Alphabeta Math
How statement and proof provenance work

The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.

  • Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
  • AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
  • AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.

These labels describe origin, not correctness: citations and verification chips remain separate evidence.

✓ 21 results · all verified · 9 also independently AI-judged
Every result on this page is machine-checked by a proof checker and read in full and owner-audited; the judge is an additional, independent cross-model AI review of the proofs. The 12 not AI-judged were verified by owner audit (typically over a confirmed judge false positive), not failures.

Solvability by Radicals and Kummer Theory

1 · Prerequisites

2 · Summary

Finite Galois extensions, roots of unity, the Galois correspondence, solvable groups, and the earlier determinant, trace, and bilinear-form pages provide the background here. The page uses the published finite-field and cyclotomic results for roots of unity and cyclic Galois actions, the Galois page for Artin's theorem and quotient fields, and the solvable-group page for composition factors and closure properties. Those prerequisites let the field-theoretic norm and trace interact directly with linear algebra, then let cyclic and abelian Galois extensions be rewritten in radical form.

The development begins with norm, trace, and the trace form, including the inseparable exponent in the embedding formulas and the separability criterion for nondegeneracy. It then proves multiplicative and additive Hilbert 90, uses the Lagrange resolvent to characterize cyclic Kummer extensions, adds the characteristic-p Artin-Schreier analogue, and packages the resulting abelian theory through the Kummer pairing and correspondence. The last block defines radical extensions and solvability by radicals, proves the two directions of Galois's solvability theorem in characteristic 0, records the solvability of the low-degree symmetric groups, and ends with the general-polynomial form of Abel-Ruffini.

3 · Logical flowchart

4 · Definitions, theorems and proofs

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The norm NK/F and trace Tr⁡K/F of a finite field extension

Definition

Let K/F be a finite field extension, so K is a finite-dimensional F-vector space of degree [K:F] (The degree [K:F]=dim⁡FK of a finite field extension). For a∈K, let

ma ⁣:K→K,x↦ax,

the F-linear operator of multiplication by a.

The norm and trace of a from K to F are

NK/F(a):=det⁡(ma),Tr⁡K/F(a):=tr⁡(ma),

where determinant and trace are those of the published linear-operator notions (The determinant of an endomorphism of a finite-dimensional vector space: its matrix determinant in an ordered basis in positive dimension, and 1 on the zero space, The basis-independent trace of an endomorphism of a finite-dimensional vector space).

Because K is a field, ma is the zero operator exactly when a=0 and is an automorphism exactly when a≠0. Later items identify these two quantities with the classical embedding formulas and the trace form.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

Norm and trace from embeddings, with the inseparable exponent in the norm formula

Statement

Let K/F be a finite field extension, let Ω/F be an algebraic closure, let Σ=Hom⁡F(K,Ω), and let [K:F]i be the inseparable degree (The inseparable degree [K:F]i=[K:F]/[K:F]s of a finite extension). Then for every a∈K,

Tr⁡K/F(a)=[K:F]i∑σ∈Σσ(a),

and

NK/F(a)=(∏σ∈Σσ(a))[K:F]i.

In particular, when K/F is separable these are the ordinary sum and product over the distinct F-embeddings of K into Ω; and when [K:F]i>1 in characteristic p>0, the trace map is identically zero because [K:F]i is a power of p.

Facts & Assumptions

Given: A finite extension K/F, an algebraic closure Ω/F, the set Σ=Hom⁡F(K,Ω) of F-embeddings, an element a∈K, the separable closure Ks of F in K, and the inseparable degree [K:F]i.

[F1]

Norm and trace are defined from the multiplication operator ma ⁣:x↦ax on the finite-dimensional F-vector space K (The norm NK/F and trace Tr⁡K/F of a finite field extension).

[L1]

The distinct automorphisms of a field are linearly independent after restriction to the multiplicative group, and the same evaluation-matrix argument applies to the distinct F-embeddings of a finite separable extension (Dedekind's linear independence theorem for distinct characters).

[L3]

If Ks is the separable closure of F in K, then [K:F]s=[Ks:F] and K/Ks is purely inseparable (For a finite extension, [K:F]s=[Ks:F], An algebraic extension is purely inseparable over its separable closure).

[L4]

For a purely inseparable extension, the inclusion into an algebraic closure is the only embedding over the base field (Pure inseparability and its conjugate, embedding, and separable-degree criteria).

Proof

technique · direct
1.1L1L2choose

Suppose first that K/F is separable. Choose an F-basis u1,…,un of K, list the embeddings as σ1,…,σn, and form the evaluation matrix A=(σi(uj))i,j. By the same argument used in Artin's fixed-field lower bound, [L1] makes A invertible.

1.2L2L3L4

For general K/F, let Ks be the separable closure of F in K. By [L3], the extension K/Ks is purely inseparable of degree [K:F]i, and the F-embeddings of K are exactly the extensions of the embeddings of Ks, one extension for each embedding because [L4] gives uniqueness over Ks. Thus Σ may be identified with Hom⁡F(Ks,Ω), and its cardinality is [K:F]s.

2.1F1step 1.1algebra

Let M be the matrix of ma in the basis u1,…,un. Because σi(auj)=σi(a)σi(uj) for every i,j, one has AM=diag⁡(σ1(a),…,σn(a))A. Hence M=A−1diag⁡(σ1(a),…,σn(a))A, so determinant and trace of [F1] give NK/F(a)=∏iσi(a),Tr⁡K/F(a)=∑iσi(a).

3.1F1step 2.1step 1.2algebra

Choose an F-basis v1,…,vs of Ks and a Ks-basis b1,…,bi of K, where i=[K:F]i. Writing abr=∑q=1icqrbq(cqr∈Ks), the matrix of ma on the product basis (bqvj) is the block matrix C♯=(M(cqr))q,r, where M(cqr) is the matrix of multiplication by cqr on Ks/F. Conjugating each block by the separable evaluation matrix of step 1.1 for Ks/F turns C♯ into a block-diagonal matrix with diagonal blocks σ(C) as σ runs through Σ. Therefore NK/F(a)=∏σ∈Σdet⁡(σ(C)),Tr⁡K/F(a)=∑σ∈Σtr⁡(σ(C)).

4.1step 3.1L3L4algebra

Over Ks, the extension K/Ks is purely inseparable of degree i, so every conjugate of a over Ks equals a. Accordingly the characteristic polynomial of the Ks-linear operator represented by C is (X−a)i, and therefore det⁡(C)=ai,tr⁡(C)=ia. Applying each σ∈Σ in step 3.1 gives det⁡(σ(C))=σ(a)i,tr⁡(σ(C))=i σ(a).

5.1step 3.1step 4.1algebra

Substituting step 4.1 into step 3.1 yields NK/F(a)=∏σ∈Σσ(a)i=(∏σ∈Σσ(a))i, and Tr⁡K/F(a)=∑σ∈Σi σ(a)=i∑σ∈Σσ(a). This is the stated formula, with the separable case already proved in step 2.1.

6.1step 5.1L2algebra∎

If char⁡F=p>0 and i>1, then [L2] makes i a positive power of p, so i⋅1F=0. The trace formula of step 5.1 is then identically zero.

Remarks

  • The inseparable exponent is load-bearing. In the separable case the norm is the product over the embeddings and the trace is their sum; outside the separable case the product must be raised to [K:F]i, and the trace may vanish identically.

  • The finite-field formulas on the earlier page are examples of this theorem. When K=Fqn over Fq, the embeddings are the Frobenius powers and the product and sum become the familiar Frobenius norm and trace.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

Norm is multiplicative, trace is F-linear, and both are transitive in towers

Statement

Let K/F be a finite extension and let a,b∈K.

  1. NK/F(ab)=NK/F(a)NK/F(b).
  2. Tr⁡K/F(a+b)=Tr⁡K/F(a)+Tr⁡K/F(b) and Tr⁡K/F(ca)=c Tr⁡K/F(a) for every c∈F.
  3. If L/K/F is a tower of finite extensions, then NL/F=NK/F∘NL/K,Tr⁡L/F=Tr⁡K/F∘Tr⁡L/K.

Facts & Assumptions

Given: Finite field extensions as in the Statement, multiplication maps mx for x∈K or x∈L, and product bases in towers.

[L1]

For same-sized square matrices over a commutative ring, determinant is multiplicative (For same-sized finite square matrices over a commutative ring, det⁡(AB)=det⁡(A)det⁡(B)).

[L2]

The embedding formulas identify norm with the product and trace with the sum of the conjugates, counted with the inseparable exponent (Norm and trace from embeddings, with the inseparable exponent in the norm formula).

[L3]

In a finite tower, a basis of the top field over the middle field times a basis of the middle field over the base is a basis of the top field over the base (Products of bases form a basis in a tower of finite extensions, Tower law for finite extensions: [L:F]=[L:K][K:F]).

[L4]

Restriction from the F-embeddings of L to the F-embeddings of K is surjective, and every fibre has cardinality [L:K]s after transporting the K-structure (Restriction partitions embeddings in a finite tower into extension fibres).

[L5]

Separable degrees multiply in finite towers: [L:F]s=[L:K]s[K:F]s (Separable degree is multiplicative in finite towers: [L:F]s=[L:K]s[K:F]s).

Proof

technique · direct
1.1F1L1algebra

Multiplication operators compose as mab=ma∘mb, because (ma∘mb)(x)=a(bx)=(ab)x. Therefore [F1] and [L1] give NK/F(ab)=det⁡(mab)=det⁡(ma)det⁡(mb)=NK/F(a)NK/F(b).

1.2F1algebra

The operator identity ma+b=ma+mb and the scalar identity mca=c ma make trace additive and F-linear. Hence [F1] gives Tr⁡K/F(a+b)=Tr⁡K/F(a)+Tr⁡K/F(b),Tr⁡K/F(ca)=c Tr⁡K/F(a).

1.3L2L4

Let L/K/F be a finite tower, fix an algebraic closure Ω/F, and write iL/F, iL/K, and iK/F for the three inseparable degrees. For an F-embedding σ:K→Ω, let Eσ be its restriction fibre in Hom⁡F(L,Ω). Applying the embedding formulas [L2] to L/K after transporting scalars along σ gives σ(NL/K(a))=(∏τ∈Eστ(a))iL/K,σ(Tr⁡L/K(a))=iL/K∑τ∈Eστ(a). The fibres Eσ partition Hom⁡F(L,Ω) by [L4].

1.4L3L5algebra

By the ordinary tower law [L3] and separable-degree multiplicativity [L5], iL/F=[L:F][L:F]s=[L:K][K:F][L:K]s[K:F]s=iL/KiK/F.

2.1L2step 1.3step 1.4algebra

Apply the outer embedding formulas [L2] for K/F to the two elements in step 1.3. Using the fibre partition and step 1.4 gives NK/F(NL/K(a))=(∏σ∏τ∈Eστ(a)iL/K)iK/F=(∏τ∈Hom⁡F(L,Ω)τ(a))iL/F=NL/F(a), and similarly Tr⁡K/F(Tr⁡L/K(a))=iK/FiL/K∑τ∈Hom⁡F(L,Ω)τ(a)=Tr⁡L/F(a).

3.1step 1.1step 1.2step 2.1∎

Steps 1.1, 1.2, and 2.1 prove the three claims.

TheoremStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

Field norm and trace agree with the determinant and trace of multiplication by an element

Statement

Let K/F be a finite extension and let a∈K. If

ma ⁣:K→K,x↦ax,

is the F-linear multiplication operator, then

NK/F(a)=det⁡(ma),Tr⁡K/F(a)=tr⁡(ma),

where the right-hand side uses the published linear-operator determinant and trace.

Facts & Assumptions

Given: A finite field extension K/F, an element a∈K, and the operator ma of multiplication by a.

[F1]

The field norm and trace were defined by NK/F(a):=det⁡(ma) and Tr⁡K/F(a):=tr⁡(ma) (The norm NK/F and trace Tr⁡K/F of a finite field extension).

Proof

technique · direct
1.1F1F2

The two displayed identities are exactly the definitions of [F1], and [F2] identifies the determinant and trace on the right with the published operator notions. Since a field extension has positive degree, the zero-dimensional determinant convention never needs a separate case here.

2.1step 1.1∎

This proves the stated dictionary identification.

Remarks

  • This is the promised dictionary item. The page is not introducing a second unrelated determinant or trace: the field-theoretic norm and trace are built from the same linear-algebra invariants already established for endomorphisms.
DefinitionDefinition: Literature-sourcedProof: Not applicableaudited 2026-08-26Open item page →

The trace form (x,y)↦Tr⁡K/F(xy) of a finite extension

Definition

Let K/F be a finite field extension. The trace form of K/F is the function

TK/F ⁣:K×K→F,TK/F(x,y):=Tr⁡K/F(xy),

where Tr⁡K/F is the field trace (The norm NK/F and trace Tr⁡K/F of a finite field extension).

Because multiplication in K is commutative and the trace is F-linear in its argument, the trace form is a symmetric bilinear form on the F-vector space K in the sense of Bilinear forms, and symmetric, skew-symmetric, and alternating bilinear forms. The later theorem on this page identifies exactly when it is nondegenerate.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The trace form of a finite extension is nondegenerate exactly when the extension is separable

Statement

Let K/F be a finite field extension and let

TK/F(x,y)=Tr⁡K/F(xy)

be its trace form (The trace form (x,y)↦Tr⁡K/F(xy) of a finite extension). Then TK/F is nondegenerate (The matrix, left and right radicals, rank, and nondegeneracy of a bilinear form on a finite-dimensional space) if and only if K/F is separable (Separable algebraic elements and separable extensions).

Facts & Assumptions

Given: A finite extension K/F, its trace form TK/F, and an F-basis e1,…,en of K.

[F1]

The trace form is the symmetric bilinear form (x,y)↦Tr⁡K/F(xy) (The trace form (x,y)↦Tr⁡K/F(xy) of a finite extension).

[F2]

A bilinear form on a finite-dimensional space is nondegenerate exactly when its matrix in one, hence every, basis is invertible (The matrix, left and right radicals, rank, and nondegeneracy of a bilinear form on a finite-dimensional space).

[L1]

The trace is the sum of the conjugates in the separable case and is identically zero in the inseparable case (Norm and trace from embeddings, with the inseparable exponent in the norm formula).

[L2]

Distinct embeddings are linearly independent, so their evaluation matrix on a suitable basis is invertible (Dedekind's linear independence theorem for distinct characters).

Proof

technique · direct
1.1F1L1

For the forward implication from inseparability to degeneracy, suppose K/F is not separable. Then [L1] makes Tr⁡K/F identically zero, so TK/F(x,y)=Tr⁡K/F(xy)=0 for every x,y∈K. Thus every vector lies in both radicals, and the form is degenerate.

1.2L2choose

For the converse direction, suppose K/F is separable and let σ1,…,σn ⁣:K→Ω be its distinct F-embeddings into an algebraic closure. Form the evaluation matrix A=(σi(ej))i,j. By [L2], A is invertible.

2.1F1L1step 1.2algebra

The matrix of TK/F in the basis e1,…,en is B=(Tr⁡K/F(eiej))i,j. Because K/F is separable, [L1] gives Tr⁡K/F(eiej)=∑r=1nσr(ei)σr(ej), so B=ATA.

3.1F2step 2.1algebra

Since A is invertible, the matrix B=ATA is invertible. Therefore [F2] makes the trace form nondegenerate.

4.1step 1.1step 3.1∎

Steps 1.1 and 3.1 prove the equivalence.

Remarks

  • The inseparable case is not a small defect but a total collapse. The trace itself vanishes, so the whole bilinear form vanishes.
CorollaryStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The trace map of a finite separable extension is surjective

Statement

If K/F is a finite separable field extension, then the trace map

Tr⁡K/F ⁣:K→F

is surjective.

Facts & Assumptions

Given: A finite separable extension K/F.

[L1]

The trace form (x,y)↦Tr⁡K/F(xy) is nondegenerate exactly when the extension is separable (The trace form of a finite extension is nondegenerate exactly when the extension is separable).

Proof

technique · direct
1.1L1algebra

By [L1], the trace form of K/F is nondegenerate. If the trace map itself were zero, then Tr⁡K/F(xy)=0 for every x,y∈K, so the vector 1∈K would lie in the radical, contradicting nondegeneracy. Therefore the trace map is not the zero linear functional.

2.1step 1.1algebra∎

The image of an F-linear map K→F is an F-subspace of the one-dimensional vector space F. A nonzero subspace of F is all of F, so the trace map is surjective.

DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

A cyclic extension is a finite Galois extension with cyclic Galois group

Definition

Let K/F be a finite Galois extension (Finite Galois extensions and Gal⁡(K/F)). It is cyclic when its Galois group Gal⁡(K/F) is a cyclic group.

When Gal⁡(K/F)=⟨σ⟩ is generated by a specific automorphism σ, that generator may be named in the body of a later theorem or example, but it is not part of the data of the definition.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

Hilbert's theorem 90 for a finite cyclic extension

Statement

Let K/F be a finite cyclic extension of degree n with Gal⁡(K/F)=⟨σ⟩. For b∈K×, the following are equivalent:

  1. NK/F(b)=1.
  2. There exists β∈K× with b=βσ(β).

Facts & Assumptions

Given: A finite cyclic extension K/F of degree n, a generator σ of its Galois group, and an element b∈K×.

[F1]

A cyclic extension is a finite Galois extension with cyclic Galois group (A cyclic extension is a finite Galois extension with cyclic Galois group).

[L1]

In a finite Galois extension, the norm is the product over the distinct F-embeddings (Norm and trace from embeddings, with the inseparable exponent in the norm formula).

[L2]

Distinct characters of a group into a field are linearly independent (Dedekind's linear independence theorem for distinct characters).

Proof

technique · direct
1.1F1L1

For the forward direction from 2 to 1, suppose b=β/σ(β) for some β∈K×. Since the embeddings of K/F are 1,σ,…,σn−1, [L1] gives NK/F(b)=∏i=0n−1σi(β)σi+1(β)=1, the numerator and denominator cancelling cyclically because σn=1.

1.2F1L2choose

For the converse, assume NK/F(b)=1. For 0≤i<n, set ci:=∏j=0i−1σj(b), with the empty product c0=1. Then the distinct automorphisms 1,σ,…,σn−1 restrict to distinct characters K×→K×, so [L2] implies that the K-linear operator T(x):=∑i=0n−1ci σi(x) is not identically zero. Choose x∈K with β:=T(x)≠0.

2.1step 1.2algebra

The coefficients satisfy ci+1=ciσi(b) for 0≤i<n−1, and cn=NK/F(b)=1. Therefore σ(ci)=b−1ci+1(0≤i<n−1),σ(cn−1)=b−1. Applying σ to β=T(x) and re-indexing the sum gives σ(β)=b−1β. Hence b=β/σ(β).

3.1step 1.1step 2.1∎

Steps 1.1 and 2.1 prove the equivalence.

Remarks

  • The proof uses only Dedekind independence. No cohomological language is needed here, although this is the classical vanishing of H1 for a finite cyclic extension.
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

Additive Hilbert 90: trace zero is the image of α↦α−σ(α)

Statement

Let K/F be a finite cyclic extension of degree n with Gal⁡(K/F)=⟨σ⟩. For b∈K, the following are equivalent:

  1. Tr⁡K/F(b)=0.
  2. There exists α∈K with b=α−σ(α).

Facts & Assumptions

Given: A finite cyclic extension K/F of degree n, a generator σ of its Galois group, and an element b∈K.

[F1]

A cyclic extension is finite Galois and therefore finite separable (A cyclic extension is a finite Galois extension with cyclic Galois group).

[L1]

In a finite separable extension, the trace map is surjective (The trace map of a finite separable extension is surjective).

Proof

technique · direct
1.1F1algebra

For the forward direction from 2 to 1, suppose b=α−σ(α). Summing the conjugates gives Tr⁡K/F(b)=∑i=0n−1σi(α−σ(α))=∑i=0n−1(σi(α)−σi+1(α))=0, again by telescoping and σn=1.

1.2F1L1choose

For the converse, assume Tr⁡K/F(b)=0. By [F1] and [L1], choose c∈K with Tr⁡K/F(c)=1. Define α:=∑i=0n−1(∑j=0i−1σj(b))σi(c), where the inner sum is 0 for i=0.

2.1step 1.2algebra

Put si=∑j=0i−1σj(b), so s0=0, sn=Tr⁡K/F(b)=0, and si=b+σ(si−1) for 1≤i≤n−1, while 0=sn=b+σ(sn−1). Applying σ to the coefficients as well gives σ(α)=∑i=0n−1σ(si)σi+1(c)=σ(sn−1)c+∑i=1n−1σ(si−1)σi(c). Therefore every coefficient of α−σ(α) is b, so α−σ(α)=b∑i=0n−1σi(c)=b Tr⁡K/F(c)=b.

3.1step 1.1step 2.1∎

Steps 1.1 and 2.1 prove the equivalence.

Remarks

  • This is the additive engine behind Artin-Schreier theory. The next theorem applies it to the trace-zero element −1 in characteristic p.
DefinitionDefinition: Literature-sourcedProof: Not applicablejudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The Lagrange resolvent attached to a cyclic action and a root of unity

Definition

Let K/F be a cyclic extension of degree n with generator σ, and let ζ be an n-th root of unity in an overfield of K (The group μn(K) of n-th roots of unity in a field, and primitive n-th roots of unity).

For x∈K, the Lagrange resolvent of x attached to (σ,ζ) is

Rσ,ζ(x):=∑i=0n−1ζ−iσi(x).

This is an F(ζ)-linear expression in the orbit of x under the cyclic action. When ζ∈F, the resolvent lies in K itself.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

If μn⊆F and char⁡F∤n, then a degree-n extension is cyclic exactly when it is F(α) with αn∈F and xn−αn irreducible

Statement

Let F be a field, let n≥1, assume char⁡F∤n, and assume F contains a primitive n-th root of unity. For a finite extension K/F of degree n, the following are equivalent:

  1. K/F is cyclic.
  2. There exists α∈K such that K=F(α), αn∈F, and xn−αn is irreducible over F.

Facts & Assumptions

Given: A field F, an integer n≥1, a primitive n-th root of unity ζ∈F, and a finite extension K/F of degree n.

[F1]

The Lagrange resolvent is Rσ,ζ(x)=∑i=0n−1ζ−iσi(x) (The Lagrange resolvent attached to a cyclic action and a root of unity).

[L1]

Distinct characters are linearly independent (Dedekind's linear independence theorem for distinct characters).

[L2]

If an extension contains one nonzero root α of xn−a, then all roots are ζiα, and the splitting field is obtained by adjoining α and the relevant roots of unity (After adjoining one nonzero root α of xn−a, all roots are ζα with ζn=1).

[L3]

When char⁡F∤n, the polynomial xn−1 is separable and a splitting field has cyclic root-of-unity group of order n (tn−1 is separable over K exactly when the characteristic does not divide n, and then a splitting field carries n distinct n-th roots of unity).

[L4]

A finite extension is Galois exactly when it is the splitting field of a separable polynomial (Equivalent characterizations of a finite Galois extension).

[L5]

Embeddings of a simple algebraic extension correspond to the distinct roots of the minimal polynomial (F-embeddings of F(α) into an algebraically closed field correspond to the distinct roots of mα).

Proof

technique · direct
1.1F1L1choose

For the forward direction, assume K/F is cyclic and choose a generator σ of Gal⁡(K/F). The distinct powers 1,σ,…,σn−1 are distinct characters K×→K×, so [L1] makes the resolvent operator Rσ,ζ of [F1] nonzero. Choose γ∈K with α:=Rσ,ζ(γ)≠0.

1.2L2L3L4algebra

For the converse direction, assume K=F(α), αn=a∈F, and xn−a is irreducible. If α=0, then a=0 and the irreducible polynomial is xn, which forces n=1. Hence K=F, so the extension is cyclic. Assume from now on that α≠0. Then [K:F]=n, and [L2] shows every root of xn−a is ζiα with ζi∈F. Because [L3] makes the n roots of xn−1 distinct, these roots ζiα are distinct as well. Hence all roots of xn−a already lie in K, and the polynomial is separable, so K is its splitting field and [L4] makes K/F Galois.

2.1step 1.2L4L5algebra

In the nonzero case of step 1.2, the root ζα of the irreducible polynomial xn−a determines, by [L5], an F-embedding K=F(α)→K sending α to ζα, hence an automorphism τ of K/F. Its powers send α to ζiα, so τ has order n. Because [K:F]=n and a finite Galois extension has at most [K:F] automorphisms, Gal⁡(K/F)=⟨τ⟩ is cyclic of order n.

2.2step 1.1algebra

Because σ(α)=∑i=0n−1ζ−iσi+1(γ)=ζα, the element αn is fixed by σ and hence by the whole cyclic Galois group, so αn∈F. If 0<d<n and αd∈F, then αd=σ(αd)=ζdαd, so ζd=1, contradicting primitivity of ζ. Therefore no smaller positive power of α lies in F. The minimal polynomial of α divides xn−αn and has degree n=[K:F], so it is exactly xn−αn and K=F(α).

3.1step 2.2step 2.1∎

Steps 2.2 and 2.1 prove the equivalence.

Remarks

  • The irreducibility clause is what forces degree exactly n. Without it, adjoining one n-th root can produce a proper divisor of n.
TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

In characteristic p, a degree-p extension is cyclic exactly when it is generated by a root of xp−x−a with a∈F and that polynomial irreducible

Statement

Let F be a field of characteristic p>0, and let K/F be a finite extension of degree p. Then the following are equivalent:

  1. K/F is cyclic.
  2. There exists α∈K and a∈F such that αp−α=a and xp−x−a is irreducible over F.

When these conditions hold, the roots of xp−x−a in K are exactly α+i for i∈Fp, so K is already the splitting field and its Galois group is generated by α↦α+1.

Facts & Assumptions

Given: A field F of characteristic p>0 and a degree-p extension K/F.

[F1]

A cyclic extension is a finite Galois extension with cyclic Galois group (A cyclic extension is a finite Galois extension with cyclic Galois group).

[L1]

In a cyclic extension, trace zero is equivalent to being of the form α−σ(α) (Additive Hilbert 90: trace zero is the image of α↦α−σ(α)).

[L2]

A finite extension is Galois exactly when it is the splitting field of a separable polynomial (Equivalent characterizations of a finite Galois extension).

Proof

technique · direct
1.1F1L1choose

For the forward direction, assume K/F is cyclic and choose a generator σ of its Galois group. Since [K:F]=p and char⁡F=p, one has Tr⁡K/F(−1)=−p=0. By [L1], choose α∈K with −1=α−σ(α), so σ(α)=α+1.

1.2L2algebra

For the converse direction, assume αp−α=a∈F and xp−x−a is irreducible. For each i∈Fp one has (α+i)p−(α+i)=αp−α=a, so the roots of the polynomial are exactly α+i for i∈Fp. Thus all roots lie in F(α), the derivative is −1≠0, and [L2] makes F(α)/F Galois. Because the polynomial is irreducible of degree p, this extension has degree p.

2.1step 1.1algebra

In characteristic p, one has (α+1)p−(α+1)=αp−α, so the element a:=αp−α is fixed by σ and therefore lies in F. Also α∉F because step 1.1 gives σ(α)≠α. Since [K:F]=p is prime and F⊆F(α)⊆K, one must have F(α)=K. The polynomial xp−x−a∈F[x] has root α and degree p=[F(α):F], so it is the minimal polynomial of α over F and is irreducible.

2.2step 1.2algebra

The rule τ(α)=α+1 permutes the root set and fixes F, so it extends to an F-automorphism of F(α). Its p-th power fixes α and each smaller positive power moves α, so τ has order p. A degree-p finite Galois extension has at most p automorphisms, hence exactly the cyclic group generated by τ. Therefore F(α)/F is cyclic of degree p.

3.1step 1.2step 2.1step 2.2∎

Steps 2.1 and 2.2 prove the equivalence, and the displayed root set in step 1.2 proves the final sentence.

DefinitionDefinition: AI-adaptedProof: Not applicableaudited 2026-08-26Open item page →

Kummer extensions from adjoining n-th roots over a base field containing μn

Definition

Let F be a field, let n≥1, assume char⁡F∤n, and assume F contains the full group μn of n-th roots of unity (The group μn(K) of n-th roots of unity in a field, and primitive n-th roots of unity).

For a subgroup B with

(F×)n⊆B⊆F×

and finite quotient B/(F×)n, choose for each b∈B an n-th root b1/n in an algebraic closure and write

F(B1/n):=F(b1/n:b∈B).

A finite Galois extension K/F is a Kummer extension of exponent dividing n when K=F(B1/n) for some such subgroup B.

The cyclic one-generator case is the earlier theorem If μn⊆F and char⁡F∤n, then a degree-n extension is cyclic exactly when it is F(α) with αn∈F and xn−αn irreducible.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The Kummer pairing Gal⁡(K/F)×B/(F×)n→μn is perfect

Statement

Let F be a field, let n≥1, assume char⁡F∤n, and assume μn⊆F. Let

(F×)n⊆B⊆F×

have finite quotient, and let K=F(B1/n) be the associated finite Kummer extension (Kummer extensions from adjoining n-th roots over a base field containing μn). Then the rule

⟨σ,  b(F×)n⟩:=σ(β)β,βn=b,

defines a well-defined bilinear pairing

Gal⁡(K/F)×B/(F×)n→μn,

and it is nondegenerate in both variables. In this sense the Kummer pairing is perfect.

Facts & Assumptions

Given: The field F, the integer n, the subgroup B, the Kummer extension K=F(B1/n), an automorphism σ∈Gal⁡(K/F), and an element b∈B with chosen n-th root β∈K.

[F1]

A Kummer extension is a finite Galois extension generated by n-th roots of elements of B, with μn⊆F (Kummer extensions from adjoining n-th roots over a base field containing μn).

[L1]

For a finite Galois extension, the fixed field of the full Galois group is the base field (Equivalent characterizations of a finite Galois extension).

Proof

technique · direct
1.1F1algebra

Because σ(β)n=σ(b)=b=βn, the quotient σ(β)/β is an n-th root of unity, so it lies in μn⊆F. If β′=ζβ is another chosen n-th root of b, then σ(β′)β′=σ(ζ)σ(β)ζβ=σ(β)β, since σ fixes F and hence ζ. If b′=bcn represents the same class in B/(F×)n, then choosing β′=βc gives the same quotient. Thus the pairing is well defined.

2.1step 1.1algebra

Bilinearity is immediate: ⟨στ,bˉ⟩=στ(β)β=σ(τ(β))τ(β)τ(β)β=⟨σ,bˉ⟩⟨τ,bˉ⟩, and for classes bˉ,cˉ represented by roots β,γ, ⟨σ,bˉcˉ⟩=σ(βγ)βγ=σ(β)βσ(γ)γ=⟨σ,bˉ⟩⟨σ,cˉ⟩.

2.2F1step 1.1

For nondegeneracy on the Galois side, let σ≠1. Since K is generated over F by the chosen n-th roots of elements of B, some such root β satisfies σ(β)≠β. For the class bˉ of b=βn, one then has ⟨σ,bˉ⟩=σ(β)β≠1. So no nontrivial automorphism lies in the left kernel.

2.3F1L1step 1.1

For nondegeneracy on the B/(F×)n side, let bˉ be nontrivial. Then β∉F, for otherwise b=βn would lie in (F×)n. By [F1] and [L1], some σ∈Gal⁡(K/F) satisfies σ(β)≠β. Hence ⟨σ,bˉ⟩=σ(β)β≠1, so no nontrivial class lies in the right kernel.

3.1step 1.1step 2.1step 2.2step 2.3∎

Steps 1.1, 2.1, 2.2, and 2.3 prove the stated well-defined bilinear pairing and its nondegeneracy in both variables.

CorollaryStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

The degree [F(an):F] is the order of a(F×)n in F×/(F×)n

Statement

Let F be a field, let n≥1, assume char⁡F∤n, assume μn⊆F, and let a∈F×. If the class of a in F×/(F×)n has order d, then

[F(an):F]=d.

Facts & Assumptions

Given: The field F, the integer n, the element a∈F×, and the order d of the class aˉ=a(F×)n in F×/(F×)n.

[L1]

The Kummer pairing on the one-generator extension is bilinear and nondegenerate in both variables (The Kummer pairing Gal⁡(K/F)×B/(F×)n→μn is perfect).

Proof

technique · direct
1.1L1

Let K=F(an). The subgroup of F×/(F×)n generated by aˉ is cyclic of order d, and [L1] gives a homomorphism ϕ ⁣:Gal⁡(K/F)→μn,ϕ(σ)=⟨σ,aˉ⟩. If ϕ(σ)=1, then σ lies in the left kernel of the pairing, so σ=1 by [L1]. Thus ϕ is injective.

2.1step 1.1L1algebra

Let H=ϕ(Gal⁡(K/F))≤μn. If ∣H∣=m<d, then every h∈H satisfies hm=1, so bilinearity makes ⟨σ,aˉ m⟩=⟨σ,aˉ⟩m=1 for every σ∈Gal⁡(K/F). Then aˉ m lies in the right kernel, contradicting [L1] because aˉ has order d. Therefore ∣H∣=d. Since ϕ is injective, this gives ∣Gal⁡(K/F)∣=d.

3.1step 2.1L1algebra∎

The extension K/F is finite Galois in this one-generator Kummer situation, so its degree equals the order of its Galois group. Hence [K:F]=d.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

Kummer theory classifies finite abelian extensions of exponent dividing n by subgroups between (F×)n and F×

Statement

Let F be a field, let n≥1, assume char⁡F∤n, and assume μn⊆F. Then the assignment

B⟼F(B1/n)

is an inclusion-preserving bijection between

  1. subgroups B with (F×)n⊆B⊆F× and finite quotient B/(F×)n, and
  2. finite abelian extensions K/F whose Galois group has exponent dividing n.

For corresponding objects one has

Gal⁡(K/F)≅Hom⁡(B/(F×)n,μn).

Facts & Assumptions

Given: The field F, the integer n, a subgroup B as above, and a finite abelian extension K/F of exponent dividing n.

[F1]

A Kummer extension is a finite Galois extension of the form F(B1/n) with μn⊆F (Kummer extensions from adjoining n-th roots over a base field containing μn).

[L1]

The Kummer pairing for K=F(B1/n) is perfect, in particular bilinear and nondegenerate in both variables (The Kummer pairing Gal⁡(K/F)×B/(F×)n→μn is perfect).

[L2]

Cyclic degree-d extensions over a base containing μd are exactly the extensions generated by a d-th root whose defining polynomial xd−a is irreducible (If μn⊆F and char⁡F∤n, then a degree-n extension is cyclic exactly when it is F(α) with αn∈F and xn−αn irreducible).

[L3]

Artin's fixed-field theorem computes degrees from finite automorphism groups (Artin's fixed-field theorem: [K:KG]=∣G∣ and Aut⁡(K/KG)=G).

[L4]

In a finite Galois extension, normal intermediate fields correspond to quotient Galois groups (Normal subgroups, conjugate fields, and quotient groups in the Galois correspondence).

[L5]

Every finite abelian group is a finite direct product Cn1×⋯×Cnr with 1<n1∣⋯∣nr (Fundamental theorem of finite abelian groups: invariant-factor form).

Proof

technique · direct
1.1F1choosealgebra

Let B be a subgroup with finite quotient. Because B/(F×)n is finite, choose elements a1,…,ar∈B whose classes generate it. Then F(B1/n)=F(a11/n,…,ar1/n). For each j, the field F(aj1/n) contains all roots of xn−aj because μn⊆F, so it is a finite Galois extension whose automorphism group embeds in μn by the rule σ↦σ(aj1/n)/aj1/n. Thus each one-generator step is cyclic of degree dividing n. The compositum K:=F(B1/n) is therefore finite Galois and abelian, and every automorphism acts on every chosen n-th root by multiplication with an element of μn. Hence the exponent of Gal⁡(K/F) divides n.

1.2L1

For this extension, the Kummer pairing of [L1] is perfect, so the natural maps Gal⁡(K/F)⟶Hom⁡(B/(F×)n,μn),B/(F×)n⟶Hom⁡(Gal⁡(K/F),μn) are isomorphisms because both groups are finite and the pairing is nondegenerate in both variables. In particular, Gal⁡(K/F)≅Hom⁡(B/(F×)n,μn).

1.3algebra

Conversely, let K/F be finite abelian of exponent dividing n and define B(K):={ b∈F×:some n-th root of b lies in K }. This contains (F×)n. For b∈B(K) choose β∈K with βn=b and define χb(σ):=σ(β)β∈μn(σ∈Gal⁡(K/F)). As in the Kummer-pairing proof, this depends only on the class of b modulo (F×)n, and if χb=1 then every σ fixes β, so β∈F and b∈(F×)n. Therefore B(K)/(F×)n↪Hom⁡(Gal⁡(K/F),μn), so the quotient is finite.

1.4L2L3L4L5algebrachoose

Let L:=F(B(K)1/n)⊆K and put G=Gal⁡(K/F). If σ∈G is nontrivial, use [L5] to write G≅Cn1×⋯×Cnr. Each nj divides n because the exponent of G divides n. Some coordinate of σ is nonzero, say in Cnj. If ζ∈F is a primitive n-th root of unity, then η=ζn/nj has order nj, so projection to this coordinate followed by a generator-preserving map Cnj→⟨η⟩ gives a character χ:G→μn with χ(σ)≠1. Put H=ker⁡χ and d=∣im⁡χ∣. Then H⊴G, d∣n, and [L4] makes E:=KH/F cyclic of degree d. Since char⁡F∤d and μd⊆F, [L2] gives E=F(γ) with γd∈F. Hence γn=(γd)n/d∈F, so γ∈L. But χ(σ)≠1 means σ∉H=Gal⁡(K/E), so σ acts nontrivially on E=F(γ) and therefore does not fix L. Thus Gal⁡(K/L)=1, and [L3] gives [K:L]=1. Therefore K=F(B(K)1/n).

2.1step 1.4

Starting from a finite abelian extension K/F, step 1.4 shows that the radicals coming from B(K) generate K, so the field-side composite is the identity.

2.2step 1.2algebra

Starting from a subgroup B and putting K=F(B1/n), one has B⊆B(K) by definition. For the reverse inclusion, let c∈B(K) and choose βc∈K with βcn=c. Step 1.2 identifies the perfect pairing with an isomorphism B/(F×)n≅Hom⁡(Gal⁡(K/F),μn), so the character χc(σ)=σ(βc)/βc equals χb for some class bˉ∈B/(F×)n represented by βbn=b with b∈B. Then for every σ∈Gal⁡(K/F) one has σ(βc/βb)βc/βb=χc(σ)χb(σ)=1, so βc/βb∈F. Hence c/b=(βc/βb)n∈(F×)n⊆B, and therefore c∈B. Thus B(K)=B.

3.1step 1.2step 2.1step 2.2∎

Inclusion preservation is immediate from the definitions, and the displayed Galois-group formula is exactly the isomorphism of step 1.2.

Remarks

  • This page uses the cohomology-free version. The proof is organized through cyclic Kummer subextensions and the concrete pairing, not through H1.
DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

A radical extension is a tower obtained by adjoining one n-th root at each step

Definition

Let L/F be a field extension. It is a radical extension when there is a finite tower

F=F0⊆F1⊆⋯⊆Fr=L

such that for each i with 1≤i≤r there are an integer mi≥1 and an element ai∈Fi−1 with

Fi=Fi−1(αi)andαimi=ai.

No Galois or root-of-unity hypothesis is built into the definition. Trivial steps with mi=1 are allowed.

DefinitionDefinition: AI-adaptedProof: Not applicablejudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

A polynomial is solvable by radicals when its splitting field lies in a radical extension

Definition

Let f∈F[x] be a nonzero polynomial and let E/F be a splitting field of f (Polynomials that split and splitting fields of a polynomial or a family of polynomials).

The polynomial f is solvable by radicals when there exists a radical extension L/F (A radical extension is a tower obtained by adjoining one n-th root at each step) such that E⊆L.

This is deliberately the weakest standard convention: the radical tower itself is not required to be Galois and is not required to contain the relevant roots of unity in advance.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

The normal closure of a radical extension is again radical

Statement

Let L/F be a finite radical extension. Then the normal closure of L/F is again a radical extension of F.

Facts & Assumptions

Given: A radical tower F=F0⊆F1⊆⋯⊆Fr=L.

[F1]

A radical extension is built by adjoining one root of one equation xmi=ai at each step (A radical extension is a tower obtained by adjoining one n-th root at each step).

[L1]

After adjoining one nonzero root of xm−a, all the remaining roots are obtained by multiplying by m-th roots of unity (After adjoining one nonzero root α of xn−a, all roots are ζα with ζn=1, The group μn(K) of n-th roots of unity in a field, and primitive n-th roots of unity).

Proof

technique · direct
1.1F1

We induct on the length r of the radical tower. For r=0, the extension is F/F, whose normal closure is itself.

1.2F1

Assume r>0, let N/F be the normal closure of Fr−1/F, and write Fr=Fr−1(α),αm=a∈Fr−1.

2.1step 1.2algebra

If α=0, then Fr=Fr−1, so the normal closure of Fr/F is just N.

2.2L1step 1.2algebra

Assume instead that α≠0. Because N/F is normal and contains Fr−1, every F-conjugate of a lies in N. Let a1,…,as∈N be the distinct conjugates of a over F, and choose roots αj with αjm=aj. Every F-conjugate of α is then a nonzero root of some polynomial xm−aj, so [L1] says it has the form ζαj for some ζ∈μm. Therefore the normal closure M of Fr/F is exactly M=N(α1,…,αs,μm).

2.3step 1.1F1

By the induction hypothesis, N/F is radical.

3.1step 2.2step 2.3F1algebra

In the case of step 2.2, the field M is radical over N: adjoin the finitely many αj one at a time, each by one equation xm−aj, and then adjoin generators of μm by roots of xm−1. Concatenating that tower with the radical tower for N/F from step 2.3 shows that M/F is radical.

4.1step 2.1step 2.2step 3.1∎

Step 2.1 handles the case α=0. Otherwise step 2.2 identifies the normal closure as M, and step 3.1 shows that M/F is radical. Thus the induction closes, so the normal closure of every finite radical extension is radical.

LemmaStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

Adjoining roots of unity to a finite Galois extension adds an abelian kernel and preserves solvability

Statement

Let K/F be a finite Galois extension, let n≥1, assume char⁡F∤n, and put L=K(μn). Then L/F is finite Galois, restriction gives a surjective homomorphism

Gal⁡(L/F)↠Gal⁡(K/F),

and its kernel is Gal⁡(L/K), which is abelian. Consequently Gal⁡(L/F) is solvable if and only if Gal⁡(K/F) is solvable.

Facts & Assumptions

Given: A finite Galois extension K/F, the cyclotomic extension F(μn)/F, and the compositum L=K(μn).

[L2]

A finite extension is Galois exactly when it is the splitting field of a separable polynomial (Equivalent characterizations of a finite Galois extension).

[L3]

In a finite Galois tower, normal intermediate fields correspond to normal subgroups and quotient Galois groups (Normal subgroups, conjugate fields, and quotient groups in the Galois correspondence).

[L4]

Subgroups and quotients of solvable groups are solvable, and an extension of solvable groups is solvable (Subgroups and quotients of solvable groups are solvable, Extensions and finite direct products of solvable groups are solvable).

Proof

technique · direct
1.1L1L2

Because K/F is finite Galois, it is the splitting field of a separable polynomial f∈F[x] by [L2]. By [L1], F(μn) is the splitting field of the separable polynomial xn−1 over F. Therefore L=K(μn) is the splitting field over F of the separable polynomial f(x)(xn−1), so L/F is finite Galois by [L2].

1.2L3

Since K/F is Galois and K is an intermediate field of the finite Galois extension L/F, [L3] gives a surjective restriction map ρ ⁣:Gal⁡(L/F)↠Gal⁡(K/F) with kernel Gal⁡(L/K).

2.1step 1.2L1L4

The kernel extension L/K is the cyclotomic extension K(μn)/K, so [L1] makes Gal⁡(L/K) abelian. If Gal⁡(L/F) is solvable, then its quotient Gal⁡(K/F) is solvable by [L4]. Conversely, if Gal⁡(K/F) is solvable, then [L4] applied to the exact sequence with abelian kernel from step 1.2 makes Gal⁡(L/F) solvable.

3.1step 1.1step 1.2step 2.1∎

This proves every part of the statement.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

In characteristic 0, a polynomial solvable by radicals has a solvable Galois group

Statement

Let F be a field of characteristic 0, and let f∈F[x] be nonzero. If f is solvable by radicals, then the Galois group of its splitting field over F is a solvable group.

Facts & Assumptions

Given: A polynomial f∈F[x] of characteristic 0 that is solvable by radicals, with splitting field E/F.

[F1]

Solvable by radicals means that E lies inside some radical extension of F (A polynomial is solvable by radicals when its splitting field lies in a radical extension).

[L1]

The normal closure of a radical extension is radical (The normal closure of a radical extension is again radical).

[L2]

Adjoining roots of unity to a finite Galois extension adds an abelian kernel and preserves solvability of the Galois group (Adjoining roots of unity to a finite Galois extension adds an abelian kernel and preserves solvability).

Proof

technique · direct
1.1F1L1choose

By [F1], choose a finite radical extension L/F with E⊆L. Replacing L by its normal closure over F, [L1] lets us assume from the start that L/F is finite Galois and radical.

2.1step 1.1L2

Let the radical tower for L/F use exponents m1,…,mr, and let N=m1⋯mr. Adjoin μN to L. By [L2], solvability of Gal⁡(L/F) is equivalent to solvability of Gal⁡(L(μN)/F). So it is enough to prove the latter solvable.

3.1step 2.1L3algebra

After adjoining μN, every step of the radical tower becomes a finite Galois extension with cyclic Galois group: if Fi=Fi−1(αi) with αimi∈Fi−1, then the enlarged lower field already contains μmi, so every root of xmi−αimi is ζαi with ζ∈μmi. Thus Fi(μN)/Fi−1(μN) is the splitting field of a separable polynomial, and every automorphism is determined by αi↦ζαi, so its Galois group embeds in the cyclic group μmi. In particular each step has solvable Galois group, and repeated use of [L3] up the tower makes Gal⁡(L(μN)/F(μN)) solvable.

4.1step 3.1L2L3

The cyclotomic extension F(μN)/F has abelian Galois group, hence solvable. Applying [L3] once more to the tower F⊆F(μN)⊆L(μN) shows that Gal⁡(L(μN)/F) is solvable. By step 2.1 the same is true of Gal⁡(L/F).

5.1step 4.1L3∎

The splitting field E is an intermediate field of the finite Galois extension L/F, so Gal⁡(E/F) is a quotient of a subgroup of Gal⁡(L/F). Therefore [L3] makes Gal⁡(E/F) solvable.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

In characteristic 0, a solvable Galois group makes a polynomial solvable by radicals

Statement

Let F be a field of characteristic 0, and let f∈F[x] be nonzero. If the splitting field E/F of f has solvable Galois group, then f is solvable by radicals.

Facts & Assumptions

Given: A polynomial f∈F[x] with splitting field E/F and solvable Galois group G=Gal⁡(E/F).

[F1]

A polynomial is solvable by radicals when its splitting field is contained in a radical extension of the base field (A polynomial is solvable by radicals when its splitting field lies in a radical extension).

[L1]

A finite solvable group has a composition series whose factors are cyclic of prime order (A finite group is solvable if and only if all its composition factors are cyclic of prime order).

[L2]

Normal subgroups in a finite Galois group correspond to Galois intermediate fields, and quotient groups give the corresponding Galois groups (Normal subgroups, conjugate fields, and quotient groups in the Galois correspondence).

[L3]

Over a base containing the needed roots of unity, a cyclic degree-d extension is generated by adjoining one d-th root (If μn⊆F and char⁡F∤n, then a degree-n extension is cyclic exactly when it is F(α) with αn∈F and xn−αn irreducible).

Proof

technique · direct
1.1L1L2

By [L1], choose a composition series G=G0▹G1▹⋯▹Gr=1 whose factors Gi−1/Gi have prime order pi. Put Ei:=EGi. Then F=E0⊆E1⊆⋯⊆Er=E, and [L2] makes each step Ei/Ei−1 a cyclic Galois extension of degree pi.

2.1step 1.1F1choose

We enlarge the lower fields one step at a time. Starting with R0=F, suppose Ri−1 is already a radical extension of F containing Ei−1. Adjoin a primitive pi-th root of unity to Ri−1. This is one radical step because it adjoins a root of xpi−1. Let the enlarged field be Ri−1′.

3.1step 2.1L3

The compositum EiRi−1′/Ri−1′ is still cyclic of degree 1 or pi, because it is a quotient of the cyclic group Gal⁡(Ei/Ei−1) after base change. Since Ri−1′ contains the primitive pi-th roots of unity, [L3] shows that EiRi−1′ is obtained from Ri−1′ by adjoining one pi-th root. Therefore Ri:=EiRi−1′ is radical over Ri−1, hence radical over F.

4.1step 3.1F1∎

Iterating step 3.1 produces a radical extension Rr/F with E=Er⊆Rr. By [F1], the polynomial f is solvable by radicals.

LemmaStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The symmetric groups Sn are solvable for n≤4

Statement

The symmetric groups Sn are solvable for every n≤4.

Facts & Assumptions

Given: The symmetric groups on at most four letters.

[L1]

A group is solvable exactly when it has a finite subnormal series with abelian factors (A group is solvable if and only if it has a subnormal series with abelian factors).

[L3]

Conjugation relabels cycles and therefore relabels a double transposition to another double transposition (Conjugating a cycle relabels each entry: g(a1 … ak)g−1=(g(a1) … g(ak))).

Proof

technique · direct
1.1L1algebra

The groups S0 and S1 are trivial, and S2 is cyclic of order 2, so all three are solvable by [L1].

1.2L1L2algebra

For S3, the subgroup A3 is normal as the sign kernel by [L2], has order 3, and the quotient S3/A3 has order 2. Thus 1◃A3◃S3 is a subnormal series with abelian factors, so S3 is solvable by [L1].

1.3L1L2L3algebra

In A4, let V4={1,(12)(34),(13)(24),(14)(23)}. Each listed double transposition is even, and direct multiplication shows that these four elements form a subgroup. By [L3], conjugation by any element of A4 permutes the three nonidentity double transpositions, so V4⊴A4. The quotient A4/V4 has order 3, hence is abelian, and V4 is abelian of order 4. Since A4 is the sign kernel in S4, the chain 1◃V4◃A4◃S4 has abelian factors V4, A4/V4, and S4/A4. Therefore S4 is solvable by [L1].

2.1step 1.1step 1.2step 1.3∎

Steps 1.1, 1.2, and 1.3 cover every n≤4.

CorollaryStatement: Literature-sourcedProof: AI-adaptedprecheck passaudited 2026-08-26Open item page →

Every polynomial of degree at most four is solvable by radicals

Statement

Every polynomial over a field of characteristic 0 of degree at most four is solvable by radicals.

Facts & Assumptions

Given: A characteristic-0 field F and a polynomial f∈F[x] of degree at most four.

[L1]

The groups Sn are solvable for n≤4 (The symmetric groups Sn are solvable for n≤4).

[L2]

The Galois group of a separable degree-m polynomial embeds in Sm (A polynomial Galois group acts faithfully on its roots).

[L3]

Subgroups of solvable groups are solvable (Subgroups and quotients of solvable groups are solvable).

[L4]

In characteristic 0, a polynomial with solvable Galois group is solvable by radicals (In characteristic 0, a solvable Galois group makes a polynomial solvable by radicals).

Proof

technique · direct
1.1L1L2L3

Let E/F be the splitting field of f, and let G be its Galois group. Since f is separable in characteristic 0, [L2] embeds G in Sm for some m≤4. By [L1] and [L3], the group G is solvable.

2.1step 1.1L4∎

Apply [L4] to G: the polynomial f is solvable by radicals.

TheoremStatement: AI-adaptedProof: AI-generatedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

For prime p, a transitive subgroup of Sp containing a transposition is all of Sp

Statement

Let p be prime and let G≤Sp act transitively on {0,1,…,p−1}. If G contains a transposition, then G=Sp.

Facts & Assumptions

Given: A prime p, a transitive subgroup G≤Sp, and a transposition τ∈G.

[F1]

A transitive action is one with a single orbit (Left group actions, transitive actions, and faithful actions).

[L1]
[L2]

If a prime divides the order of a finite group, the group contains an element of that prime order (Cauchy's theorem: if a prime p divides ∣G∣, then G has an element of order p).

[L3]

The order of a permutation is the least common multiple of its nontrivial cycle lengths (The order of a permutation is the least positive common multiple of its nontrivial cycle lengths, with value 1 for the identity).

[L4]

The adjacent transpositions generate the full symmetric group on p letters (Adjacent transpositions generate the finite symmetric group Sn).

Proof

technique · direct
1.1F1L1L2

Because the action of G on {0,1,…,p−1} is transitive, the orbit of any point has size p. Hence [L1] gives p∣∣G∣. By [L2], the group G contains an element c of order p.

1.2L3L5algebra

By [L3], a permutation of order p in Sp must be a p-cycle: every nontrivial cycle length divides p, so each is 1 or p, and there must be one nontrivial cycle. Conjugating inside Sp, we may relabel so that c=(0 1 … p−1).

2.1step 1.2L5algebra

Write the given transposition as τ=(i j) with i≠j, and put d≡j−i(modp), choosing d∈{1,…,p−1}. For each k, ckτc−k=(i+k  j+k), so G contains every transposition of the form (x  x+d), with indices read modulo p. Because p is prime, d is invertible modulo p, so the sequence 0, d, 2d, …, (p−1)d lists all p symbols exactly once modulo p. Therefore the p−1 transpositions joining consecutive terms in that order all lie in G.

3.1L4L5step 2.1∎

Let π be the relabelling permutation carrying k to kd modulo p. By step 2.1 and [L5], the conjugates π(k  k+1)π−1 for 0≤k≤p−2 are exactly the p−1 transpositions joining consecutive terms in the ordering of step 2.1, so they lie in G. Since [L4] says the standard adjacent transpositions generate Sp, their conjugates also generate Sp. Hence G contains a generating set of Sp, so G=Sp.

TheoremStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

The general polynomial of degree n has Galois group Sn

Statement

Let F be a field and let X1,…,Xn be algebraically independent over F. Write e1,…,en for the elementary symmetric polynomials in the Xi, let

E:=F(e1,…,en)⊆L:=F(X1,…,Xn),

and let

g(t):=∏i=1n(t−Xi)=tn−e1tn−1+⋯+(−1)nen∈E[t].

Then L is the splitting field of g over E, and

Gal⁡(L/E)≅Sn.

Facts & Assumptions

Given: The field F, the indeterminates X1,…,Xn, the rational function field L=F(X1,…,Xn), the subfield E=F(e1,…,en), and the polynomial g(t) above.

[L1]

Every symmetric polynomial is uniquely a polynomial in the elementary symmetric polynomials (Fundamental theorem of symmetric polynomials: unique expression as a polynomial in e1,…,en).

[L2]

For every field A, the one-variable rational function field A(t) is the field of fractions of A[t] (For a field F, F(t)=Frac⁡(F[t]) is its rational function field; in particular R(t)=Frac⁡(R[t])).

[L3]

Artin's fixed-field theorem says that if a finite automorphism group G acts on a field L, then [L:LG]=∣G∣ and Aut⁡(L/LG)=G (Artin's fixed-field theorem: [K:KG]=∣G∣ and Aut⁡(K/KG)=G).

[L4]

A finite extension is Galois exactly when it is the splitting field of a separable polynomial (Equivalent characterizations of a finite Galois extension).

Proof

technique · direct
1.1givenalgebra

The polynomial g(t) belongs to E[t] by construction and splits in L with roots X1,…,Xn. Since L is generated over F by the Xi, it is generated over E by those same roots, so L is the splitting field of g over E.

1.2L1algebra

Every permutation π∈Sn acts on L by π(Xi)=Xπ(i). Because each ej is symmetric, [L1] makes every ej fixed by this action, so Sn≤Aut⁡(L/E).

2.1L1L2L3algebra

Conversely, let h∈LSn. Iterating [L2] and clearing coefficient denominators writes h=f/g with f,g∈F[X1,…,Xn] and g≠0. Put D:=∏π∈Snπ(g). Then D is a nonzero symmetric polynomial, and hD=f∏π∈Snπ≠1π(g) is also a polynomial. Both h and D are fixed by Sn, so hD is symmetric as well. By [L1], both D and hD are polynomials in e1,…,en, hence h=(hD)/D∈F(e1,…,en)=E. The reverse inclusion is already in step 1.2, so LSn=E. Now [L3] yields [L:E]=∣Sn∣=n! and Aut⁡(L/E)=Sn.

3.1step 1.1step 2.1L4∎

Step 1.1 makes L/E the splitting field of g, and step 2.1 identifies its full automorphism group with Sn. Therefore Gal⁡(L/E)≅Sn.

CorollaryStatement: Literature-sourcedProof: AI-adaptedprecheck passjudge pass (gpt-5.6-terra)audited 2026-08-26Open item page →

For n≥5 over a characteristic-zero base, the general polynomial of degree n is not solvable by radicals

Statement

Let F be a field of characteristic 0. For every n≥5, the general polynomial of degree n over the rational function field F(e1,…,en) is not solvable by radicals.

Facts & Assumptions

Given: A field F of characteristic 0, an integer n≥5, and the general polynomial of degree n of the previous theorem.

[L1]

The general polynomial of degree n has Galois group Sn (The general polynomial of degree n has Galois group Sn).

[L2]

The group Sn is not solvable for n≥5 (A5 and Sn for n≥5 are not solvable).

[L3]

In characteristic 0, a polynomial solvable by radicals has solvable Galois group (In characteristic 0, a polynomial solvable by radicals has a solvable Galois group).

Proof

technique · direct
1.1L1L2

By [L1], the Galois group of the general polynomial is Sn, and [L2] says that group is not solvable for n≥5.

2.1step 1.1L3∎

If the polynomial were solvable by radicals, [L3] would force its Galois group to be solvable, contradicting step 1.1. Therefore it is not solvable by radicals.

5 · Examples, counterexamples and false statements

None yet.

Sources