opposite
Dedekind-finite	yes	$a b = 1 \Rightarrow b a = 1$ is unchanged by reversing the product
Von Neumann regular	yes	the condition $a \in a R a$ is its own mirror
Noetherian	no	$(\begin{matrix} ℚ & ℚ \\ 0 & ℤ \end{matrix})$
Artinian	no	$(\begin{matrix} ℝ & ℝ \\ 0 & ℚ \end{matrix})$
Primitive	no	Bergman's example of a left primitive ring that is not right primitive
Perfect	no	left perfect and right perfect are independent
Hereditary	no	independent, by an example of Small

How standard constructions interact with the opposite
	Self-opposite?	Explicit anti-automorphism	Fails when
Commutative ring $R$	yes	identity	never
$M_{n} (R)$ , $R$ commutative	yes	transpose	never
$M_{n} (R)$ , $R$ general	partial	transpose into $M_{n} (R^{op})$	$R not ≅ R^{op}$
Group ring $k G$ , $k$ commutative	yes	$g \mapsto g^{- 1}$	never
Real quaternions $ℍ$	yes	conjugation	never
Central division algebra $D / F$	partial	exists iff $[D]$ has order $\leq 2$	$[D]$ of order $\geq 3$ in $Br (F)$
Triangular $(\begin{matrix} R & M \\ 0 & S \end{matrix})$	partial	swap the corners	the two module structures on $M$ differ

How standard constructions interact with the opposite

Relationship Map

What the opposite ring fixes, mirrors and destroys.

$R \mapsto R^{op}$ — an involutive operation on the class of rings
- leaves unchanged
  - the additive group and the underlying set
  - the centre $Z (R)$ and the unit group $U (R)$ as a group under reversed product
  - the lattice of two-sided ideals
  - $rad R$ , the prime radical, and simplicity
- swaps
  - left ideals with right ideals
  - left modules with right modules
  - left noetherian with right noetherian
  - left primitive with right primitive
- can genuinely change the isomorphism class
  - central division algebras of Brauer order at least three
  - certain triangular rings with asymmetric bimodules

R

commutative

R

has an involution

R ≅ R^{op}

left theory

=

right theory for

R

None of these arrows reverses. A ring can be self-opposite without carrying an involution, and left theory can accidentally match right theory for a ring that is not self-opposite.

Failure Modes and Common Mistakes

Do not write $a^{op} b^{op} = (a b)^{op}$ . The whole point is the reversal.
Do not assume $Hom$ and $\otimes$ keep their sides. ${Hom}_{R} (M, N)$ for left modules is naturally a right module over $End (_{R} M)$ , hence carries an $op$ if you insist on writing it on the left.
Do not treat $R \times R^{op}$ as $R \otimes R^{op}$ . The enveloping algebra of an $F$ -algebra is the tensor product, and it is what represents bimodules.

Best Practices

Prove the version whose argument is more natural, then state the mirror explicitly as a corollary obtained by applying the result to $R^{op}$ .
When constructing a counterexample, record its mirror in the same breath; readers otherwise reconstruct it needlessly.
Keep the superscript upright — $R^{op}$ , not $R^{op}$ — so it is not read as a product of variables.
For algebras over a commutative base $k$ , check that your anti-automorphism is $k$ -linear; a $k$ -semilinear one gives a different and weaker conclusion.
When a computer algebra system returns an endomorphism ring, determine whether it computed $End$ of a left or a right module before comparing with a hand calculation.

Quick Reference

Definition

a^{op} b^{op} = (b a)^{op}

, same addition

Involutive

(R^{op})^{op} = R

Modules

_{R} Mod = {Mod}_{R^{op}}

Endomorphisms

End (_{R} R) ≅ R^{op}

Matrices

M_{n} (R)^{op} ≅ M_{n} (R^{op})

by transpose

Self-opposite

R ≅ R^{op}

iff

an anti-automorphism exists

Brauer

[D^{op}] = - [D]

;

D \otimes_{F} D^{op} ≅ {End}_{F} (D)

Centre

Z (R^{op}) = Z (R)

Mirror pairs to keep in mind
Left-hand notion	Right-hand notion	Coincide?
Maximal left ideal	Maximal right ideal	no, but their intersections agree
Left artinian	Right artinian	no
Left primitive	Right primitive	no
Left semisimple	Right semisimple	yes
Left zero-divisor	Right zero-divisor	no
Left inverse	Right inverse	only in Dedekind-finite rings

Frequently Asked Questions

R^{op}

ever equal to

R

, rather than merely isomorphic?

Yes, and exactly when $R$ is commutative. Then the identity map on the underlying set is a ring isomorphism, so the two rings are the same ring with the same multiplication. For noncommutative $R$ the identity map is never a homomorphism, though some other bijection may be.

R ≅ R^{op}

, does that mean every left theorem holds on the right?

For that particular ring, yes — any left-handed property $R$ enjoys is enjoyed on the right, because the isomorphism transports it. It says nothing about other rings, and it does not make the property side-neutral as a general notion.

Why does the Jacobson radical come out symmetric while primitivity does not?

Because the radical has a characterisation that never mentions a side: $y \in rad R$ if and only if $1 - x y z$ is a unit for all $x, z$ . Invertibility is two-sided, so the resulting set is the same computed either way. Primitivity is defined by the existence of a faithful simple left module, and no side-free reformulation exists — Bergman's example shows none can.

How does the opposite ring interact with tensor products of algebras?

For $F$ -algebras, $(A \otimes_{F} B)^{op} ≅ A^{op} \otimes_{F} B^{op}$ . The important consequence is in Brauer theory: for a central simple $F$ -algebra $A$ of dimension $n^{2}$ , $A \otimes_{F} A^{op} ≅ M_{n} (F) ≅ {End}_{F} (A)$ , which is why $[A^{op}]$ inverts $[A]$ in the Brauer group.

Does a group ring always equal its opposite?

For a commutative coefficient ring $k$ and any group $G$ , the $k$ -linear extension of $g \mapsto g^{- 1}$ is an involution of $k G$ , so $k G ≅ (k G)^{op}$ . The computation needs $k$ commutative: the coefficients must be allowed to swap past one another when the product is reversed.

Is there a ring with no anti-automorphism at all?

Yes. Any central division algebra whose Brauer class has order greater than two — for instance the invariant $1 / 3$ algebra over a $p$ -adic field — fails to be isomorphic to its opposite, and an anti-automorphism would supply such an isomorphism.

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §1 (pp. 3–5), with the primitivity example in §11.
N. Jacobson, Basic Algebra II, 2nd edition, W. H. Freeman, 1989, Chapters 3 and 4.
F. W. Anderson and K. R. Fuller, Rings and Categories of Modules, 2nd edition, Graduate Texts in Mathematics 13, Springer-Verlag, 1992, §4.
L. W. Small, “An example in Noetherian rings”, Proceedings of the National Academy of Sciences USA 54 (1965), 1035–1036.
M.-A. Knus, A. Merkurjev, M. Rost and J.-P. Tignol, The Book of Involutions, American Mathematical Society Colloquium Publications 44, 1998, Chapter I.

AI Suggested Questions

Give a concrete presentation of a division algebra of degree 3 that is not isomorphic to its opposite.
Prove that a ring is left semisimple if and only if it is right semisimple.
Sketch Bergman's construction of a left primitive ring that is not right primitive.
How does the opposite ring interact with Morita equivalence, and is Morita equivalence side-neutral?
Which finite-dimensional algebras over a field admit an involution, and how does Albert's classification organise them?
Show that ${Hom}_{R} (M, N)$ is a right $End (_{R} M)$ -module and explain where the opposite ring hides.
Are left perfect and right perfect genuinely independent, and what is the standard separating example?

Opposite Rings and Left–Right Duality

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

What the opposite ring is for

Modules change sides, functors do not

When is $R$ its own opposite?

Key Results

Worked Example

Opposites of triangular rings

Small's example and its mirror

A self-opposite triangular ring

Process and Workflow

Comparison and Classification

Relationship Map

Failure Modes and Common Mistakes

Best Practices

Quick Reference

Frequently Asked Questions

References

AI Suggested Questions

Continue learning

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

What the opposite ring is for

Modules change sides, functors do not

When is R its own opposite?

06Key Results

07Worked Example

Opposites of triangular rings

Small's example and its mirror

A self-opposite triangular ring

08Process and Workflow

09Comparison and Classification

10Relationship Map

11Failure Modes and Common Mistakes

12Best Practices

13Quick Reference

14Frequently Asked Questions

15Related KEVOS Topics

16References

17AI Suggested Questions

Continue learning

Executive Summary

Overview

Learning Objectives

Definitions

Core Concepts

When is $R$ its own opposite?

Key Results

Worked Example

Process and Workflow

Comparison and Classification

Relationship Map

Failure Modes and Common Mistakes

Best Practices

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions