The Brauer–Albert Theorem on Conjugate Maximal Subfields

Q: Why can a division algebra never be generated by one element?

The F-subalgebra generated by a single element a consists of polynomials in a with central coefficients, and any two such expressions commute. So a one-generated subalgebra is commutative, and equals D only if D is a field. Two generators is therefore the minimum, and (15.17) achieves it.

Q: Where does separability actually get used?

Only to apply the primitive element theorem and write K = F() with a single of degree r. Everything after that is linear algebra. In characteristic zero separability is automatic, so the reliance on (15.12) is invisible; in characteristic p it is essential.

Q: Is the choice of canonical?

No. Cyclic vectors are plentiful — they form the complement of a proper closed subset — and different choices give genuinely different bases and different conjugates '. The theorem is an existence statement about a rich supply of bases, not a normal form.

Q: Does (15.17) imply that D is a crossed product?

No. Crossed product structure requires a maximal subfield Galois over F, and Amitsur showed that central division algebras without one exist. The Brauer–Albert theorem needs only separability, which is always available, so it cannot imply the stronger statement.

Q: How does this compare with the fact that any finite-dimensional algebra has a finite generating set?

Trivially any F-basis generates. The content here is the shape: exactly two generators, both with the same separable minimal polynomial of degree r, and a basis whose elements are indexed by two exponents in the range 0 to r-1. Cardinality is not the point; structure is.

Q: Can the argument be run with left multiplication instead of right?

Yes, symmetrically: left multiplication by is a right K-linear operator on D_K, of the same dimension r by (15.8), and the same cyclic vector argument applies. The resulting basis is the mirror image and the two versions are exchanged by passing to D^op.

Executive Summary

Let $D$ be a division ring of dimension $r^{2}$ over its centre $F$ . The Brauer–Albert theorem produces two elements $α, β \in D$ whose $r^{2}$ products $α^{i} β α^{j}$ , for $0 \leq i, j < r$ , form an $F$ -basis of $D$ . Multiplying that basis on the right by $β^{- 1}$ converts it into ${α^{i} (α')^{j}}$ with $α' = β α β^{- 1}$ conjugate to $α$ .

The consequence is the noncommutative analogue of the primitive element theorem: a central division algebra is generated over its centre by two conjugate elements. The proof is linear algebra applied to right multiplication by a primitive element of a separable maximal subfield — the operator has minimal polynomial of degree $r$ on a space of dimension $r$ , hence admits a cyclic vector, and that vector is $β$ .

$r^{2}$ Size of the basis

$2$ Generators needed

ConjugateRelation between them

$(15.12)$ Prerequisite: separable maximal subfield

Overview

For a finite separable field extension the primitive element theorem gives a single generator and the clean basis $1, α, \dots, α^{r - 1}$ . A noncommutative division algebra cannot be generated by one element — a single element generates a commutative subfield — so the honest question is how few generators are needed and how explicit a basis can be made.

D = ⨁_{0 \leq i, j < r} F α^{i} β α^{j} = ⨁_{0 \leq i, j < r} F α^{i} (α')^{j}, α' = β α β^{- 1} .

(15.16)–(15.17)

Two bases of the same algebra; the second is obtained from the first by right multiplication by $β^{- 1}$ .

The answer is as good as possible: two, and they may be taken conjugate. The construction is not abstract nonsense — every step is a computation with a primitive element $α$ of a separable maximal subfield $K = F (α)$ , supplied by Existence of Separable Maximal Subfields, and a cyclic vector for right multiplication by $α$ .

A basis is not a presentation: knowing ${α^{i} β α^{j}}$ spans $D$ says nothing directly about how two basis elements multiply. What the theorem delivers is generation and a dimension-matched spanning set, which is exactly what is needed to bound the complexity of a presentation.

Learning Objectives

State $(15.16)$ and $(15.17)$ with the hypothesis $\dim_{F} D = r^{2}$ , $F = Z (D)$ .
Explain why $D$ is a faithful module over $K \otimes_{F} K$ .
Show $1, T, \dots, T^{r - 1}$ are $K$ -independent for $T$ right multiplication by $α$ .
Apply the cyclic vector criterion to obtain $β$ .
Derive the conjugate-generator basis by right multiplication by $β^{- 1}$ .
Carry out the construction concretely in $ℍ_{ℚ}$ and check the resulting basis.

Definitions

Standing setup: $D$ is a division ring with centre $F$ and $\dim_{F} D = r^{2} < \infty$ ; $K = F (α)$ is a maximal subfield separable over $F$ , with $α$ of minimal polynomial $f (t) \in F [t]$ of degree $r$ .
$_{K} D$: $D$ viewed as a left $K$ -vector space, of dimension $r$ by $(15.8)$ .
$T$: The operator $T (v) = v α$ of right multiplication by $α$ . It is left $K$ -linear because left and right multiplications commute.
Cyclic operator: A linear operator on a finite-dimensional space whose minimal polynomial has degree equal to the dimension; equivalently one admitting a cyclic vector.
$K \otimes_{F} K$: The commutative $F$ -algebra $⨁_{i = 0}^{r - 1} K \otimes α^{i}$ , acting on $D$ by $(a \otimes b) \cdot v = a v b$ .

The action of $K \otimes_{F} K$ on $D$ is the restriction of the $D \otimes_{F} K$ -module structure of (15.3); since $K$ is commutative, no opposite ring is needed.

Core Concepts

One operator, two roles

The element $α$ appears twice in the argument, in different capacities. On the left it generates the maximal subfield $K$ , so it supplies the scalars. On the right it acts as the operator $T$ , whose powers generate the second index of the basis. The two roles commute — that is the associative law — which is why the products $α^{i} β α^{j}$ are unambiguous.

α

on the left: scalars

K = F (α)

α

on the right: operator

T

β

: cyclic vector for

T

Basis

α^{i} β α^{j}

Faithfulness supplies independence

A relation $\sum_{i} a_{i} T^{i} = 0$ with $a_{i} \in K$ says $\sum_{i} a_{i} v α^{i} = 0$ for every $v \in D$ , i.e. that the element $\sum_{i} a_{i} \otimes α^{i}$ of $K \otimes_{F} K$ annihilates $D$ . Since $D$ is a faithful $D \otimes_{F} K$ -module by $(15.3)$ , and $K \otimes_{F} K$ embeds in $D \otimes_{F} K$ , that element is zero. The decomposition $K \otimes_{F} K = ⨁_{i < r} K \otimes α^{i}$ then forces every $a_{i} = 0$ .

So $T$ has minimal polynomial exactly $f$ , of degree $r$ , on a space $_{K} D$ of dimension $r$ . That is the numerical coincidence the proof lives on.

Cyclic vectors

For a linear operator on an $n$ -dimensional space, the minimal polynomial has degree $n$ if and only if some vector $β$ has $β, T β, \dots, T^{n - 1} β$ linearly independent — a cyclic vector, or in numerical language a starting vector whose Krylov sequence is a basis. Applying this to $T$ on $_{K} D$ gives $β \in D$ with $D = ⨁_{j < r} K β α^{j}$ , and expanding $K = ⨁_{i < r} F α^{i}$ finishes the count.

Key Results

Theorem(15.16)Brauer–Albert basis theorem

Let $D$ be a division ring of dimension $r^{2}$ over its centre $F$ . Then there exist $α, β \in D$ such that the $r^{2}$ elements

{α^{i} β α^{j} : 0 \leq i, j < r}

form an $F$ -basis of $D$ . One may take $α$ to be a primitive element of any maximal subfield of $D$ separable over $F$ .

Proof

**Choice of $α$ .** By $(15.12)$ there is a maximal subfield $K \subseteq D$ separable over $F$ , and by $(15.8)$ $\dim_{F} K = r$ . Separability and the primitive element theorem give $K = F (α)$ with $f (t) \in F [t]$ the minimal polynomial of $α$ , of degree $r$ .

The module structure. As in $(15.3)$ — with $K^{op}$ identified with $K$ , since $K$ is commutative — $D$ is a faithful left module over $D \otimes_{F} K$ via $(d \otimes a) \cdot v = d v a$ . Restricting along the injection $K \otimes_{F} K ↪ D \otimes_{F} K$ makes $D$ a faithful $K \otimes_{F} K$ -module.

The operator. Let $T : D \to D$ , $T (v) = v α$ . Since left and right multiplications commute, $T$ is a left $K$ -linear endomorphism of $_{K} D$ , a $K$ -space of dimension $r$ by $(15.8)$ . Because the coefficients of $f$ lie in $F \subseteq Z (D)$ , $f (T) (v) = v f (α) = 0$ , so $f (T) = 0$ .

Independence of the powers. Suppose $\sum_{i = 0}^{r - 1} a_{i} T^{i} = 0$ with $a_{i} \in K$ . Evaluating at $v \in D$ gives $\sum_{i} a_{i} v α^{i} = 0$ for all $v$ , i.e. the element $\sum_{i} a_{i} \otimes α^{i} \in K \otimes_{F} K$ annihilates $D$ . Faithfulness makes it zero, and since $K \otimes_{F} K = ⨁_{i = 0}^{r - 1} K \otimes α^{i}$ — the elements $1, α, \dots, α^{r - 1}$ being an $F$ -basis of the right-hand factor — all $a_{i} = 0$ .

A cyclic vector. Hence the minimal polynomial of $T$ on $_{K} D$ is $f$ , of degree $r = \dim (_{K} D)$ . By the standard criterion of linear algebra, $T$ then has a cyclic vector: there is $β \in D$ with $β, T β, \dots, T^{r - 1} β$ a $K$ -basis of $_{K} D$ , that is

D = ⨁_{j = 0}^{r - 1} K β α^{j} .

Conclusion. Substituting $K = ⨁_{i = 0}^{r - 1} F α^{i}$ gives $D = \sum_{0 \leq i, j < r} F α^{i} β α^{j}$ . This is a spanning set of $r^{2}$ elements in an $F$ -space of dimension $r^{2}$ , so it is a basis.

Corollary(15.17)Generation by two conjugate elements

Let $D$ be a division ring of dimension $r^{2}$ over its centre $F$ . Then there exist conjugate elements $α, α' \in D$ — that is, $α' = β α β^{- 1}$ for some $β \in D^{\times}$ — such that ${α^{i} (α')^{j} : 0 \leq i, j < r}$ is an $F$ -basis of $D$ . In particular $D$ is generated as an $F$ -algebra by the two conjugate elements $α$ and $α'$ .

Proof

Take $α, β$ from $(15.16)$ ; $β \neq 0$ , so $β^{- 1}$ exists. Right multiplication by $β^{- 1}$ is an $F$ -linear bijection of $D$ , hence carries the basis ${α^{i} β α^{j}}$ to another basis. Its members are

α^{i} β α^{j} β^{- 1} = α^{i} (β α β^{- 1})^{j} = α^{i} (α')^{j}, α' : = β α β^{- 1},

using $β α^{j} β^{- 1} = (β α β^{- 1})^{j}$ . Since these $r^{2}$ elements form a basis, the $F$ -subalgebra they generate is all of $D$ , and that subalgebra is generated by $α$ and $α'$ .

Remark—What conjugacy does and does not mean

$α$ and $α'$ have the same minimal polynomial over $F$ , since conjugation is an $F$ -algebra automorphism of $D$ . So $D$ is generated by two elements with the same minimal polynomial, both generating separable maximal subfields. This does not make $D$ a cyclic algebra or a crossed product: no assumption is made, and none is obtained, about the Galois behaviour of $F (α) / F$ .

Proof Techniques and Method

The reusable moves behind these proofs.

Move 1

Turn multiplication into an operator

Right multiplication by $α$ on $_{K} D$ is a $K$ -linear map because left and right multiplications commute. This converts a question about products in $D$ into a question about a single matrix over $K$ .

Move 2

Use faithfulness to bound the minimal polynomial from below

The minimal polynomial cannot drop below degree $r$ , because a lower-degree relation would give a nonzero annihilator in $K \otimes_{F} K$ . Faithfulness of the module is the only input.

Move 3

Convert a basis by an invertible right factor

Right multiplication by any unit permutes bases. Choosing the unit to be $β^{- 1}$ turns a two-sided pattern $α^{i} β α^{j}$ into a product of powers of two conjugates.

Move 2 is worth isolating: minimal polynomial degree $\geq r$ plus $\dim = r$ forces equality and cyclicity, so the seemingly weak faithfulness statement is doing all the work that a dimension count usually does.

Worked Example

The rational quaternions, worked end to end

Let $D = ℍ_{ℚ}$ , $F = ℚ$ , $\dim_{F} D = 4$ , so $r = 2$ . Take the separable maximal subfield $K = ℚ (i)$ with primitive element $α = i$ and minimal polynomial $f (t) = t^{2} + 1$ . As a left $K$ -space, $D = K \cdot 1 \oplus K \cdot j$ , of dimension $2$ .

The operator is $T (v) = v i$ . In the ordered left $K$ -basis $(1, j)$ , note $j i = - i j = (- i) j$ , so $T (1) = i \cdot 1$ and $T (j) = j i = (- i) \cdot j$ : the matrix of $T$ is $diag (i, - i)$ , with minimal polynomial $t^{2} + 1 = f$ , of degree $2 = \dim (_{K} D)$ , as the theorem predicts.

Finding a cyclic vector. Not every $β$ works: for $β = 1$ we get $T β = i = i \cdot 1$ , dependent with $β$ ; for $β = j$ we get $T β = (- i) \cdot j$ , again dependent. Take $β = 1 + j$ , with coordinates $(1, 1)$ ; then $T β = (1 + j) i = i + j i = i \cdot 1 + (- i) \cdot j$ , coordinates $(i, - i)$ , and the determinant $1 \cdot (- i) - 1 \cdot i = - 2 i \neq 0$ . So $β = 1 + j$ is cyclic.

The Brauer–Albert basis ${α^{i} β α^{j}}$ for $α = i$ , $β = 1 + j$
$(i, j)$	Element	Value in $1, i, j, k$
$(0, 0)$	$β$	$1 + j$
$(0, 1)$	$β α$	$i - k$
$(1, 0)$	$α β$	$i + k$
$(1, 1)$	$α β α$	$- 1 + j$

The computations use $i j = k$ , $j i = - k$ , $k i = j$ : $β α = (1 + j) i = i - k$ ; $α β = i (1 + j) = i + k$ ; $α β α = (i + k) i = i^{2} + k i = - 1 + j$ . The four coordinate vectors $(1, 0, 1, 0)$ , $(0, 1, 0, - 1)$ , $(0, 1, 0, 1)$ , $(- 1, 0, 1, 0)$ are visibly $ℚ$ -independent, so this is a basis.

Passing to conjugate generators

Here $β^{- 1} = \frac{1}{2} (1 - j)$ , since $(1 + j) (1 - j) = 1 - j^{2} = 2$ . Then

α' = β α β^{- 1} = \frac{1}{2} (i - k) (1 - j) = \frac{1}{2} (- 2 k) = - k,

(E.1)

A conjugate of $i$ ; indeed $(- k)^{2} = - 1$ , the same minimal polynomial $t^{2} + 1$ .

Right-multiplying each basis element by $β^{- 1}$ gives $1$ , $- k$ , $i$ and $j$ respectively — check the last one: $(- 1 + j) \cdot \frac{1}{2} (1 - j) = \frac{1}{2} (- 1 + j + j - j^{2}) = \frac{1}{2} (2 j) = j$ . So the conjugate-generator basis is

{α^{i} (α')^{j}}_{0 \leq i, j < 2} = {1, - k, i, i (- k)} = {1, - k, i, j},

(E.2)

Using $i k = - j$ , so $i (- k) = j$ . This is manifestly a $ℚ$ -basis of $ℍ_{ℚ}$ .

Process and Workflow

Find a separable maximal subfieldGuaranteed by

(15.12)

; in characteristic zero any maximal subfield will do.

Take a primitive elementSeparability plus the primitive element theorem give

K = F (α)

with

\deg f = r

Form the operator $T (v) = v α$ Left

K

-linear on

_{K} D

, of dimension

r

; its minimal polynomial is

f

Find a cyclic vector $β$ Test candidates by checking that

β, β α, \dots, β α^{r - 1}

are left

K

-independent. Most elements work;

β \in K

never does.

Read off the basis

{α^{i} β α^{j}}

, or after right multiplication by

β^{- 1}

{α^{i} (α')^{j}}

with

α' = β α β^{- 1}

Your candidate $β$ fails to be cyclic. What now?

$β \in K$ Guaranteed failure: then all

β α^{j}

lie in

K

, a

1

-dimensional left

K

-subspace of

D

. Choose

β

outside

K

$β \notin K$ but still dependentPossible when

r > 2

. Perturb: cyclic vectors form a Zariski-dense set, so a random modification succeeds with high probability.

Every candidate failsImpossible — it would contradict the minimal polynomial having degree

r

. Recheck that

α

generates a maximal subfield and that

\deg f = r

Comparison and Classification

Bases and generators of a central division algebra of degree $r$
Description	Basis shape	Hypothesis needed	Gives a presentation?
Primitive element theorem (commutative case)	$1, α, \dots, α^{r - 1}$	$K / F$ finite separable	yes
Cyclic algebra	$α^{i} u^{j}$ with $u a u^{- 1} = σ (a)$	$K / F$ cyclic Galois inside $D$	yes
Crossed product	$α^{i} u_{σ}$ , $σ \in Gal (K / F)$	$K / F$ Galois inside $D$	yes
Brauer–Albert $(15.16)$	$α^{i} β α^{j}$	$\dim_{F} D = r^{2}$ only	no
Conjugate form $(15.17)$	$α^{i} (α')^{j}$ , $α' \sim α$	$\dim_{F} D = r^{2}$ only	no

What each construction assumes and delivers
	Always available	Two generators	Structure constants explicit
Brauer–Albert basis	yes	yes	no
Conjugate generators	yes	yes	no
Cyclic algebra presentation	no	yes	yes
Crossed product presentation	no	no	yes

What each construction assumes and delivers

Relationship Map

This is the terminal result of §15: it consumes almost everything proved earlier in the section.

$(15.1)$ — tensor products and centralizersSimplicity of

D \otimes_{F} K

$(15.3)$ — $D$ is a faithful $D \otimes_{F} K$ -moduleSupplies the faithfulness used for independence of

1, T, \dots, T^{r - 1}

$(15.8)$ — $\dim (_{K} D) = r$ , $\dim_{F} D = r^{2}$ Supplies the dimension count

$(15.12)$ — a separable maximal subfield existsSupplies the primitive element

α

; then

(15.16)

and

(15.17)

(15.12)

separable

K

K = F (α)

T

cyclic on

_{K} D

(15.16)

basis

(15.17)

conjugate generators

Applications and Industry Use

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

Structure theory

A noncommutative primitive element theorem

The result bounds how complicated a central division algebra can be to describe: two elements suffice, and they can be taken with the same minimal polynomial. Generation statements of this kind underpin the theory of generic division algebras.

Computer algebra

Compact representation

Storing an algebra by two generators and their relations is far cheaper than a full $r^{2} \times r^{2} \times r^{2}$ structure-constant table. Systems that construct division algebras exploit exactly this economy.

Numerical linear algebra

The Krylov analogy

The cyclic vector step is the algebraic form of a Krylov subspace computation: iterate an operator on a starting vector until the images span. The same non-degeneracy condition governs convergence of Krylov methods.

Coding theory

Generating a codebook

Space–time codes built from a division algebra transmit $F$ -linear combinations of basis elements. A basis indexed by two exponents, as here, translates into a codeword parametrisation with two independent index ranges.

Computational Notes

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

Given structure constants for $D$ over $F$ and a primitive element $α$ , building the matrix of $T$ in a left $K$ -basis costs $O (r^{3})$ operations in $K$ ; testing a candidate $β$ is one rank computation on an $r \times r$ matrix over $K$ .
Cyclic vectors are generic: the set of non-cyclic vectors is a proper Zariski-closed subset, so over an infinite field a random $β$ succeeds with probability $1$ in the algebraic sense, and with high probability in any sufficiently large finite sample.
The basis ${α^{i} β α^{j}}$ does not come with multiplication rules; converting it to structure constants requires $r^{4}$ products, each expanded in the basis, so the compact description is cheap to store and expensive to unfold.
In characteristic $p$ the first step — finding a separable maximal subfield — is the expensive one, since random search may repeatedly return inseparable or small-degree elements. $(15.12)$ guarantees the search terminates in principle but gives no bound.

Failure Modes and Common Mistakes

Do not assume $β$ can be chosen in $K$ ; elements of $K$ are never cyclic vectors for $T$ , since $K$ is a one-dimensional left $K$ -subspace.
Do not confuse the two occurrences of $α$ : the left exponent indexes scalars from $K$ , the right exponent indexes powers of the operator $T$ .
Do not expect the result for algebras that are merely simple rather than division rings without checking the dimension hypothesis; the count $r^{2}$ is what matches the spanning set to the dimension.
Do not read $(15.17)$ as a uniqueness statement — many pairs of conjugate generators exist, and none is canonical.

Quick Reference

Hypothesis

D

a division ring,

F = Z (D)

\dim_{F} D = r^{2}

$(15.16)$

{α^{i} β α^{j} : 0 \leq i, j < r}

is an

F

-basis

$(15.17)$

{α^{i} (α')^{j}}

is an

F

-basis with

α' = β α β^{- 1}

Generation

D = F ⟨ α, α' ⟩

with

α, α'

conjugate

Choice of $α$ primitive element of a separable maximal subfield,

\deg f = r

Choice of $β$ cyclic vector for

T (v) = v α

_{K} D

Key decomposition

K \otimes_{F} K = ⨁_{i < r} K \otimes α^{i}

Quaternion instance

α = i

β = 1 + j

α' = - k

, basis

{1, - k, i, j}

Proof skeleton
Step	Content
1	$(15.12)$ and the primitive element theorem give $K = F (α)$ separable, $\deg f = r$
2	$D$ is a faithful $K \otimes_{F} K$ -module, by restriction from $(15.3)$
3	$T (v) = v α$ is $K$ -linear on $_{K} D$ with minimal polynomial $f$ of degree $r = \dim (_{K} D)$
4	$T$ has a cyclic vector $β$ ; expanding $K$ over $F$ gives the $r^{2}$ basis elements
5	Right multiplication by $β^{- 1}$ converts the basis to ${α^{i} (α')^{j}}$

Frequently Asked Questions

Why can a division algebra never be generated by one element?

The $F$ -subalgebra generated by a single element $a$ consists of polynomials in $a$ with central coefficients, and any two such expressions commute. So a one-generated subalgebra is commutative, and equals $D$ only if $D$ is a field. Two generators is therefore the minimum, and $(15.17)$ achieves it.

Where does separability actually get used?

Only to apply the primitive element theorem and write $K = F (α)$ with a single $α$ of degree $r$ . Everything after that is linear algebra. In characteristic zero separability is automatic, so the reliance on $(15.12)$ is invisible; in characteristic $p$ it is essential.

Is the choice of

β

canonical?

No. Cyclic vectors are plentiful — they form the complement of a proper closed subset — and different choices give genuinely different bases and different conjugates $α'$ . The theorem is an existence statement about a rich supply of bases, not a normal form.

Does

(15.17)

imply that

D

is a crossed product?

No. Crossed product structure requires a maximal subfield Galois over $F$ , and Amitsur showed that central division algebras without one exist. The Brauer–Albert theorem needs only separability, which is always available, so it cannot imply the stronger statement.

How does this compare with the fact that any finite-dimensional algebra has a finite generating set?

Trivially any $F$ -basis generates. The content here is the shape: exactly two generators, both with the same separable minimal polynomial of degree $r$ , and a basis whose elements are indexed by two exponents in the range $0$ to $r - 1$ . Cardinality is not the point; structure is.

Can the argument be run with left multiplication instead of right?

Yes, symmetrically: left multiplication by $α$ is a right $K$ -linear operator on $D_{K}$ , of the same dimension $r$ by $(15.8)$ , and the same cyclic vector argument applies. The resulting basis is the mirror image and the two versions are exchanged by passing to $D^{op}$ .

References

T. Y. Lam, A First Course in Noncommutative Rings, Graduate Texts in Mathematics 131, Springer-Verlag, 1991, §15, results (15.16)–(15.17) (pp. 259–260).
N. Jacobson, Structure of Rings, American Mathematical Society Colloquium Publications 37, 1956; revised edition 1964 — the treatment Lam's presentation follows.
A. A. Albert, Structure of Algebras, American Mathematical Society Colloquium Publications 24, 1939.
P. K. Draxl, Skew Fields, London Mathematical Society Lecture Note Series 81, Cambridge University Press, 1983.
N. Jacobson, Basic Algebra I, 2nd edition, W. H. Freeman, 1985, Chapter 3 (cyclic operators, minimal and characteristic polynomials).

AI Suggested Questions

Prove the linear algebra criterion that an operator with minimal polynomial of degree equal to the dimension has a cyclic vector.
Carry out the Brauer–Albert construction for a degree- $3$ cyclic algebra over $ℚ$ and write out the nine basis elements.
Show that the set of non-cyclic vectors for a cyclic operator is a proper Zariski-closed subset.
Give an example of a division algebra of degree $3$ and two conjugate generators for it, with their common minimal polynomial.
How would one convert a Brauer–Albert basis into explicit structure constants, and what is the cost?
What is the analogue of $(15.16)$ for central simple algebras that are not division rings?
Relate the cyclic vector step to the theory of Krylov subspaces and companion matrices.

01Executive Summary

02Overview

03Learning Objectives

04Definitions

05Core Concepts

One operator, two roles

Faithfulness supplies independence

Cyclic vectors

06Key Results

07Proof Techniques and Method

Turn multiplication into an operator

Use faithfulness to bound the minimal polynomial from below

Convert a basis by an invertible right factor

08Worked Example

The rational quaternions, worked end to end

Passing to conjugate generators

09Process and Workflow

10Comparison and Classification

11Relationship Map

12Applications and Industry Use

A noncommutative primitive element theorem

Compact representation

The Krylov analogy

Generating a codebook

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

Key Results

Proof Techniques and Method

Worked Example

Process and Workflow

Comparison and Classification

Relationship Map

Applications and Industry Use

Computational Notes

Failure Modes and Common Mistakes

Quick Reference

Frequently Asked Questions

Related KEVOS Topics

References

AI Suggested Questions