Bourbaki

Corollary 4

Tag 06N7, permanent. Cite it as [Bourbaki, Tag 06N7].

Algebra, V, p. 68, § 10. Galois extensions

Let E be a subfield of N containing K and let A=g(E)A = g(E). For E to be Galois over K it is necessary and sufficient that A should be a normal subgroup of Γ\Gamma. When this is so, the restriction homomorphism of Γ\Gamma into Gal(E/K)\mathrm{Gal}(E/K) defines by passage to quotients a topological group isomorphism of Γ/Δ\Gamma / \Delta onto Gal(E/K)\mathrm{Gal}(E/K).

Since N is separable over K, the same is true of E (V, p. 36, Prop. 1). Therefore E is Galois over K if and only if it is quasi-Galois over K; this also means that σ(E)=E\sigma(E) = E for every K-automorphism σ\sigma of N (V, p. 52, Prop. 1 and p. 54, Prop. 3). By Cor. 3 this is equivalent to σΔσ−1=A\sigma \Delta \sigma^{-1} = A for all σ∈Γ\sigma \in \Gamma.

The restriction homomorphism φ:Gal(N/K)→Gal(E/K)\varphi : \mathrm{Gal}(N/K) \to \mathrm{Gal}(E/K) is continuous and surjective (V, p. 60, Prop. 3) and its kernel is clearly equal to A=Gal(N/E)A = \mathrm{Gal}(N/E). Since Γ\Gamma is compact, the homomorphism of Γ/Δ\Gamma / \Delta onto Gal(E/K)\mathrm{Gal}(E/K) derived from φ\varphi by passage to quotients is an isomorphism of topological groups (Gen. Top., I, p. 87, Cor. 2).

Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.

Cited by

Cites

Transcribed from the printed edition, 2003, Springer. Not translated.