Galois Theory HT25, Galois groups and Galois extensions


Overview

Let $K/F$ be a field extension and let $G = \text{Aut} _ F(K)$. $K/F$ is called a Galois extension if any of these equivalent conditions are met:

  1. $K/F$ is the splitting field of a separable polynomial (taken as the definition in Prof. Konstantin Ardakov’s notes)
  2. $F = K^G$ (taken as the definition in Prof. Damian Rössler’s notes)
  3. $K/F$ is normal and separable
  4. $ \vert G \vert = [K : F]$

If $K/F$ is Galois, then we write $G = \text{Gal}(K/F)$, “the Galois group”. The proofs that these are equivalent are here:

-(3) implies (1)-(4) implies (2)-(2) implies (1)-(1) implies (4)- (1) implies (3):(1) implies normal,(1) implies separableThose proofs are sufficient to show that all the conditions are equivalent. However, if you wanted to prove that e.g. (2) implies (3), it would be inconvenient to show (2) implies (1) implies (3). Some of the other pairwise proofs are here:

-(2) implies (3)### Flashcards

Definition of a Galois group

Suppose:

  • $F \subset \mathbb C$
  • $K/F$ is a Galois extension

@Define the Galois group of $K/F$, written $\text{Gal}(K/F)$.


The group of all $F$-linear field automorphisms of $K$.

@Define an $F$-linear field automorphism of $K$ where $K/F$.


A bijective map $\theta : K \to K$ such that for all $k _ 1, k _ 2 \in K$, and $f \in F$:

\[\begin{aligned} \theta(k _ 1 k _ 2) &= \theta(k _ 1) \theta(k _ 2) \\\\ \theta(k _ 1 + k _ 2) &= \theta(k _ 1) + \theta(k _ 2) \\\\ \theta(f k _ 1) &= f \theta(k _ 1) \end{aligned}\]

Suppose:

  • $K/F$ is an extension
  • $H$ is a subgroup of $\text{Gal}(K/F)$

@Define the fixed field of $H$, written $K^H$.


\[K^H = \\{x \in K \mid \forall \sigma \in H,\text{ }\sigma(x) = x \\}\]

Splitting fields

Suppose:

  • $F \subset \mathbb C$ is a subfield
  • $f \in F[x]$ is a non-constant polynomial

@Define a splitting field of $f$ over $F$.


$K \supseteq F$ is said to be a splitting field of $f$ if:

  • $f$ splits completely in $K[t]$, and
  • $K$ is generated as a field by $F$ together with the roots of $f$.

Suppose:

  • $K/F$ is a field extension
  • $\alpha _ 1, \ldots, \alpha _ n \in K$ are algebraic over $F$

Then:

\[[F(\alpha _ 1, \ldots, \alpha _ n) : F] < \infty\]

@State a useful corollary of this fact related to splitting fields.


If $f \in F[t]$ and $K$ is a splitting field of $f$, then $[K : F] < \infty$.

Suppose:

  • $f \in F[t]$ is irreducible

@Prove that you can always find a (simple!) extension field $K \supset F$ which includes a root of $f$ (note that $F$ can’t contain any roots of $f$, or it wouldn’t be irreducible).


Since $f$ is irreducible in $F[t]$, $\langle f \rangle$ must be a maximal ideal in $F[t]$ and hence $K := \frac{F[t]}{\langle f\rangle}$ is a field. Then $\alpha := t + \langle f \rangle$ is a root of $f$ in $K$.

Since $\alpha$ generates $K$ as a ring together with $F$, $K = F[\alpha]$.

Since $\alpha$ is algebraic over $F$, it follows

\[F(\alpha) = F[\alpha] = K\]

and $K / F$ is the required extension.

Suppose:

  • $f \in F[t]$ be any polynomial

@Prove that there exists a splitting field $K$ of $f$.


Induct on $d := \deg f$.

Base case. $d = 1$. Then $F$ is already a splitting field.

Inductive step. $d > 1$. Suppose that $g$ is an irreducible factor of $f$. By the result that says:

If $g \in F[t]$ is irreducible, there exists a simple extension $F(\alpha)$ where $\alpha$ is a root of $g$.

we can construct $L := F(\alpha)$. Since $g \mid f$, $f(\alpha) = 0$ and hence $(t - \alpha) \mid f$.

Define $h := \frac{g}{(t - \alpha)} \in L[t]$. Since $\deg h < d$, by induction there exists a splitting field $K$ of $h$.

Since $a \in L \subset K$, $f = (t - \alpha) \cdot h$ splits completely in $K[t]$.

Since the roots of $f$ in $K$ generate $K$ together with $F$, it follows that $K$ is a splitting field of $f$.

Suppose:

  • $f \in F[t]$ is a polynomial

Is it true that any two splitting fields of $f$ are isomorphic?


Yes.

We have a result that says

If:

  • $\varphi : F \to \tilde F$ is an isomorphism of fields
  • $f \in F[t]$ is a separable polynomial
  • $K$ is a splitting field for $f$
  • $\tilde f = \varphi(f)$
  • $\tilde K$ is a splitting field for $\tilde f$

then there are at least $[K : F]$ distinct isomorphisms extending $\varphi$ to an isomorphism $K \to \tilde K$.

Use this to @prove that if $f \in F[t]$ is a separable polynomial, then any two splitting fields of $f$ are isomorphic.


Take $\varphi = \text{id} _ K$. Then since $[K : F] \ge 1$, there is at least one isomorphism $\varphi^\ast$ extending $\varphi$ to $K \to \tilde K$, and so they are isomorphic.

Suppose:

  • $f \in F[t]$ is a polynomial
  • $K, K’$ are splitting fields

@Prove (without assuming $F$ is separable!) that $K$ and $K’$ are isomorphic.


We induct on the degree of $f$.

If $\deg f = 1$, then there is nothing to prove.

If $\deg f > 1$, pick some root $a \in K$ of $f$ in $K$ and let $m _ {F, \alpha}$ be its minimal polynomial. Then $m _ {F, \alpha}$ splits in $K$ and in $K’$ (since it divides $P(x)$).

Let $\alpha _ 1$ be a root of $m _ {F, \alpha}$ in $K$. Notice that $K/F(\alpha)$ is a splitting extension of $f(x)/(x - \alpha) \in F(\alpha)$.

Similarly, if $K/F(\alpha _ 1)$ is a splitting extension of $f(x)/(x - \alpha _ 1)$ in $F(\alpha _ 1)$.

Let $J := F[x]/\langle m _ {F,\alpha}(x)\rangle$. The ring $J$ is a field, since $m _ {F, \alpha}$ is irreducible and furthermore, there are natural isomorphisms $J \cong F(\alpha)$ and $J \cong F(\alpha _ 1)$.

Consider the two $F$-extensions $K/J$ and $K’/J$ arising from these isomorphisms. By the inductive hypothesis, these two $J$-extensions are isomorphic. By construction, an isomorphism $K \cong K’$ of $J$-extensions is also an isomorphism of $F$-extensions, so we are done.


Alternatively, apply the result that says

If:

  • $\varphi : F \to \tilde F$ is an isomorphism between two fields
  • $K / F$ and $\tilde K / \tilde F$ are finite extensions
  • $\alpha \in K$, $\tilde \alpha \in \tilde K$
  • $\varphi(m _ {F, \alpha})(\tilde \alpha) = 0$

Then there is a unique extension of $\varphi$ to $\varphi^\ast$ where

\[\varphi^\ast : F(\alpha) \to \tilde F(\tilde \alpha)\]

such that $\varphi^\ast (\alpha) = \tilde \alpha$.

To each of the roots in the splitting field in turn.

@Prove that if:

  • $K/F$ is a splitting extension for $f(x)$
  • $L/F$ is an extension

then $\text{Im}(\varphi)$ is the same for every $F$-linear homomorphism $\varphi : K \to L$ (i.e. a map of $F$-extensions).


If there are no such homomorphisms, the statement is empty.

Suppose there is such a map $\varphi : K \to L$. Since $K$ is generated by $F$ by the roots of $f$, the image of $\varphi$ is generated over $F$ by the images of these roots in $L$ under $\varphi$. But the images are the roots of $f(x)$ in $L$.

To see this, let $\alpha _ 1, \ldots, \alpha _ d$ be the (possibly duplicate) roots of $f(x)$ in $K$. Then we have

\[f(x) = x^d - \sigma_1 (\alpha_1, \ldots, \alpha_d) x^{d-1} + \cdots + (-1)^d \sigma_d(\alpha_1, \ldots, \alpha_d)\]

Thus the elements $\varphi(\alpha _ 1), \cdots, \varphi(\alpha _ d)$ are roots of

\[\begin{aligned} &x^d - \sigma_d(\varphi(\alpha_1), \ldots, \varphi(\alpha_d)) x^{d-1} + \cdots + (-1)^d \sigma_d(\varphi(\alpha_1), \cdots, \varphi(\alpha_d)) \\\\ &=x^d - \varphi(\sigma(\alpha_1, \ldots, \alpha_d))x^{d-1} + \cdots + (-1)^d \varphi(\sigma_d(\alpha_1, \ldots, \alpha_d)) \\\\ &= f(x) \end{aligned}\]

since the coefficients of $f(x)$ lie in $F$.

Now the set of roots of $f(x)$ in $L$ does not depend on $\varphi$, hence the assertion.

Suppose:

  • $K = \mathbb F _ {125}(t)$
  • $f(x) = x^{124} - t \in K[x]$
  • $E$ is a splitting field for $f$ over $K$

Can you show that

\[E \cong \frac{K[x]}{\langle f\rangle}\]

?


Since any two splitting fields of a polynomial are isomorphic, this follows from showing that $\frac{K[x]}{\langle f \rangle}$ is a splitting field for $f$. Clearly $\frac{K[x]}{\langle f \rangle}$ contains a root $\alpha := \langle f \rangle$ of $f$, and note that if $u \in \mathbb F _ {125}^\times$, then $u\alpha$ is also a root (since $u^{124} = 1$). Since $ \vert \mathbb F _ {125}^\times \vert = 124$ and $u \in K$, it follows $\frac{K[x]}{\langle f \rangle}$ contains all the roots and is generated by them, so it is a splitting field.

@example~ @exam~

Galois extensions

Suppose:

  • $F$ is a field
  • $K / F$ is a finite field extension

@Define what it means for $K/F$ to be a Galois extension.


There is a separable polynomial $f \in F[x]$ such that $K/F$ is the splitting field of $f$.

Alternative characterisations of being Galois

Suppose:

  • $K/F$ is a finite extension
  • $G = \text{Gal}(K/F)$

@State 4 equivalent conditions to the condition that $K/F$ is Galois.


The following are equivalent:

  • $K/F$ is Galois (i.e. $K$ is the splitting field of a separable polynomial in $F[x]$)
  • $ \vert G \vert = [K : F]$
  • $F = K^G$
  • $K/F$ is normal and separable.

Suppose:

  • $K/F$ is a Galois extension
  • $G = \text{Gal}(K/F)$
  • $ \vert G \vert = [K : F]$

@State a result about the relationship between $F$, $K$ and $G$.


\[F = K^G\]

@Prove that if:

  • $K/F$ is a finite extension
  • $G = \text{Gal}(K/F)$
  • $ \vert G \vert = [K : F]$

Then:

\[F = K^G\]

This is a consequence of three other results:

  • $G = \text{Gal}(K/K^G)$
  • $ \vert G \vert \le [K : K^G]$ (applying the upper bound on the size of Galois groups to $\text{Gal}(K/K^G)$)
  • $[K : K^G] \le [K : F]$ by the tower law, since $K/K^G/F$.

Then:

\[|G| \le [K : K^G] \le [K : F] = |G|\]

and so $[K : K^G] = [K : F]$ and hence $[K^G : F] = 1$, and so $K^G = F$.

@important~ (main theorem)

Suppose:

  • $K/F$ is a finite field extension
  • $G = \text{Gal}(K/F)$
  • $F = K^G$

@State a result about what you can then conclude about $K/F$.


$K/F$ is Galois (i.e. $K$ is the splitting field of a separable polynomial in $F[x]$).

@Prove that if:

  • $K/F$ is a finite field extension
  • $G = \text{Gal}(K/F)$
  • $F = K^G$

Then:

  • $K/F$ is Galois (i.e. $K$ is the splitting field of a separable polynomial in $F[x]$).

Let $\{z _ 1, \ldots, z _ n\}$ be an $F$-basis for $K$. Since $G$ is finite (by the result that says finite field extensions lead to finite Galois groups), it follows that the set

\[X := \bigcup^n _ {i = 1} G\cdot z _ i\]

is finite as well as $G$-stable.

Then we can construct the polynomial $f _ X$ where

\[f _ X := \prod _ {y \in X} (t - y)\]

which is a separable polynomial with coefficients in $K^G$ (this is not immediate, but follows from considering the fact the coefficients are symmetric functions). Since $K^G = F$, it follows $f _ X \in F[t]$.

Since $f _ X$ splits completely over $K$ ($G \cdot z _ i \subseteq K$) and since $K$ is generated by the roots of $f _ X$ in $K$, it is the splitting field of $f _ X$ over $F$.

@important~

@Prove that if:

  • $K/F$ is a finite field extension
  • $G = \text{Gal}(K/F)$
  • $K/F$ is normal and separable

Then:

  • $K/F$ is Galois

::

Let $\alpha _ 1, \ldots, \alpha _ k$ be generators of $K$ over $F$. Define

\[f(x) = \prod^k_{i=1} m_{\alpha_i, K} (x)\]

Then $f$ is a separable polynomial (it is the product of irreducible polynomials, all of which are separable). But then $K$ is a splitting field for $f$, and so we are done.

A useful separable polynomial

Suppose:

  • $H$ is a finite group of automorphisms of a field $L$
  • $X \subseteq L$, $X$ is finite

@Define a polynomial $f _ X$ which satisfies the properties:

  1. $f _ X$ is separable.
  2. If $X$ is $H$-stable, then $f _ X$ has coefficients in $L^H$.

\[f _ X := \prod _ {y \in X} (t - y) \in L[t]\]

Useful for:

  • Property 1 is used in showing that $t^{p^n} - t$ is separable, and hence that $F$ is a splitting field of $t^{p^n} - t$ iff it has order $p^n$.
  • Both properties are used in showing that the minimal polynomial of $\alpha$ over $K/F$ is the product of $(x - g(\alpha))$ for each $g \in \text{Gal}(K/F)$.
  • Both properties are used in showing that $F = K^G$ implies $K/F$ is Galois.
  • Both properties are used in showing $F=K^G$ implies that $K/F$ is normal and separable.
  • Both properties are used in showing that the $n$-th cyclotomic polynomial lives in $\mathbb Q[t]$.

Suppose:

  • $H$ is a finite group of automorphisms of a field $L$
  • $X \subseteq L$, $X$ is finite
  • $f _ X(t) := \prod _ {y \in X} (t - y) \in L[t]$

@Prove that then:

  1. $f _ X$ is separable.
  2. If $X$ is $H$-stable, then $f _ X$ has coefficients in $L^H$.

Separable: Any factor (and hence any irreducible factor) of $f _ X$ must have the form $f _ Y$ for $Y \subseteq X$, and so it suffices to just check that $D(f _ X) \ne 0$. Note that

\[D(f _ X) = \sum _ {y \in X} \prod _ {z \in X \setminus \\{y\\} }(t - z)\]

Choose any value $u \in X$. If $y \ne u$, then $\prod _ {z \in X \setminus \{y\} } (u - z) = 0$ since the product includes the factor $u - z$ with $z = u$. Therefore

\[D(f _ X)(u) = \prod _ {z \in X \setminus \\{u\\} } (u - z) \ne 0\]

which is non-zero, and so $f _ X$ is separable.


If $X$ is $H$-stable, then $f _ X$ has coefficients in $L^H$: $H$ acts on $L[t]$ by applying each automorphism to the coefficients and fixing $t$.

Under the assumption $X$ is $H$-stable, the set of linear polynomials

\[\\{t - x : x \in X\\}\]

is $H$-stable. Since the $H$-action respects multiplication in $L[t]$, it follows that the product of these linear polynomials is fixed by $H$:

\[f _ X \in L[t]^H = L^H[t]\]

and so $f _ X$ has coefficients in $L^H$.

(Alternatively, can consider that the coefficients are given by symmetric functions of the roots).

@Prove that if:

  • $K/F$ is a finite field extension
  • $F = K^G$

then:

  • $K/F$ is normal and separable

(do this directly from the definition of $F = K^G$, rather than appealing to lots of other results)

::

$K/F$ is normal and separable is equivalent to showing that $m _ {\alpha, F}$ splits and is separable for all $\alpha \in K$. Fix some $\alpha \in K$. Let

\[f(x) := f_{G \cdot \alpha}(x) = \prod_{\beta \in G \cdot \alpha}(x - \beta)\]

By results about $f _ X$ in general, $f$ is separable, has coefficients in $K^G$ and clearly splits completely. Since $m _ {\alpha, F} \mid f$, it follows that $m _ {\alpha, F}$ is also separable and splits completely.

Abelian extensions

@Define what it means for a Galois extension $K/\mathbb Q$ to be abelian.


\[\text{Gal}(K / \mathbb Q)\]

is an abelian group.

Useful identification with a subgroup of $S _ n$

@Prove that if:

  • $f \in F[t]$ has $n$ distinct roots in a splitting field $K$

then:

  • $G = \text{Gal}(K/F)$ is naturally isomorphic to a subgroup of $S _ n$
  • Further, $f$ is irreducible then this subgroup is transitive

Naturally isomorphic to a subgroup of $S _ n$:

We show there is an injection $\rho : G \to S _ n$. Let $V(f)$ be the set of roots of $f$. Then the action of $G$ on $K$ restricts to an action on $V(f)$. Hence let

\[\rho : G \to \text{Sym}(V(f)) \cong S _ n\]

This map is injective, since if $\rho(\sigma) = \text{id}$ for some $\rho \in G$, then $\sigma$ fixes $V(f)$ pointwise. But $V(f)$ generates $K$ as a field together with $F$, so $\sigma$ fixes all elements of $K$ and hence $\sigma = 1$. Therefore

\[G \cong \rho(G) \le S _ n\]

If $f$ irreducible, then this subgroup is transitive:

If $f$ is irreducible, then $f = m _ {F, \alpha}$ where $\alpha \in V(f)$. Then applying the result

If

  • $\varphi : F \to \tilde F$ is an isomorphism between two fields
  • $K / F$ and $\tilde F / \tilde K$ are finite extensions
  • $\alpha \in K$, $\tilde \alpha \in \tilde K$
  • $\varphi(m _ {F, \alpha})(\tilde \alpha) = 0$

Then there is a unique extension of $\varphi$ to

\[\varphi^\ast : F(\alpha) \to \tilde F(\tilde \alpha)\]

such that $\varphi^\ast (\alpha) = \tilde \alpha$.

where $\varphi$ is the identity and $\tilde \alpha$ is any other root. This gives at least one automorphism of $K$ that swaps $\alpha$ and $\tilde \alpha$, so $G$ is transitive.

Relationship between the Galois group of a field extension and the Galois groups of minimal polynomials of elements of that extension

@State three results concerning the minimal polynomial of any element of a Galois extension.


Suppose:

  • $K/F$ is a Galois extension
  • $\alpha \in K$

Then:

  • $m _ {F, \alpha}$ is separable
  • $m _ {F, \alpha} = \prod _ {\psi \in \text{Gal}(K/F) \cdot \alpha} (x - \psi)$
  • There is a surjective group homomorphism $\text{Gal}(K/F) \to \text{Gal} _ F(m _ {F, \alpha})$

(This result is useful mainly because it comes up in the proof that field extensions with solvable Galois groups are radical. Also, since $K/F$ is defined to be a separable extension whenever the minimal polynomial of every $\alpha \in K$ is separable, this also shows that all Galois extensions are separable).

@Prove that if:

  • $K/F$ is a Galois extension
  • $\alpha \in K$

Then:

  • $m _ {F, \alpha}$ is separable, and
  • $m _ {F, \alpha} = \prod _ {\psi \in \text{Gal}(K/F) \cdot \alpha} (x - \psi)$
  • There is a surjective group homomorphism $\text{Gal}(K/F) \to \text{Gal} _ F(m _ {F, \alpha})$

(You may assume previous results).


  • $m _ {F, \alpha}$ is irreducible over $F$ and has a zero in $K$ (namely $\alpha$)
  • Since $K/F$ is Galois, it is particular normal.
  • Therefore $m _ {F, \alpha}$ splits completely in $K$, and so $K$ contains a splitting field of $m _ {F, \alpha}$, say $L$.
  • By another result, $m _ {F, \alpha} = m _ {K^G, \alpha} = f _ {G \cdot \alpha} = \prod _ {\beta \in G\cdot\alpha}(t - \beta)$.
  • Since $f _ X$ is always separable, it follows $m _ {F, \alpha}$ is always separable.

  • Therefore $L/F$ is Galois (it is the splitting field of some separable polynomial).
  • $\text{Gal} _ F(m _ {F, \alpha}) = \text{Gal}(L/F)$ and so the restriction map $\text{Gal}(K/F) \to \text{Gal}(L/F)$ is a surjective group homomorphism.

Composite field extensions

Suppose:

  • $K \supseteq L _ 1, L _ 2 \supseteq F$
  • $L _ 1/F$ and $L _ 2/F$ are Galois extensions
  • $L$ is the smallest subfield of $K$ containing $L _ 1$ and $L _ 2$ ($L = L _ 1 L _ 2$, their “compositum”)

@State a result about how the Galois groups of each of these extensions relate to one another.


  • $L/F$ is also Galois
  • There is an injective homomorphism $\phi : \text{Gal}(L/F) \to \text{Gal}(L _ 1/F) \times \text{Gal}(L _ 2 / F)$
  • If $L _ 1 \cap L _ 2 = F$ (“linearly disjoint”), then this homomorphism is an isomorphism

@exam~

Suppose:

  • $K \supseteq L _ 1, L _ 2 \supseteq F$
  • $L _ 1/F$ and $L _ 2/F$ are Galois extensions
  • $L$ is the smallest subfield of $K$ containing $L _ 1$ and $L _ 2$ ($L = L _ 1 L _ 2$, their “compositum”)

@Prove:

  • $L/F$ is also Galois
  • There is an injective homomorphism $\phi : \text{Gal}(L/F) \to \text{Gal}(L _ 1/F) \times \text{Gal}(L _ 2 / F)$
  • If $L _ 1 \cap L _ 2 = F$ (“linearly disjoint”), then this homomorphism is an isomorphism

$L/F$ is Galois:

$L _ 1$ and $L _ 2$ are the splitting fields of separable polynomials $f _ 1, f _ 2 \in F[x]$. Then $F$ is the splitting field of $f _ 1 f _ 2$, which is also separable. Hence $L/F$ is also Galois.


There is an injective homomorphism $\phi : \text{Gal}(L/F) \to \text{Gal}(L _ 1/F) \times \text{Gal}(L _ 2 / F)$:

Let $G := \text{Gal}(L/F)$ and $H := \text{Gal}(L _ 1/F) \times \text{Gal}(L _ 2 / F)$. Let $\phi : G \to H$ be the map which sends $\sigma \to (\sigma \vert _ {L _ 1}, \sigma \vert _ {L _ 2})$. This map is well-defined as $L _ 1 / F$ and $L _ 2/F$ are both Galois extensions contained in $K$.

To see it is injective, suppose $\phi(\sigma) = (\text{id} _ {L _ 1}, \text{id} _ {L _ 2})$. Since $L$ is generated by the union of the generators for $L _ 1$ and $L _ 2$, it must also be the identity on $L$.


If $L _ 1 \cap L _ 2 = F$, then this homomorphism is an isomorphism:

It suffices to show that $ \vert G \vert = \vert H \vert $. Let $\iota$ be the restriction map

\[\iota : \text{Gal}(L / L _ 2) \to \text{Gal}(L _ 1 / F)\]

which sends $\sigma$ to $\sigma \vert _ {L _ 1}$. Since $L _ 2$ contains $F$, the image of $\iota$ fixes $F$. Since $L _ 1$ is Galois, $\iota$ is well-defined and it is injective since any element of its kernel must be the identity on both $L _ 1$ and $L _ 2$.

To see $\iota$ is also surjective, we show that

\[[F : L _ 2] = |S| = |\text{Gal}(L _ 1 / K)| = [L _ 1 : K]\]

as then the size of the domain and range are equal. Consider the subgroup $\iota(\text{Gal}(L/L _ 2))$. Then

\[[L _ 1 : L _ 1^\iota(\text{Gal}(L/L _ 2))] = |\text{Gal}(L/L _ 2)|\]

as the action of $\iota(\text{Gal}(L/L _ 2))$ on $L _ 1$ is faithful. If we can show $L _ 1^{\iota(\text{Gal}(L/L _ 2))} = L _ 1 \cap L _ 2$, then we will be done as $L _ 1 \cap L _ 2 = F$.

Note

\[\\{x \in L : g(x) = x \text{ } \forall g \in \text{Gal}(L/L _ 2)\\} = L _ 2\]

by the Galois correspondence on $L/F$. Therefore

\[L _ 1^{\iota(\text{Gal}(L/L _ 2))} = \\{x \in L _ 1 : g(x) = x \text{ } \forall g \in \text{Gal}(F/L _ 2)\\} = L _ 1 \cap L _ 2\]

and hence

\[|H| = |\text{Gal}(L _ 1 / F)| \times |\text{Gal}(L _ 2 / F)| = [L _ 1 : F][L _ 2 : F] = [L : L _ 2][L _ 2 : F] = [L : F] = |G|\]

as required.

@exam~

Galois groups of rational polynomials with mostly real roots

@State a result which gives a test for when the Galois group of a prime-degree polynomial will be the full symmetric group $S _ p$.


Suppose:

  • $f$ is a degree $p$ polynomial where $p$ prime
  • $f$ has exactly $p - 2$ real roots

Then:

  • $\text{Gal} _ \mathbb Q(f) \cong S _ p$

@Justify why if

  • $f$ is a degree $p$ polynomial where $p$ prime
  • $f$ has exactly $p - 2$ real roots

Then:

  • $\text{Gal} _ \mathbb Q(f) \cong S _ p$

  • The Galois group must contain a transposition corresponding to complex conjugation
  • The Galois group must also contain a $p$-cycle by Cauchy’s theorem
  • Therefore the Galois group contains the whole symmetric group

Examples and counterexamples

Give an @example of a field extension that is not Galois because of normality reasons.


Consider $\mathbb Q[\alpha] / \mathbb Q$ where $\alpha = \sqrt[3]{2}$. Then $m _ \mathbb Q(x) = x^3 - 2$, but $m(x)$ does not split completely, since it has two roots which are not in $\mathbb Q[\alpha]$.

Give an @example of a field extension that is not Galois because of separability reasons.


Consider $F := \mathbb F _ p(t)$ be the field of fractions over the finite field of size $p$, and let $K$ be a field extension of $F$ containing a root $\alpha$ of $f(x) := x^p - t$. But then $K / F$ is not Galois as it is not separable, since $m _ {F, \alpha}$ is not separable.

Give an @example to show that if you have the sequence of extensions $K ‘ / K / F$ where $K/F$ and $K’/K$ are both Galois, it doesn’t need to be the case that $K’ / F$ is Galois.


  • Let $F = \mathbb Q$, $K = \mathbb Q(\sqrt 2)$, $K’ = \mathbb Q(2^{1/4})$.
  • Then:
    • $K/F$ is Galois (it’s the splitting field of $x^2 - 2$)
    • $K’/F$ is Galois (it’s the splitting field of $x^2 - \sqrt{2}$)
    • $K’/F$ is not Galois, since $m _ {\mathbb Q, 2^{1/4} } = x^4 - 2$, which does not split over $K’ \subseteq \mathbb R$.

@exam~

@Prove every degree 2 extension $L/K$ where $K$ is a field of odd characteristic is a Galois extension.


Pick any $\alpha \in L \setminus K$. Then $m _ {K, \alpha}(x) = x^2 + bx + c$ for some $b, c \in K$ and this polynomial is irreducible. Since $m _ {K, \alpha}’(x) = 2x + b$, it follows that $m _ {K, \alpha}$ is separable.

We aim to show that $L$ is a splitting field for $m _ {K, \alpha}$. By the quadratic formula,

\[\begin{aligned} m _ {K, \alpha}(x) &= \left(x - \frac{-b +\sqrt{b^2 - 4c} }{2}\right) \left(x - \frac{-b - \sqrt{b^2 - 4c} }{2}\right) \\\\ &= (x - \alpha)(x - (\alpha - b)) \end{aligned}\]

Since $\alpha, \alpha - b \in L$ and $L = F(\alpha)$, so $L$ is the splitting field.

@exam~ @example~




Related posts