Properties of the Separable Closure (Theorem # 3334)
Theorem
Let $k$ be a field. Then:
(i) $k^{\mathrm{sep}}/k$ is a Galois extension.
(ii) $k^{\mathrm{sep}} = \overline{k}$ if and only if $k$ is perfect.
(iii) The extension $\overline{k} / k^{\mathrm{sep}}$ is purely inseparable: every element of $\overline{k}$ is purely inseparable over $k^{\mathrm{sep}}$.
(iv) The absolute Galois group $G_k = \operatorname{Gal}(k^{\mathrm{sep}}/k)$ is a profinite group, and the Galois correspondence gives a bijection between closed subgroups of $G_k$ and intermediate fields $k \subset F \subset k^{\mathrm{sep}}$.
Group Theory
Algebra
Abstract Algebra
Group Theory
Discussion
No discussion available for this theorem.
Proof
[proofplan]
We prove each part of the theorem separately. For (i), we show that $k^{\mathrm{sep}}$ is both normal and separable over $k$ by expressing it as a directed union of splitting fields of separable polynomials. For (ii), we use the [Classification of Perfect Fields](/theorems/3324) to relate the Frobenius surjectivity condition to whether every algebraic element is separable. For (iii), we take an arbitrary $\alpha \in \overline{k}$ and show that its minimal polynomial over $k^{\mathrm{sep}}$ has only one distinct root, using the [Decomposition into Separable and Purely Inseparable Parts](/theorems/3332). For (iv), we invoke the [Galois Group as Inverse Limit](/theorems/2362) to realise $G_k$ as an inverse limit of finite groups, and then apply the [Fundamental Theorem of Infinite Galois Theory](/theorems/2387).
[/proofplan]
[step:Show that $k^{\mathrm{sep}}/k$ is separable]
By definition, $k^{\mathrm{sep}}$ is the set of all elements in $\overline{k}$ that are separable over $k$. Let $\alpha \in k^{\mathrm{sep}}$. Then the minimal polynomial $\operatorname{min}(\alpha, k) \in k[x]$ is separable, so $\alpha$ is separable over $k$. Since every element of $k^{\mathrm{sep}}$ is separable over $k$, the extension $k^{\mathrm{sep}}/k$ is separable.
We must verify that $k^{\mathrm{sep}}$ is indeed a field. Let $\alpha, \beta \in k^{\mathrm{sep}}$. The extension $k(\alpha, \beta)/k$ is finite and generated by separable elements. By the [Characterisation of Separable Extensions](/theorems/1265), $k(\alpha, \beta)/k$ is separable, so every element of $k(\alpha, \beta)$ is separable over $k$. In particular, $\alpha + \beta$, $\alpha \beta$, and (when $\alpha \neq 0$) $\alpha^{-1}$ all lie in $k^{\mathrm{sep}}$. Thus $k^{\mathrm{sep}}$ is a subfield of $\overline{k}$.
[guided]
The separable closure $k^{\mathrm{sep}}$ is defined as the set of all elements of $\overline{k}$ that are separable over $k$. We need to verify two things: that $k^{\mathrm{sep}}$ is a field, and that $k^{\mathrm{sep}}/k$ is separable.
For the field property, the concern is closure under arithmetic. Take $\alpha, \beta \in k^{\mathrm{sep}}$. Each is separable over $k$, but why should $\alpha + \beta$ be separable over $k$? The key is that $k(\alpha, \beta)/k$ is a finite extension generated by elements whose minimal polynomials over $k$ are separable. The [Characterisation of Separable Extensions](/theorems/1265) tells us that a finite extension generated by elements with separable minimal polynomials is itself separable, meaning every element of $k(\alpha, \beta)$ is separable over $k$. Since $\alpha + \beta, \alpha\beta, \alpha^{-1} \in k(\alpha, \beta)$, they all lie in $k^{\mathrm{sep}}$.
That $k^{\mathrm{sep}}/k$ is separable is then immediate from the definition: every element of $k^{\mathrm{sep}}$ has a separable minimal polynomial over $k$.
[/guided]
[/step]
[step:Show that $k^{\mathrm{sep}}/k$ is normal]
We must show that $k^{\mathrm{sep}}/k$ is a normal extension, i.e., for every $\alpha \in k^{\mathrm{sep}}$, the minimal polynomial $\operatorname{min}(\alpha, k)$ splits completely in $k^{\mathrm{sep}}[x]$.
Let $\alpha \in k^{\mathrm{sep}}$ and let $f = \operatorname{min}(\alpha, k) \in k[x]$. Since $\alpha$ is separable over $k$, the polynomial $f$ is separable. Let $\alpha_1, \ldots, \alpha_n$ be the roots of $f$ in $\overline{k}$. Since $f$ is separable, these roots are distinct, so $\deg f = n$. Each $\alpha_i$ is a root of the same irreducible polynomial $f \in k[x]$, so $\operatorname{min}(\alpha_i, k) = f$, which is separable. Therefore each $\alpha_i \in k^{\mathrm{sep}}$, and $f$ splits completely in $k^{\mathrm{sep}}[x]$.
Since $k^{\mathrm{sep}}/k$ is both separable and normal, it is Galois by the [Equivalent Characterisations of Galois Extensions](/theorems/3310) (applied to every finite sub-extension: each finite sub-extension $L/k$ with $L \subset k^{\mathrm{sep}}$ is separable, and the normality of $k^{\mathrm{sep}}/k$ passes to $L/k$ when $L$ is the splitting field of a separable polynomial). More precisely, $k^{\mathrm{sep}} = \bigcup_{\alpha \in k^{\mathrm{sep}}} K_\alpha$, where $K_\alpha$ is the splitting field of $\operatorname{min}(\alpha, k)$ over $k$. Each $K_\alpha/k$ is a finite Galois extension by the [Equivalent Characterisations of Galois Extensions](/theorems/3310), since $K_\alpha$ is the splitting field of a separable polynomial. The union of a directed family of finite Galois extensions is Galois, completing the proof of part (i).
[guided]
We need to show normality: for every $\alpha \in k^{\mathrm{sep}}$, all conjugates of $\alpha$ over $k$ also lie in $k^{\mathrm{sep}}$.
Take $\alpha \in k^{\mathrm{sep}}$ and let $f = \operatorname{min}(\alpha, k)$. By definition of $k^{\mathrm{sep}}$, $f$ is separable. Now consider any root $\beta$ of $f$ in $\overline{k}$. Since $\beta$ is also a root of $f$, and $f$ is irreducible over $k$, we have $\operatorname{min}(\beta, k) = f$. The polynomial $f$ is separable, so $\beta$ is separable over $k$, hence $\beta \in k^{\mathrm{sep}}$.
This shows that $f$ splits completely in $k^{\mathrm{sep}}[x]$, which is exactly the normality condition.
Now, $k^{\mathrm{sep}}/k$ is both normal and separable. For a finite extension, the [Equivalent Characterisations of Galois Extensions](/theorems/3310) would immediately give us that it is Galois. For the possibly infinite extension $k^{\mathrm{sep}}/k$, we argue as follows. For each $\alpha \in k^{\mathrm{sep}}$, the splitting field $K_\alpha$ of $\operatorname{min}(\alpha, k)$ over $k$ is contained in $k^{\mathrm{sep}}$ (since all roots are separable), and $K_\alpha/k$ is a finite Galois extension by the [Equivalent Characterisations of Galois Extensions](/theorems/3310) (it is the splitting field of a separable polynomial). Since $k^{\mathrm{sep}} = \bigcup_\alpha K_\alpha$ and this union is directed (given $K_\alpha, K_\beta$, both are contained in $K_\gamma$ where $\gamma$ is a primitive element of $K_\alpha K_\beta / k$, which exists by the [Primitive Element Theorem](/theorems/1267) since the composite of separable extensions is separable), $k^{\mathrm{sep}}/k$ is a directed union of finite Galois extensions, hence Galois.
[/guided]
[/step]
[step:Prove that $k^{\mathrm{sep}} = \overline{k}$ if and only if $k$ is perfect]
($\Rightarrow$) Suppose $k^{\mathrm{sep}} = \overline{k}$. Then every element of $\overline{k}$ is separable over $k$, which means every irreducible polynomial in $k[x]$ is separable. If $\operatorname{char}(k) = 0$, then $k$ is perfect by the [Classification of Perfect Fields](/theorems/3324). If $\operatorname{char}(k) = p > 0$, suppose for contradiction that the Frobenius $\varphi_p: k \to k$, $x \mapsto x^p$, is not surjective. Then there exists $a \in k$ with $a \notin k^p$. The polynomial $x^p - a \in k[x]$ is irreducible (if it factored, say $x^p - a = g(x) h(x)$ with $\deg g = d$, $0 < d < p$, then in $\overline{k}$ we have $x^p - a = (x - \alpha)^p$ where $\alpha^p = a$, so the constant term of $g$ would be $\pm \alpha^d$, giving $\alpha^d \in k$; since $\gcd(d, p) = 1$, Bezout gives integers $r, s$ with $rd + sp = 1$, so $\alpha = \alpha^{rd + sp} = (\alpha^d)^r \cdot (\alpha^p)^s = (\alpha^d)^r a^s \in k$, contradicting $a = \alpha^p \notin k^p$ since $\alpha \in k$ would give $a \in k^p$). But $x^p - a = (x - \alpha)^p$ in $\overline{k}[x]$, so $x^p - a$ has a repeated root and is inseparable. This contradicts the assumption that every irreducible polynomial is separable. Therefore $\varphi_p$ is surjective, and $k$ is perfect by the [Classification of Perfect Fields](/theorems/3324).
($\Leftarrow$) Suppose $k$ is perfect. We show every $\alpha \in \overline{k}$ is separable over $k$. Let $f = \operatorname{min}(\alpha, k)$. By the [Criterion for Inseparability of Irreducible Polynomials](/theorems/3330), if $f$ is inseparable then $\operatorname{char}(k) = p > 0$ and $f(x) = g(x^p)$ for some $g \in k[x]$. Write $g(x) = \sum_{i=0}^m c_i x^i$. Since $k$ is perfect, the Frobenius is surjective, so each $c_i = b_i^p$ for some $b_i \in k$. Then $f(x) = \sum_{i=0}^m b_i^p x^{ip} = \left(\sum_{i=0}^m b_i x^i\right)^p$, contradicting the irreducibility of $f$ (since $\deg f = mp > m$). Therefore $f$ is separable, $\alpha \in k^{\mathrm{sep}}$, and $k^{\mathrm{sep}} = \overline{k}$.
[guided]
We prove both directions of part (ii).
**Forward direction.** Assume $k^{\mathrm{sep}} = \overline{k}$, meaning every algebraic element over $k$ is separable. We want to show $k$ is perfect. If $\operatorname{char}(k) = 0$, the [Classification of Perfect Fields](/theorems/3324) immediately gives perfectness. The interesting case is $\operatorname{char}(k) = p > 0$. We need to show the Frobenius $\varphi_p: k \to k$, $x \mapsto x^p$, is surjective.
Suppose for contradiction that some $a \in k$ has no $p$-th root in $k$, i.e., $a \notin k^p$. Consider the polynomial $f(x) = x^p - a \in k[x]$. In the algebraic closure, if $\alpha \in \overline{k}$ satisfies $\alpha^p = a$, then $x^p - a = x^p - \alpha^p = (x - \alpha)^p$ (using the Frobenius identity $(x - \alpha)^p = x^p - \alpha^p$ in characteristic $p$). So $f$ has $\alpha$ as a root of multiplicity $p$, making it inseparable.
We verify that $f$ is irreducible over $k$. If $f = g \cdot h$ with $1 \le d = \deg g < p$, then since $f = (x - \alpha)^p$ in $\overline{k}[x]$, we have $g(x) = (x - \alpha)^d$. The constant term of $g$ is $(-\alpha)^d \in k$, so $\alpha^d \in k$. Since $d < p$ and $p$ is prime, $\gcd(d, p) = 1$. By Bezout's identity, there exist integers $r, s$ with $rd + sp = 1$. Then $\alpha = \alpha^{rd + sp} = (\alpha^d)^r \cdot (\alpha^p)^s$. We have $\alpha^d \in k$ and $\alpha^p = a \in k$, so $\alpha \in k$, giving $a = \alpha^p \in k^p$, a contradiction. So $f$ is irreducible and inseparable, contradicting $k^{\mathrm{sep}} = \overline{k}$.
**Backward direction.** Assume $k$ is perfect. Take any $\alpha \in \overline{k}$ and let $f = \operatorname{min}(\alpha, k)$. We want to show $f$ is separable. By the [Criterion for Inseparability of Irreducible Polynomials](/theorems/3330), if $f$ were inseparable, then $\operatorname{char}(k) = p > 0$ and $f(x) = g(x^p)$ for some $g \in k[x]$. Write $g(x) = \sum_{i=0}^m c_i x^i$. Since $k$ is perfect, the Frobenius is surjective by the [Classification of Perfect Fields](/theorems/3324), so each $c_i = b_i^p$ for some $b_i \in k$. Then:
\begin{align*}
f(x) = \sum_{i=0}^m b_i^p \, x^{ip} = \left(\sum_{i=0}^m b_i \, x^i\right)^p
\end{align*}
where the last equality uses the Frobenius identity $(\sum b_i x^i)^p = \sum b_i^p x^{ip}$ in characteristic $p$. This contradicts the irreducibility of $f$, since $\deg g = m < mp = \deg f$. Therefore every irreducible polynomial over $k$ is separable, so every element of $\overline{k}$ is separable over $k$, giving $k^{\mathrm{sep}} = \overline{k}$.
[/guided]
[/step]
[step:Show that $\overline{k} / k^{\mathrm{sep}}$ is purely inseparable]
Let $\alpha \in \overline{k}$ and let $f = \operatorname{min}(\alpha, k^{\mathrm{sep}}) \in k^{\mathrm{sep}}[x]$. We must show that $f$ has only one distinct root in $\overline{k}$, i.e., $f(x) = (x - \alpha)^m$ for some $m \ge 1$.
If $\operatorname{char}(k) = 0$, then $k$ is perfect, so $k^{\mathrm{sep}} = \overline{k}$ by part (ii), and the statement is vacuous (every element of $\overline{k}$ already lies in $k^{\mathrm{sep}}$).
Assume $\operatorname{char}(k) = p > 0$. Consider the finite extension $k(\alpha)/k$. By the [Decomposition into Separable and Purely Inseparable Parts](/theorems/3332), there exists a unique intermediate field $F$ with $k \subset F \subset k(\alpha)$ such that $F/k$ is separable and $k(\alpha)/F$ is purely inseparable. Since $F/k$ is separable, $F \subset k^{\mathrm{sep}}$.
Since $k(\alpha)/F$ is purely inseparable, there exists $e \ge 0$ such that $\alpha^{p^e} \in F \subset k^{\mathrm{sep}}$. Set $\beta = \alpha^{p^e} \in k^{\mathrm{sep}}$. The polynomial $x^{p^e} - \beta = (x - \alpha)^{p^e}$ in $\overline{k}[x]$ (using the Frobenius identity in characteristic $p$), and $\alpha$ is a root. The minimal polynomial $\operatorname{min}(\alpha, k^{\mathrm{sep}})$ divides $x^{p^e} - \beta = (x - \alpha)^{p^e}$, so $\operatorname{min}(\alpha, k^{\mathrm{sep}}) = (x - \alpha)^{p^r}$ for some $0 \le r \le e$. This polynomial has $\alpha$ as its only root, so $\alpha$ is purely inseparable over $k^{\mathrm{sep}}$.
[guided]
We want to show that every element of $\overline{k}$ is purely inseparable over $k^{\mathrm{sep}}$. Recall that $\alpha$ is purely inseparable over a field $L$ if and only if $\operatorname{min}(\alpha, L)$ has only one distinct root in $\overline{k}$.
If $\operatorname{char}(k) = 0$, then $k$ is perfect, so by part (ii) $k^{\mathrm{sep}} = \overline{k}$, and there is nothing to prove: every element of $\overline{k}$ already lies in $k^{\mathrm{sep}}$ and has minimal polynomial $x - \alpha$.
The substantive case is $\operatorname{char}(k) = p > 0$. Take $\alpha \in \overline{k}$. The idea is to use the separable-inseparable decomposition of $k(\alpha)/k$.
By the [Decomposition into Separable and Purely Inseparable Parts](/theorems/3332), applied to the finite extension $k(\alpha)/k$, there is a unique intermediate field $F$ with $k \subset F \subset k(\alpha)$ such that $F/k$ is separable and $k(\alpha)/F$ is purely inseparable. Since $F/k$ is separable, every element of $F$ is separable over $k$, so $F \subset k^{\mathrm{sep}}$.
Now, $k(\alpha)/F$ is purely inseparable, so there exists $e \ge 0$ with $\alpha^{p^e} \in F$. Set $\beta = \alpha^{p^e}$. Then $\beta \in F \subset k^{\mathrm{sep}}$. Consider the polynomial $x^{p^e} - \beta \in k^{\mathrm{sep}}[x]$. In characteristic $p$, the Frobenius identity gives:
\begin{align*}
x^{p^e} - \beta = x^{p^e} - \alpha^{p^e} = (x - \alpha)^{p^e}.
\end{align*}
Since $\alpha$ is a root of $x^{p^e} - \beta$ and this polynomial lies in $k^{\mathrm{sep}}[x]$, the minimal polynomial $\operatorname{min}(\alpha, k^{\mathrm{sep}})$ divides $(x - \alpha)^{p^e}$. Every divisor of $(x - \alpha)^{p^e}$ has $\alpha$ as its only root, so $\operatorname{min}(\alpha, k^{\mathrm{sep}})$ has only one distinct root, confirming that $\alpha$ is purely inseparable over $k^{\mathrm{sep}}$.
Note: if $\alpha$ were already in $k^{\mathrm{sep}}$, then $e = 0$, $\beta = \alpha$, and $\operatorname{min}(\alpha, k^{\mathrm{sep}}) = x - \alpha$. The purely inseparable condition is trivially satisfied with inseparable degree $1$.
[/guided]
[/step]
[step:Establish that $G_k$ is profinite via the inverse limit representation]
By part (i), $k^{\mathrm{sep}}/k$ is a Galois extension. Let $\mathcal{I}$ denote the directed set of all finite Galois sub-extensions $L/k$ with $L \subset k^{\mathrm{sep}}$, ordered by inclusion. This set is directed: given $L_1, L_2 \in \mathcal{I}$, the composite $L_1 L_2$ is a finite Galois extension of $k$ (as the composite of finite Galois extensions is Galois) contained in $k^{\mathrm{sep}}$, so $L_1 L_2 \in \mathcal{I}$.
By the [Galois Group as Inverse Limit](/theorems/2362), the restriction maps $\operatorname{Gal}(L'/k) \to \operatorname{Gal}(L/k)$ for $L \subset L'$ in $\mathcal{I}$ form an inverse system, and
\begin{align*}
G_k = \operatorname{Gal}(k^{\mathrm{sep}}/k) \cong \varprojlim_{L/k \in \mathcal{I}} \operatorname{Gal}(L/k)
\end{align*}
as topological groups, where each $\operatorname{Gal}(L/k)$ carries the discrete topology. Each $\operatorname{Gal}(L/k)$ is a finite group (by the [Order of the Galois Group](/theorems/3325), $|\operatorname{Gal}(L/k)| = [L:k] < \infty$). An inverse limit of finite groups with the inverse limit topology is, by definition, a profinite group. By the [Compactness of the Galois Group](/theorems/2386), $G_k$ is compact and Hausdorff in the Krull topology, confirming the profinite structure.
[guided]
A profinite group is, by definition, a compact, Hausdorff, totally disconnected topological group, or equivalently, an inverse limit of finite groups. We show $G_k$ is profinite by realising it as such an inverse limit.
Since $k^{\mathrm{sep}}/k$ is Galois (part (i)), every $\alpha \in k^{\mathrm{sep}}$ has a separable minimal polynomial over $k$ that splits in $k^{\mathrm{sep}}$. The splitting field of $\operatorname{min}(\alpha, k)$ is a finite Galois extension of $k$ contained in $k^{\mathrm{sep}}$. Let $\mathcal{I}$ be the collection of all such finite Galois sub-extensions $L/k$. This forms a directed set under inclusion: given $L_1, L_2 \in \mathcal{I}$, the composite $L_1 L_2$ is finite (as both $L_1$ and $L_2$ are finite over $k$), normal (as the composite of normal extensions is normal), and separable (by the [Multiplicativity of Separable Degree](/theorems/3331)), hence Galois by the [Equivalent Characterisations of Galois Extensions](/theorems/3310).
The [Galois Group as Inverse Limit](/theorems/2362) applies to the Galois extension $k^{\mathrm{sep}}/k$ with this directed system. For each pair $L \subset L'$ in $\mathcal{I}$, the restriction homomorphism $\operatorname{Gal}(L'/k) \to \operatorname{Gal}(L/k)$, $\sigma \mapsto \sigma|_L$, is well-defined and surjective. The theorem gives:
\begin{align*}
G_k = \operatorname{Gal}(k^{\mathrm{sep}}/k) \cong \varprojlim_{L/k \in \mathcal{I}} \operatorname{Gal}(L/k)
\end{align*}
as topological groups. Each $\operatorname{Gal}(L/k)$ is finite by the [Order of the Galois Group](/theorems/3325), with $|\operatorname{Gal}(L/k)| = [L:k]$. An inverse limit of finite discrete groups, equipped with the product topology (restricted to the inverse limit), is compact by Tychonoff and Hausdorff, hence profinite. The [Compactness of the Galois Group](/theorems/2386) confirms that $G_k$ is compact and Hausdorff in the Krull topology.
[/guided]
[/step]
[step:Derive the Galois correspondence for intermediate fields of $k^{\mathrm{sep}}/k$]
Since $k^{\mathrm{sep}}/k$ is a Galois extension (part (i)) and $G_k = \operatorname{Gal}(k^{\mathrm{sep}}/k)$ is profinite (as just shown), the [Fundamental Theorem of Infinite Galois Theory](/theorems/2387) applies. It gives a bijection
\begin{align*}
\{F : k \subset F \subset k^{\mathrm{sep}}\} &\longleftrightarrow \{H \le G_k : H \text{ is closed}\} \\
F &\longmapsto \operatorname{Gal}(k^{\mathrm{sep}}/F) \\
(k^{\mathrm{sep}})^H &\longmapsfrom H.
\end{align*}
We verify the hypotheses of the [Fundamental Theorem of Infinite Galois Theory](/theorems/2387): $k^{\mathrm{sep}}/k$ is Galois by part (i). The theorem requires no further conditions beyond the extension being Galois.
Moreover, an intermediate field $F/k$ is finite if and only if $\operatorname{Gal}(k^{\mathrm{sep}}/F)$ is open in $G_k$, and $F/k$ is Galois if and only if $\operatorname{Gal}(k^{\mathrm{sep}}/F)$ is normal in $G_k$, in which case $\operatorname{Gal}(F/k) \cong G_k / \operatorname{Gal}(k^{\mathrm{sep}}/F)$.
This completes the proof of part (iv) and of the theorem.
[guided]
The [Fundamental Theorem of Infinite Galois Theory](/theorems/2387) is the infinite analogue of the [Fundamental Theorem of Galois Theory](/theorems/1274). The finite version gives a bijection between all subgroups of $\operatorname{Gal}(K/k)$ and intermediate fields. For infinite Galois extensions, one must restrict to **closed** subgroups (in the Krull topology) to maintain the bijection. Not every subgroup of the profinite group $G_k$ is closed — for instance, the commutator subgroup of a profinite group may not be closed — and the subgroups that fail to be closed do not correspond to any intermediate field.
The theorem applies directly to $k^{\mathrm{sep}}/k$ because this extension is Galois (part (i)). The correspondence is:
\begin{align*}
F &\longmapsto \operatorname{Gal}(k^{\mathrm{sep}}/F), \\
(k^{\mathrm{sep}})^H &\longmapsfrom H.
\end{align*}
These maps are mutually inverse: $(k^{\mathrm{sep}})^{\operatorname{Gal}(k^{\mathrm{sep}}/F)} = F$ (which is the content of the Galois correspondence), and $\operatorname{Gal}(k^{\mathrm{sep}} / (k^{\mathrm{sep}})^H) = H$ for $H$ closed (which fails for non-closed $H$, since $\operatorname{Gal}(k^{\mathrm{sep}} / (k^{\mathrm{sep}})^H) = \overline{H}$, the closure of $H$).
The additional statements about finiteness and normality are part of the [Fundamental Theorem of Infinite Galois Theory](/theorems/2387): $F/k$ is finite if and only if $\operatorname{Gal}(k^{\mathrm{sep}}/F)$ has finite index in $G_k$, which, in a profinite group, is equivalent to being open. The extension $F/k$ is Galois if and only if $\operatorname{Gal}(k^{\mathrm{sep}}/F)$ is normal, yielding the quotient isomorphism $G_k / \operatorname{Gal}(k^{\mathrm{sep}}/F) \cong \operatorname{Gal}(F/k)$.
[/guided]
[/step]
Explore Further
Equivalent Characterisations of Galois Extensions
Galois Theory
Classification of Finite Fields
Galois Theory
Galois Group as a Subgroup of $S_n$
Galois Theory
Solvability of Symmetric Groups
Galois Theory
Lattice Isomorphism via Galois Correspondence
Galois Theory
Normality and Conjugate Embeddings
Galois Theory
Order of the Galois Group
Galois Theory
Galois Groups of Quartics
Galois Theory