Blog Archive

Friday, June 13, 2025

ANT.2.2

 

Galois Theory and Field Extensions

Minimal Polynomial via Galois Conjugates

Lemma

Lemma (Minimal Polynomial via Galois Conjugates): If E/K is a Galois extension and αE, then

mα,K(X)=βOrb(α)(Xβ)

where Orb(α)={σ(α):σGal(E/K)} is the Galois orbit of α.


Proof (Galois-theoretic approach):

Let G=Gal(E/K) and m=mα,K(X) be the minimal polynomial of α over K.

Step 1: Let F be the splitting field of m(X) over K.

Since m(X) is separable (as E/K being Galois implies separability), the extension F/K is Galois.

Step 2: Establish the subfield relation and group homomorphism.

Clearly αE, so K(α)E. Since F is the splitting field of m(X) and m(X) is the minimal polynomial of α, we have F=K(α1,,αd), where α1=α,α2,,αd are all roots of m(X).

Since these roots are all algebraic conjugates of α, they all lie in E, hence FE.

By the Fundamental Theorem of Galois Theory, there exists a surjective homomorphism:

ρ:G=Gal(E/K)Gal(F/K)

Step 3: Analyze the relationship between orbits and roots.

Let the roots of m(X) be {α1,α2,,αd}, where α1=α.

For any σG, σ(α) is an algebraic conjugate of α, i.e., m(σ(α))=0. This follows because:

m(σ(α))=σ(m(α))=σ(0)=0

Therefore, Orb(α){α1,α2,,αd}.

Step 4: Prove that the orbit equals the root set.

Conversely, for any root αi of m(X), since F/K is Galois, there exists τGal(F/K) such that τ(α)=αi.

Since ρ:GGal(F/K) is surjective, there exists σG such that ρ(σ)=τ.

This means σ|F=τ, hence σ(α)=τ(α)=αi.

Therefore αiOrb(α), which gives us {α1,α2,,αd}Orb(α).

Step 5: Conclusion.

Combining Steps 3 and 4, we obtain: Orb(α)={α1,α2,,αd}

Therefore:

mα,K(X)=i=1d(Xαi)=βOrb(α)(Xβ)

This completes the proof.

Cyclotomic Polynomials

Definition

Consider the Galois extension QQ(ζn), let G=Gal(Q(ζn)/Q)U(Z/nZ), define

Φn=σG(xσ(ζn))

i.e. Φn(α)=0αn1=0 with ord(α)=n.

By the Lemma above, Φn is the minimal polynomial of ζn. Hence it is irreducible.

Proposition
xn1=d|nΦd

Proof. Let Gal(Q(ζn)/Q) acts on 1,ζn,...,ζnn1, then xn1=d|nΦd comes from the orbit decomposition directly.

ζniζnjord(ζni)=ord(ζnj)

Trace and Norm

Historical Note

Alexander Grothendieck (1928-2014) revolutionized algebraic geometry and number theory through his radical reconceptualization of mathematical foundations. In the 1950s and 1960s, working primarily at the Institut des Hautes Études Scientifiques (IHÉS) near Paris, Grothendieck developed a vast framework that transformed how mathematicians approach abstract structures.

The tensor product approach to field extensions presented here reflects Grothendieck's profound influence. While classical Galois theory had been established in the 19th century, Grothendieck's functorial perspective and scheme theory provided powerful new tools for understanding these structures. His development of étale cohomology, descent theory, and the formalism of derived categories created a language where field extensions could be viewed within a broader categorical context.

Let LK be a finite separable extension, then K=L[θ]L[x]/(f(x)).

Hence

KLLL[x]/(xθ1)×...×L[x]/(xθn)Ln

For kK, we have

k1(σ1(k),...,σn(k))

Here σ1,...,σn is all the L-embedding.

The linear map mk(x1)=k(x1) correspond to the matrix

mk=diag(σ1(k),...,σn(k))

The trace of this matrix is i=1nσi(k), and det(mk)=i=1nσi(k). That is, the trace and norm of k.

Compositional Properties of Trace and Norm in Tower Extensions

Let LMK be finite separable extensions with

[M:L]=m,[K:M]=n,[K:L]=mn

and fix an algebraic closure L to split all polynomials.

Base Field Extension and Matrix Diagonalization
Step 1: Extension of M over L

Extending M from L to L, we have

MLLLm

corresponding to the set of L-embeddings {σ1,,σm}=HomL(M,L).

For any yM, its multiplication operator my on Lm is

my=diag(σ1(y),,σm(y))

Therefore

TrM/L(y)=i=1mσi(y),NM/L(y)=i=1mσi(y)

Remark.

By the infinite Galois corresponding, the fixed field of Gal(L/L) is L.

Hence we have TrM/L(y)=i=1mσi(y),NM/L(y)=i=1mσi(y)L.

Step 2: Extension of K over M

Extending K from M to the same L, we have

KMLLn

corresponding to each σi in HomL(M,L), there is a set of extension embeddings

{τi1,,τin}HomM(K,L)

Therefore for xK, the multiplication operator mx on this n-dimensional space is

diag(τi1(x),,τin(x))

yielding

j=1nτij(x)=TrK/M(x),j=1nτij(x)=NK/M(x)

where the results belong to M, then embedded into L by σi.

Step 3: Combining into one step

Performing the two-stage base‐change

MLLLmKMLLn

is equivalent to a single extension

KLL(MLL)MLLmn

We now spell out the two‐step unfolding of each coordinate:

  1. First–level: σi–coordinates.
    Under

    MLLLm,x1(σ1(x),,σm(x))

    each copy of L is indexed by an L–embedding σiHomL(M,L).

  2. Second–level: τij–coordinates.
    Now view each factor σi(x)L as coming from

    σi:MLKLL

    and extend scalars again along ML. This splits each σi–line into n lines, indexed by

    {τi1,,τin}HomM(K,L)

    Concretely,

    σi(x)(τi1(x),,τin(x))
  3. Combined coordinate map.
    Putting these two steps together, an elementary tensor x1 in KLL corresponds to the concatenated tuple

    (x1)(τ11(x),,τ1n(x)σ1-block||τm1(x),,τmn(x)σm-block)Lmn

Thus the full block–diagonal form of the multiplication operator mk on Lmn has m blocks (one for each σi), each block being the n×n diagonal matrix diag(τij(k))j=1n.

Step 4: Trace composition
Tr(mk)=i=1mj=1nτij(k)
=i=1m(j=1nτij(k))
=i=1mσi(TrK/M(k))
=TrM/L(TrK/M(k))

This gives us

TrK/L(k)=TrM/LTrK/M(k)
Step 5: Norm composition
det(mk)=i=1mj=1nτij(k)
=i=1m(j=1nτij(k))
=i=1mσi(NK/M(k))
=NM/L(NK/M(k))

That is,

NK/L(k)=NM/LNK/M(k)

Popular Posts