Bourbaki

§ 16. Other descriptions of the brauer group

In this section, if F is an abelian group and gg an automorphism of F, then we write g⋅fg\cdot f for g(f)g(f).

1. τ\tau-Extensions of Groups

In this subsection, we fix a group G, an abelian group F written multiplicatively, and a group homomorphism τ\tau from G to the automorphism group Aut(F) of F. We denote the identity element of G by ee and the identity element of F by 1.

Recall (I, §6, No. 1, p. 65) that an extension E\mathscr{E} of G by F is a triple (Γ,ι,π)(\Gamma , \iota , \pi ), where Γ is a group, π:Γ→G\pi : \Gamma \rightarrow G is a surjective homomorphism, and ι\iota is an injective homomorphism from F to Γ such that Im(ι\iota ) == Ker(π\pi ). Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be such an extension. For every γ∈Γ\gamma \in \Gamma, the mapping φγ:F→F\varphi_{\gamma}: F\rightarrow F defined by

ι(φγ(f))=γι(f)γ−1 \iota (\varphi_{\gamma}(f)) =\gamma \iota (f)\gamma^{-1}

for f∈Ff\in F is an automorphism of F. Since F is commutative, for every f∈Ff\in F, the automorphism defined by ι(f)\iota (f) is the identity mapping on F. By passing to the quotient, we obtain a homomorphism IntE_{\mathscr{E}} from G to Aut(F) characterized by

γι(f)γ−1=ι\gamma \iota (f)\gamma^{-1}=\iota(IntE(π(γ))⋅f_{\mathscr{E}}(\pi (\gamma ))\cdot f)

for γ∈Γ\gamma \in \Gamma and f∈Ff\in F.

AτA\tau-extension of G by F is an extension E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) such that IntE_{\mathscr{E}} is equal to τ\tau, in other words, that satisfies the relation

γι(f)γ−1=ι(τ(π(γ))⋅f)(1) \gamma \iota (f)\gamma^{-1}=\iota (\tau (\pi (\gamma ))\cdot f) \tag{1}

for γ∈Γ\gamma \in \Gamma and f∈Ff\in F. If E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) and E′=(Γ′,ι′,π′)\mathscr{E}'= (\Gamma ', \iota ', \pi ') are τ\tau-extensions, then a morphism of τ\tau-extensions from E\mathscr{E} to E′\mathscr{E}' is a morphism of extensions from E\mathscr{E} to E′\mathscr{E}' (I, §6, No. 1, p. 65), that is, a group homomorphism u:Γ→Γ′u: \Gamma \rightarrow \Gamma ' such that π′∘u=π\pi '\circ u=\pi and ι′=u∘ι\iota '=u\circ \iota. Note that the τ\tau-extensions form a species of structures for which every morphism is an isomorphism (Set Theory, IV, §1, No. 5, p. 264 and I, §6, No. 1, p. 66, Proposition 1). Denote by Isoτ(E,F)_{\tau}(\mathscr{E},\mathscr{F}) the relation

“E\mathscr{E} and F\mathscr{F} are isomorphic τ\tau-extensions.”

This is an equivalence relation; the class of the extension E\mathscr{E} is the class of objects equivalent to E\mathscr{E} for Isoτ(_{\tau}(Set Theory, II, §6, No. 6, p. 122).

Lemma 1 00I9

The relation

“α\alpha is the class of a τ\tau-extension for Isoτ_{\tau}”

is collectivizing in α\alpha.

Set E0=F×GE_0= F\times G, and consider the mappings ι0:F→E0\iota_0: F\rightarrow E_0 and π0:E0→G\pi_0: E_0\rightarrow G defined by ι0(f)=(f,e)\iota_0(f) = (f, e) for f∈Ff\in F and π0(f,g)=g\pi_0(f, g) =g for (f,g)∈F×G(f, g)\in F\times G. Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension of G by F. The mapping π\pi is surjective, so has a section σ:G→Γ\sigma : G\rightarrow \Gamma such that σ(e)=e\sigma (e) =e. The mapping u:(f,g)↦ι(f)σ(g)u: (f, g)\mapsto \iota (f)\sigma (g) from F×GF\times G to Γ is bijective. We endow F×GF\times G with the group law obtained by transfer of structure. The triple (E0,ι0,π0)(E_0, \iota_0, \pi_0) is then a τ\tau-extension isomorphic to E\mathscr{E}. Lemma 1 now follows from Set Theory, II, §6, No. 6, p. 122.

We denote by Exτ(G,F)_{\tau}(G,F) the set of classes of τ\tau-extensions for the relation Isoτ_{\tau}.

Example 1 00IA

The external semidirect product (F×τG,i,p)(F\times_{\tau}G, i, p) (I, §6, No. 1, p. 66, Proposition 3) is a τ\tau-extension; we denote it by Iτ\mathscr{I}_{\tau}. We say that a τ\tau-extension is semitrivial if it is isomorphic to the τ\tau-extension Iτ\mathscr{I}_{\tau}.

Example 2 00IB

For i∈{1,2}i\in \{1,2\}, take a group GiG_i, an abelian group FiF_i, and a group homomorphism τi\tau_i from GiG_i to the automorphism group of FiF_i. We denote by τ1×τ2:G1×G2→\tau_1\times \tau_2: G_1\times G_2\rightarrow Aut(F1×F2_1\times F_2) the homomorphism defined by

(τ1×τ2)(g1,g2)⋅(f1,f2)=(τ1(g1)⋅f1,τ2(g2)⋅f2) (\tau_1\times \tau_2)(g_1, g_2)\cdot (f_1, f_2) = (\tau_1(g_1)\cdot f_1, \tau_2(g_2)\cdot f_2)

for g1∈G1,g2∈G2,f1∈F1g_1\in G_1,g_2\in G_2,f_1\in F_1, and f2∈F2f_2\in F_2. Let E1=(Γ1,ι1,π1)\mathscr{E}_1= (\Gamma_1, \iota_1, \pi_1) be a τ1\tau_1-extension of G1G_1 by F1F_1 and E2=(Γ2,ι2,π2)\mathscr{E}_2= (\Gamma_2, \iota_2, \pi_2) a τ2\tau_2-extension of G2G_2 by F2F_2. Then the triple (Γ1×Γ2,ι1×ι2,π1×π2)(\Gamma_1\times \Gamma_2, \iota_1\times \iota_2, \pi_1\times \pi_2) is a τ1×τ2\tau_1\times \tau_2-extension of G1×G2G_1\times G_2 by F1×F2F_1\times F_2; it is called the exterior product of the extensions E1\mathscr{E}_1 and E2\mathscr{E}_2 and denoted by E1×E2\mathscr{E}_1\times \mathscr{E}_2.

2. Inverse Image of a τ\tau-Extension

Let G and G′G' be groups. Let F be an abelian group, and let τ:G→\tau : G\rightarrow Aut(F) and u:G′→Gu: G'\rightarrow G be group homomorphisms. Consider the group homomorphism τ′=τ∘u:G′→\tau '=\tau \circ u: G'\rightarrow Aut(F), and denote by Γ′\Gamma ' the fiber product Γ×GG′\Gamma \times_GG' with respect to π\pi and uu (I, §4, No. 8, p. 46). It is the subgroup of Γ×G′\Gamma \times G' consisting of the pairs (γ,g′)(\gamma , g') such that π(γ)=u(g′)\pi (\gamma ) =u(g'). Let ι′:F→Γ′\iota ': F\rightarrow \Gamma ' be the group homomorphism given by ι′(α)=(ι(α),e)\iota '(\alpha ) = (\iota (\alpha ), e) for α∈F\alpha \in F, and denote the second projection by π′:Γ′→G′\pi ': \Gamma '\rightarrow G'. Then the morphism ι′\iota ' is injective, the morphism π′\pi ' is surjective because π\pi is, and the image of ι′\iota ' coincides with the kernel of π′\pi '. Moreover, for every α∈F\alpha \in F and γ′=(γ,g)∈Γ′\gamma '= (\gamma , g)\in \Gamma ', we have the relations

γ′ι′(α)γ′−1=(γ,g)(ι(α),e)(γ,g)−1=(ι(τ(π(γ)).α),e)=ι′(τ′(π′(γ′)).α) \gamma '\iota '(\alpha )\gamma^{'-1}= (\gamma , g)(\iota (\alpha ), e)(\gamma , g)^{-1}= (\iota (\tau (\pi (\gamma )).\alpha ), e) =\iota '(\tau '(\pi '(\gamma ')).\alpha )

Consequently, (Γ′,ι′,π′)(\Gamma ', \iota ', \pi ') is a τ′\tau '-extension of G′G' by F; we call it the inverse image of E\mathscr{E} by uu and denote it by u∗(E)u^*(\mathscr{E}). The first projection is a group homomorphism φ:Γ′→Γ\varphi : \Gamma '\rightarrow \Gamma that we call canonical.

Proposition 1 00IC

The following diagram commutes:

F ι′^{\iota'} // Γ′π′{\Gamma'}^{\pi'} // G′G'

(2)

φu\varphi u

F ι^{\iota} // Γ π^{\pi} // G.

Moreover, if E1′=(Γ1′,ι1′,π1′)\mathscr{E}_1'= (\Gamma '_1, \iota '_1, \pi '_1) is a τ′\tau '-extension and φ1:Γ1′→Γ\varphi_1: \Gamma '_1\rightarrow \Gamma is a group homomorphism such that the diagram

F ι1′^{\iota'_1} // Γ1′π1′{\Gamma'_1}^{\pi'_1} // G′G'

φ1u\varphi_1u

F ι^{\iota} // Γ π^{\pi} // G

commutes, then there exists a unique morphism ψ\psi of τ′\tau '-extensions from E1′\mathscr{E}_1' to u∗(E)u^*(\mathscr{E}) such that we have φ1=φ∘ψ\varphi_1=\varphi \circ \psi.

The commutativity of the first diagram follows from the definition of φ\varphi. The existence and uniqueness of ψ\psi follow from Lemma 2 below.

Lemma 2 00ID

Let F1′F'_1 be an abelian group, and let w:F1′→w: F'_1\rightarrow F and τ1:G′→\tau_1: G'\rightarrow Aut(F1′'_1) be group homomorphisms such that

w(τ1(g)(f))=τ(u(g))(w(f)) w(\tau_1(g)(f)) =\tau (u(g))(w(f))

for every f∈F1′f\in F'_1 and g∈G′g\in G'. Let E1′=(Γ1′,ι1′,π1′)\mathscr{E}_1'= (\Gamma '_1, \iota '_1, \pi '_1) be a τ1\tau_1-extension of G′G' by F1′F'_1 and φ1:Γ1′→Γ\varphi_1: \Gamma '_1\rightarrow \Gamma a group homomorphism such that the following diagram commutes:

F1′ι1′{F'_1}^{\iota'_1} // Γ1′π1′{\Gamma'_1}^{\pi'_1} // G′G'

wφ1uw\varphi_1u

F ι^{\iota} // Γ π^{\pi} // G.

Then there exists a unique group homomorphism ψ:Γ1′→Γ′\psi : \Gamma '_1\rightarrow \Gamma ' such that the diagram

F1′ι1′{F'_1}^{\iota'_1} // Γ1′π1′{\Gamma'_1}^{\pi_1'} // G′G'

wψw\psi

F ι′^{\iota'} // Γ′π′{\Gamma'}^{\pi'} // G′G'

commutes and φ1=φ∘ψ\varphi_1=\varphi \circ \psi.

If the group homomorphism ψ:Γ1′→Γ′\psi : \Gamma '_1\rightarrow \Gamma ' has the desired properties, then it satisfies the relations

ψ(γ)=(φ∘ψ(γ),π′∘ψ(γ))=(φ1(γ),π1′(γ)) \psi (\gamma ) = (\varphi \circ \psi (\gamma ), \pi '\circ \psi (\gamma )) = (\varphi_1(\gamma ), \pi '_1(\gamma ))

for every γ∈Γ1′\gamma \in \Gamma '_1. Conversely, the group homomorphism from Γ1′\Gamma '_1 to Γ×G′\Gamma \times G' defined by γ↦(φ1(γ),π1′(γ))\gamma \mapsto (\varphi_1(\gamma ), \pi_1'(\gamma )) has values in the fiber product Γ×GG′\Gamma \times_GG' because π∘φ1(γ)=u∘π1′(γ)\pi \circ \varphi_1(\gamma ) =u\circ \pi_1'(\gamma ) for every γ∈Γ1′\gamma \in \Gamma '_1.

Corollary 1 00IE

Let E1\mathscr{E}_1 and E2\mathscr{E}_2 be τ\tau-extensions of G by F, and let ψ\psi be a morphism of τ\tau-extensions from E1\mathscr{E}_1 to E2\mathscr{E}_2. Denote by φ1(\varphi_1(resp. φ2)\varphi_2) the canonical homomorphism for u∗(E1)(u^*(\mathscr{E}_1) (resp. u∗(E2))u^*(\mathscr{E}_2)). Then there exists a unique morphism of τ′\tau '-extensions from u∗(E1)u^*(\mathscr{E}_1) to u∗(E2)u^*(\mathscr{E}_2), denoted by u∗(ψ)u^*(\psi ), such that φ2∘u∗(ψ)=ψ∘φ1\varphi_2\circ u^*(\psi ) =\psi \circ \varphi_1.

It suffices to apply Proposition 1 to ψ∘φ1\psi \circ \varphi_1.

The class of the τ′\tau '-extension u∗(E)u^*(\mathscr{E}) therefore depends only on the class of E\mathscr{E}. We also denote by u∗:u^*: Exτ(G,F)→_{\tau}(G,F)\rightarrow Exτ′(G′,F)_{\tau'}(G',F) the mapping that sends the class of a τ\tau-extension E\mathscr{E} to the class of the τ′\tau '-extension u∗(E)u^*(\mathscr{E}).

Corollary 2 00IF

Let u′:G′′→G′u': G''\rightarrow G' be a group homomorphism, and let E\mathscr{E} be a τ\tau-extension of G by F. Set τ′′=τ′∘u′\tau ''=\tau '\circ u', and denote by φ(\varphi (resp. φ′,φ′′)\varphi ',\varphi '') the canonical homomorphism associated with the τ′\tau '-extension u∗(E)(u^*(\mathscr{E}) (resp. the τ′′\tau ''-extension u′∗(u∗(E))u^{'*}(u^*(\mathscr{E})), the τ′′\tau ''-extension (u∘u′)∗(E))(u\circ u')^*(\mathscr{E})). Then there exists a unique morphism ψ\psi from the τ′′\tau ''-extension u′∗(u∗(E))u^{'*}(u^*(\mathscr{E})) to the τ′′\tau ''-extension (u∘u′)∗(E)(u\circ u')^*(\mathscr{E}) such that φ′′∘ψ=φ∘φ′\varphi ''\circ \psi =\varphi \circ \varphi '.

Example 00RI

Let H be a subgroup of G and j:H→Gj: H\rightarrow G the canonical injection. Then for every τ\tau-extension E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ), the τ∘j\tau \circ j-extension j∗(E)j^*(\mathscr{E}) is isomorphic to (π−1(H),ι′,π′)(\overset{-1}{\pi}(H), \iota ', \pi '), where ι′:F→π−1(H)\iota ': F\rightarrow \overset{-1}{\pi}(H) (resp. π′:π−1(H)→H\pi ':\overset{-1}{\pi}(H)\rightarrow H) is the group homomorphism f↦ι(f)f\mapsto \iota (f) (resp. γ↦π(γ)\gamma \mapsto \pi (\gamma )). More generally, if the group homomorphism u:G′→Gu: G'\rightarrow G is injective, then the canonical homomorphism φ\varphi is injective with image π−1(u(G′))\overset{-1}{\pi}(u(G')).

3. Direct Image of a τ\tau-Extension

Let G be a group, let F and F′F' be abelian groups, let τ\tau (resp. τ′\tau ') be a group homomorphism from G to the automorphism group of F (resp. F′F'), and let v:F→F′v: F\rightarrow F' be a group homomorphism such that

v(τ(g)⋅f)=τ′(g)⋅v(f)(3) v(\tau (g)\cdot f) =\tau '(g)\cdot v(f) \tag{3}

for every g∈Gg\in G and f∈Ff\in F. Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension of G by F. Let Γ~\widetilde{\Gamma} be the external semidirect product F′×τ′∘πΓF'\times_{\tau'\circ\pi}\Gamma. We denote by ı~:F′→Γ~\widetilde{\imath}: F'\rightarrow \widetilde{\Gamma} the homomorphism f↦(f,e)f\mapsto (f, e) and by p~:Γ~→Γ\widetilde{p}:\widetilde{\Gamma}\rightarrow \Gamma the first projection. Let j:F→Γ~j: F\rightarrow \widetilde{\Gamma} be the mapping defined by f↦(v(f),ι(f)−1)f\mapsto (v(f), \iota (f)^{-1}). Since the image of ι\iota is contained in the kernel of τ′∘π\tau '\circ \pi, the mapping jj is a group homomorphism; it is injective because ι\iota is injective. We have the relations

(f′,γ)j(f)(f′,γ)−1=(f′,γ)(v(f),ι(f)−1)(f′,γ)−1 (f', \gamma )j(f)(f', \gamma )^{-1}= (f', \gamma )(v(f), \iota (f)^{-1})(f', \gamma )^{-1}
=(τ′(π(γ))⋅v(f),ι(τ(π(γ))⋅f)−1)(4) = (\tau '(\pi (\gamma ))\cdot v(f), \iota (\tau (\pi (\gamma ))\cdot f)^{-1}) \tag{4}
=j(τ(π(γ))⋅f) =j(\tau (\pi (\gamma ))\cdot f)

for f∈F,γ∈Γf\in F,\gamma \in \Gamma, and f′∈F′f'\in F'. Consequently, the image of jj is a normal subgroup of Γ~\widetilde{\Gamma}. We denote by Γ′\Gamma ' the quotient of Γ~\widetilde{\Gamma} by the image of jj. We denote by ι′\iota ' the composition of the canonical surjection from Γ~\widetilde{\Gamma} to Γ′\Gamma ' and ı~\widetilde{\imath}. The kernel of the homomorphism π∘p~\pi \circ \widetilde{p} is the product F′×ιF'\times \iota(F), which contains the image of jj. We define π′:Γ′→G\pi ': \Gamma '\rightarrow G as the group homomorphism deduced from π∘p~\pi \circ \widetilde{p} by passing to the quotient. Since ι\iota is injective, the intersection of j(F)j(F) and ı~(F′)\widetilde{\imath}(F') is reduced to the identity element of Γ~\widetilde{\Gamma}; it follows that ι′\iota ' is injective. The mapping from F′×FF'\times F to F′×ι(F)=F'\times \iota (F) = Ker(π∘p~\pi \circ \widetilde{p}) given by

(f′,f)⟼(f′v(f),ι(f)−1) (f', f)\longmapsto (f'v(f), \iota (f)^{-1})

is a group isomorphism. The image of ι′\iota ' therefore coincides with the kernel of π′\pi '. Since π\pi and p~\widetilde{p} are surjective, so is π′\pi '. This proves that E′=(Γ′,ι,π′)\mathscr{E}'= (\Gamma ', \iota , \pi ') is a τ′\tau '-extension of G by F′F'; we call it the direct image of E\mathscr{E} by vv and denote it by v∗(E)v_*(\mathscr{E}). The composition of the canonical surjection from Γ~\widetilde{\Gamma} to Γ′\Gamma ' and the group homomorphism from Γ to Γ~\widetilde{\Gamma} given by γ↦(1,γ)\gamma \mapsto (1, \gamma ) is a group homomorphism φ:Γ→Γ′\varphi : \Gamma \rightarrow \Gamma ' that we call canonical.

Proposition 2 00IG

In the notation above, the following diagram commutes:

F ι^{\iota} // Γ π^{\pi} // G

vφ(5) v\varphi \tag{5}

F′ι′{F'}^{\iota'} // Γ′π′{\Gamma'}^{\pi'} // G .

Let E1′=(Γ1′,ι1′,π1′)\mathscr{E}_1'= (\Gamma '_1, \iota '_1, \pi_1') be a τ′\tau '-extension of G by F′F', and let φ1:Γ→Γ1′\varphi_1: \Gamma \rightarrow \Gamma '_1 be a group homomorphism such that the following diagram commutes:

F ι^{\iota} // Γ π^{\pi} // G

vφ1v\varphi_1

F′ι1′{F'}^{\iota'_1} // Γ1′π1′{\Gamma'_1}^{\pi'_1} // G.

Then there exists a unique morphism ψ\psi of τ′\tau '-extensions from v∗(E)v_*(\mathscr{E}) to E1′\mathscr{E}_1' such that we have φ1=ψ∘φ\varphi_1=\psi \circ \varphi.

The commutativity of the first diagram follows by construction. The existence and uniqueness of ψ\psi follow from Lemma 3 below.

Lemma 3 00IH

Let G1′G'_1 be a group, and let w:G→G1′w: G\rightarrow G'_1 and τ1:G1′→\tau_1: G'_1\rightarrow Aut(F′') be group homomorphisms such that τ′=τ1∘w\tau '=\tau_1\circ w. Let E1′=(Γ1′,ι1′,π1′)\mathscr{E}_1'= (\Gamma '_1, \iota '_1, \pi '_1) be a τ1\tau_1-extension of G1′G'_1 by F′F', and let φ1:Γ→Γ1′\varphi_1: \Gamma \rightarrow \Gamma '_1 be a group homomorphism such that the following diagram commutes:

F ι^{\iota} // Γ π^{\pi} // G

vφ1wv\varphi_1w

F′ι1′{F'}^{\iota'_1} // Γ1′π1′{\Gamma'_1}^{\pi'_1} // G1′G'_1.

Then there exists a unique group homomorphism ψ:Γ′→Γ1′\psi : \Gamma '\rightarrow \Gamma '_1 such that the diagram

F′ι′{F'}^{\iota'} // Γ′π′{\Gamma'}^{\pi'} // G

ψw\psi w

F′ι1′{F'}^{\iota'_1} // Γ1′π1′{\Gamma'_1}^{\pi'_1} // G1′G'_1 commutes and φ1=ψ∘φ\varphi_1=\psi \circ \varphi.

For any (f,γ)∈F′×Γ(f, \gamma )\in F'\times \Gamma, denote the class of (f,γ)(f, \gamma ) in Γ′\Gamma ' by (f,γ)(f, \gamma ). If the group homomorphism ψ:Γ′→Γ1′\psi : \Gamma '\rightarrow \Gamma '_1 has the desired properties, then it satisfies the relations

ψ(f′,γ)=ψ(ι′(f′)φ(γ))=ι1′(f′)φ1(γ) \psi (f', \gamma )=\psi (\iota '(f')\varphi (\gamma )) =\iota '_1(f')\varphi_1(\gamma )

for every f′∈F′f'\in F' and γ∈Γ\gamma \in \Gamma. Conversely, the mapping ψ~\widetilde{\psi} from F′×τ′∘πΓF'\times_{\tau'\circ\pi}\Gamma to Γ1′\Gamma '_1 given by (f,γ)↦ι1′(f)φ1(γ)(f, \gamma )\mapsto \iota '_1(f)\varphi_1(\gamma ) is a group homomorphism. Indeed, we have the relations

ι1′(f)φ1(γ)ι1′(f′)φ1(γ′)=ι1′(fτ1(π1′(φ1(γ)))⋅f′)φ1(γγ′) \iota '_1(f)\varphi_1(\gamma )\iota '_1(f')\varphi_1(\gamma ') =\iota '_1(f \tau_1(\pi '_1(\varphi_1(\gamma )))\cdot f')\varphi_1(\gamma \gamma ')
=ι1′(fτ′(π(γ))⋅f′)φ1(γγ′) =\iota '_1(f \tau '(\pi (\gamma ))\cdot f')\varphi_1(\gamma \gamma ')

for f,f′∈F′f, f'\in F' and γ,γ′∈Γ\gamma , \gamma '\in \Gamma. The kernel of ψ~\widetilde{\psi} contains the image of jj because ι1′(v(f))=φ1(ι(f))\iota '_1(v(f)) =\varphi_1(\iota (f)) for f∈Ff\in F, and the morphism ψ\psi deduced from ψ~\widetilde{\psi} by passing to the quotient has the desired properties.

Remark 00II

Denote by Σ the species of structures of τ′\tau '-extensions, and define α\alpha-mappings to be mappings from Γ to a group Γ′\Gamma ' underlying a τ′\tau '-extension that are group homomorphisms and that make the following diagram commute:

F ι^{\iota} // Γ π^{\pi} // G

vφ(6) v\varphi \tag{6}

F′ι′{F'}^{\iota'} // Γ′π′{\Gamma'}^{\pi'} // G .

Proposition 2 expresses that v∗(E)v_*(\mathscr{E}) is a solution of the corresponding universal mapping problem (Set Theory, IV, §3, No. 1, p. 284).

Corollary 1 00IJ

Let E1\mathscr{E}_1 and E2\mathscr{E}_2 be τ\tau-extensions of G by F, and let ψ\psi be a morphism of τ\tau-extensions from E1\mathscr{E}_1 to E2\mathscr{E}_2. Denote by φ1(\varphi_1(resp. φ2)\varphi_2) the canonical homomorphism for v∗(E1)(v_*(\mathscr{E}_1) (resp. v∗(E2))v_*(\mathscr{E}_2)). Then there exists a unique morphism of τ′\tau '-extensions from v∗(E1)v_*(\mathscr{E}_1) to v∗(E2)v_*(\mathscr{E}_2), denoted by v∗(ψ)v_*(\psi ), such that φ2∘ψ=v∗(ψ)∘φ1\varphi_2\circ \psi =v_*(\psi )\circ \varphi_1.

It suffices to apply Proposition 2 to φ2∘ψ\varphi_2\circ \psi.

The class of the τ′\tau '-extension v∗(E)v_*(\mathscr{E}) therefore depends only on the class of E\mathscr{E}. We also denote by v∗:v_*: Exτ(G,F)→_{\tau}(G,F)\rightarrow Exτ′(G,F′)_{\tau'}(G,F') the mapping that sends the class of a τ\tau-extension E\mathscr{E} to the class of the τ′\tau '-extension v∗(E)v_*(\mathscr{E}).

Corollary 2 00RJ

We keep the notation of the proposition. Let F′′F'' be an abelian group, and let τ′′:G→\tau '': G\rightarrow Aut(F′′'') and v′:F′→F′′v': F'\rightarrow F'' be group homomorphisms such that

τ′′(g)⋅v′(f)=v′(τ′(g)⋅f) \tau ''(g)\cdot v'(f) =v'(\tau '(g)\cdot f)

for g∈Gg\in G and f∈F′f\in F'. Let E\mathscr{E} be a τ\tau-extension of G by F, and denote by φ(\varphi (resp. φ′,φ′′)\varphi ',\varphi '') the canonical homomorphism associated with v∗(E)(v_*(\mathscr{E}) (resp. v∗′(v∗(E))v'_*(v_*(\mathscr{E})), (v′∘v)∗(E))(v'\circ v)_*(\mathscr{E})). Then there exists a unique morphism ψ\psi from the τ′′\tau ''-extension v∗′(v∗(E))v'_*(v_*(\mathscr{E})) of G by F′′F'' to the τ′′\tau ''-extension (v′∘v)∗(E)(v'\circ v)_*(\mathscr{E}) such that φ′′=ψ∘φ′∘φ\varphi ''=\psi \circ \varphi '\circ \varphi.

Example 1 00IK

Let j:{1}→Fj:\{1\} \rightarrow F be the canonical injection. The semitrivial extension Iτ\mathscr{I}_{\tau} is isomorphic to j∗((G,ij_*((G, i, IdG))_G)), where i:{e}→Gi:\{e\} \rightarrow G is the canonical injection. Let c:F→Fc: F\rightarrow F be the constant homomorphism f↦1f\mapsto 1, and let E\mathscr{E} be a τ\tau-extension. The τ\tau-extension c∗(E)c_*(\mathscr{E}) is also isomorphic to Iτ\mathscr{I}_{\tau}.

Example 2 00IL

Let E be a subgroup of F stable under the action of G. Denote by F′F' the quotient of F by E, by v:F→F′v: F\rightarrow F' the canonical homomorphism, and by τ′:G→\tau ': G\rightarrow Aut(F′') the action of G on F′F' characterized by

τ′(g)⋅v(f)=v(τ(g)⋅f) \tau '(g)\cdot v(f) =v(\tau (g)\cdot f)

for g∈Gg\in G and f∈Ff\in F. Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension of G by F. Then ι(E)\iota (E) is a normal subgroup of Γ, and the τ′\tau '-extension v∗(E)v_*(\mathscr{E}) of G by F′F' is isomorphic to the extension (Γ/ι(E),ι,π)(\Gamma /\iota (E), \iota , \pi ), where ι‾\overline{\iota} and π‾\overline{\pi} are the group homomorphisms deduced from ι\iota and π\pi by passing to the quotients. By this isomorphism, the canonical homomorphism φ\varphi associated with v∗(E)v_*(\mathscr{E}) corresponds to the canonical homomorphism from Γ to Γ/ι(E)\Gamma /\iota (E).

Proposition 3 00IM

Let G and G′G' be groups, and let F and F′F' be abelian groups. Let τ:G→\tau : G\rightarrow Aut(F),τ′:G→,\tau ': G\rightarrow Aut(F′'),u:G′→G,u: G'\rightarrow G, and v:F→F′v: F\rightarrow F' be group homomorphisms such that

τ′(g)⋅v(f)=v(τ(g)⋅f) \tau '(g)\cdot v(f) =v(\tau (g)\cdot f)

for g∈Gg\in G and f∈Ff\in F. We write τ′′=τ′∘u\tau ''=\tau '\circ u. Let E\mathscr{E} be a τ\tau-extension of G by F. We denote by φu(\varphi_u(resp. φv,φu′,φv′)\varphi_v,\varphi '_u,\varphi '_v) the canonical homomorphism corresponding to the τ∘u\tau \circ u-extension u∗(E)(u^*(\mathscr{E}) (resp. to the τ′\tau '-extension v∗(E)v_*(\mathscr{E}), to the τ′′\tau ''-extensions u∗(v∗(E))u^*(v_*(\mathscr{E})) and v∗(u∗(E)))v_*(u^*(\mathscr{E}))). Then there exists a unique morphism ψ\psi of τ′′\tau ''-extensions from v∗(u∗(E))v_*(u^*(\mathscr{E})) to u∗(v∗(E))u^*(v_*(\mathscr{E})) such that φv∘φu=\varphi_v\circ \varphi_u= φu′∘ψ∘φv′\varphi '_u\circ \psi \circ \varphi '_v.

We denote the τ∘u\tau \circ u-extension u∗(E)u^*(\mathscr{E}) (resp. the τ′′\tau ''-extension u∗(v∗(E))u^*(v_*(\mathscr{E}))) by (Γu,ιu,πu)(\Gamma_u, \iota_u, \pi_u) (resp. (Γu′,ιu′,πu′)(\Gamma '_u, \iota '_u, \pi '_u)). Applying Lemma 2 of VIII, p. 287 to φv∘φu\varphi_v\circ \varphi_u, we find that there exists a group homomorphism ψ1:Γu→Γu′\psi_1: \Gamma_u\rightarrow \Gamma '_u such that the diagram

F ιu^{\iota_u} // Γuπu\Gamma_u^{\pi_u} // G′G'

vψ1v\psi_1

F′ιu′{F'}^{\iota'_u} // Γu′πu′{\Gamma'_u}^{\pi'_u} // G′G'

commutes and φu′∘ψ1=φv∘φu\varphi '_u\circ \psi_1=\varphi_v\circ \varphi_u. The existence of ψ\psi follows from Proposition 2 applied to ψ1\psi_1. Conversely, if ψ′\psi ' also has the desired properties, then we have ψ′∘φv′=ψ1\psi '\circ \varphi '_v=\psi_1 by Lemma 2, so ψ′=ψ\psi '=\psi (Proposition 2).

4. Group Law on the Classes of τ\tau-Extensions

Let G be a group, F be an abelian group, and τ:G→\tau : G\rightarrow Aut(F) be a group homomorphism. We denote by δ:G→G×G\delta : G\rightarrow G\times G the diagonal mapping g↦g\mapsto (g,g)(g, g) and by m:F×F→Fm: F\times F\rightarrow F the multiplication homomorphism (f1,f2)↦(f_1, f_2)\mapsto f1f2f_1f_2. We denote by s:F→Fs: F\rightarrow F the group homomorphism given by f↦f−1f\mapsto f^{-1}. Let E1=(Γ1,ι1,π1)\mathscr{E}_1= (\Gamma_1, \iota_1, \pi_1) and E2=(Γ2,ι2,π2)\mathscr{E}_2= (\Gamma_2, \iota_2, \pi_2) be τ\tau-extensions of G by F. Since we have the relation

m(((τ×τ)∘δ)(g)⋅(f1,f2))=τ(g)⋅m(f1,f2) m(((\tau \times \tau )\circ \delta )(g)\cdot (f_1, f_2)) =\tau (g)\cdot m(f_1, f_2)

for every g∈Gg\in G and all f1,f2∈Ff_1, f_2\in F, the extension m∗(δ∗(E1×E2))m_*(\delta^*(\mathscr{E}_1\times \mathscr{E}_2)) is a τ\tau-extension; we call it the product of the τ\tau-extensions E1\mathscr{E}_1 and E2\mathscr{E}_2 and denote it by E1E2\mathscr{E}_1\mathscr{E}_2. The class of this extension depends only on the classes of the extensions E1\mathscr{E}_1 and E2\mathscr{E}_2 (VIII, p. 288, Corollary 1 and VIII, p. 291, Corollary 1). So this gives a law of composition on Exτ(G,F)_{\tau}(G,F).

Remark 00IN

Let E1=(Γ1,ι1,π1)\mathscr{E}_1= (\Gamma_1, \iota_1, \pi_1) and E2=(Γ2,ι2,π2)\mathscr{E}_2= (\Gamma_2, \iota_2, \pi_2) be τ\tau-extensions of G by F. Let E1E2=(Γ,ι,π)\mathscr{E}_1\mathscr{E}_2= (\Gamma , \iota , \pi ) be the product of these extensions. Following the example of VIII, p. 289, the construction gives a surjective group homomorphism from the fiber product Γ1×GΓ2\Gamma_1\times_G\Gamma_2 to Γ whose kernel is the image of the group homomorphism f↦(ι1(f),ι2(f)−1)f\mapsto (\iota_1(f), \iota_2(f)^{-1}) from F to Γ1×GΓ2\Gamma_1\times_G\Gamma_2.

Proposition 4 00IO

The product of τ\tau-extensions endows the set Exτ(G,F)_{\tau}(G,F) with the structure of an abelian group. Its identity element is the class of the semitrivial extension Iτ\mathscr{I}_{\tau}. The inverse of the class of a τ\tau-extension E\mathscr{E} is the class of s∗(E)s_*(\mathscr{E}).

The associativity of the law follows from the commutativity of the diagrams

G→δG×G↓δ↓Id⁡×δG×G→δ×Id⁡G×G×G \begin{array}{ccc} G & \xrightarrow{\delta} & G \times G \\ \downarrow \delta & & \downarrow \operatorname{Id} \times \delta \\ G \times G & \xrightarrow{\delta \times \operatorname{Id}} & G \times G \times G \end{array}
and
F×F×F→m×Id⁡F×F↓Id⁡×m↓mF×F→mF \begin{array}{ccc} F \times F \times F & \xrightarrow{m \times \operatorname{Id}} & F \times F \\ \downarrow \operatorname{Id} \times m & & \downarrow m \\ F \times F & \xrightarrow{m} & F \end{array}

and from Corollary 2 of VIII, p. 288; Corollary 2 of VIII, p. 291; and Proposition 3 of VIII, p. 292.

Let Δ:F→F×F\Delta : F \to F \times F be the diagonal mapping f↦(f,f)f \mapsto (f, f). Let E=(Γ,ι,π)\mathscr{E} = (\Gamma, \iota, \pi) be a τ\tau-extension. Let Δ~:Γ→Γ×GΓ\tilde{\Delta} : \Gamma \to \Gamma \times_G \Gamma be the group homomorphism given by γ↦(γ,γ)\gamma \mapsto (\gamma, \gamma). The following diagram commutes:

F→ιΓ→πG↓Δ↓Δ~∥F×F⟶Γ×GΓ⟶G. \begin{array}{ccccc} F & \xrightarrow{\iota} & \Gamma & \xrightarrow{\pi} & G \\ \downarrow \Delta & & \downarrow \tilde{\Delta} & & \parallel \\ F \times F & \longrightarrow & \Gamma \times_G \Gamma & \longrightarrow & G. \end{array}

By Proposition 2 of VIII, p. 290, it follows that the (τ×τ)∘δ(\tau \times \tau) \circ \delta-extension δ∗(E×E)\delta^*(\mathscr{E} \times \mathscr{E}) is isomorphic to Δ∗(E)\Delta_*(\mathscr{E}).

Denote by c:F→Fc : F \to F the constant homomorphism f↦1f \mapsto 1. by Example 1 of VIII, p. 292, the fact that Iτ\mathscr{I}_\tau is an identity element for this law of composition follows from the isomorphism from δ∗(E×E)\delta^*(\mathscr{E} \times \mathscr{E}) to Δ∗(E)\Delta_*(\mathscr{E}) and the commutative diagram

F→(Id⁡F×c)∘ΔF×F(c×Id⁡F)∘Δ↓Id⁡F↓mF×F→mF. \begin{array}{ccc} F & \xrightarrow{(\operatorname{Id}_F \times c) \circ \Delta} & F \times F \\ (c \times \operatorname{Id}_F) \circ \Delta & & \downarrow \operatorname{Id}_F \\ & & \downarrow m \\ F \times F & \xrightarrow{m} & F. \end{array}

The last assertion follows from the commutative diagram

F→(Id⁡F×s)∘ΔF×F(s×Id⁡F)∘Δ↓c↓mF×F→mF. \begin{array}{ccc} F & \xrightarrow{(\operatorname{Id}_F \times s) \circ \Delta} & F \times F \\ (s \times \operatorname{Id}_F) \circ \Delta & & \downarrow c \\ & & \downarrow m \\ F \times F & \xrightarrow{m} & F. \end{array}

Let E1=(Γ1,ι1,π1)\mathscr{E}_1 = (\Gamma_1, \iota_1, \pi_1) and E2=(Γ2,ι2,π2)\mathscr{E}_2 = (\Gamma_2, \iota_2, \pi_2) be τ\tau-extensions. The group isomorphism Γ1×Γ2→Γ2×Γ1\Gamma_1 \times \Gamma_2 \to \Gamma_2 \times \Gamma_1 given by (γ1,γ2)↦(γ2,γ1)(\gamma_1, \gamma_2) \mapsto (\gamma_2, \gamma_1) restricts to a group isomorphism σ:Γ1×GΓ2→Γ2×GΓ1\sigma : \Gamma_1 \times_G \Gamma_2 \to \Gamma_2 \times_G \Gamma_1. Because of the relations

σ(ι1(f),ι2(f)−1)=(ι2(f−1),ι1(f−1)−1) \sigma(\iota_1(f), \iota_2(f)^{-1}) = (\iota_2(f^{-1}), \iota_1(f^{-1})^{-1})

for f∈Ff\in F, the group homomorphism σ\sigma induces, by passing to the quotients, a morphism of τ\tau-extensions from E1E2\mathscr{E}_1\mathscr{E}_2 to E2E1\mathscr{E}_2\mathscr{E}_1. So the law of composition is commutative.

Proposition 5 00IP

a) Let G and G′G' be groups, and let F be an abelian group. Let τ:G→\tau : G\rightarrow Aut(F) and u:G′→Gu: G'\rightarrow G be group homomorphisms. The mapping u∗:u^*: Exτ(G,F)→_{\tau}(G,F)\rightarrow Exτ∘u(G′,F)_{\tau\circ u}(G',F) is a group homomorphism.

b) Let G be a group, and let F and F′F' be abelian groups. Let τ:G→\tau : G\rightarrow Aut(F),τ′:G→,\tau ': G\rightarrow Aut(F′'), and v:F→F′v: F\rightarrow F' be group homomorphisms such that

τ′(g)⋅v(f)=v(τ(g)⋅f) \tau '(g)\cdot v(f) =v(\tau (g)\cdot f)

for g∈Gg\in G and f∈Ff\in F. The mapping v∗:v_*: Exτ(G,F)→_{\tau}(G,F)\rightarrow Exτ′(G,F′)_{\tau'}(G,F') is a group homomorphism.

This follows from the commutativity of the diagrams

G′δ{G'}^{\delta} // G′×G′F×FmG'\times G'F\times F^m // F

uu×u_uu_{\times u} and v×vvv_{\times v}v

G δ^{\delta} // G×GF′×F′mG\times GF'\times {F'}^m // F′F'.

5. Cohomological Description

Let G be a group, let F be an abelian group, and let τ:G→\tau : G\rightarrow Aut(F) be a group homomorphism. For any g∈Gg\in G and f∈Ff\in F, we also write gf^gf for τ(g)⋅f\tau (g)\cdot f. A 2-cocycle of G with values in F is a mapping c:G×G→Fc: G\times G\rightarrow F such that for every (g1,g2,g3)∈G×G×G(g_1, g_2, g_3)\in G\times G\times G, we have

g1c(g2,g3)c(g1,g2g3)=c(g1,g2)c(g1g2,g3)(7) g_{_1}c(g_2, g_3)c(g_1, g_2g_3) =c(g_1, g_2)c(g_1g_2, g_3) \tag{7}

Since F is commutative, the set of 2-cocycles is a subgroup of the group of mappings from G×GG\times G to F; we denote it by Z2(G,F)Z^2(G,F). We denote the group of mappings from G to F by C1(G,F)C^1(G,F). For every h∈C1(G,F)h\in C^1(G,F) and (g1,g2)∈G×G(g_1, g_2)\in G\times G, set

(∂h)(g1,g2)=g1h(g2)h(g1g2)−1h(g1)(8) (\partial h)(g_1, g_2) =^{g_1}h(g_2)h(g_1g_2)^{-1}h(g_1) \tag{8}

A simple calculation shows that the mapping ∂h:G×G→F\partial h: G\times G\rightarrow F is a 2-cocycle. The mapping ∂:C1(G,F)→Z2(G,F)\partial : C^1(G,F)\rightarrow Z^2(G,F) is a group homomorphism. For any h∈C1(G,F)h\in C^1(G,F), the 2-cocycle ∂h\partial h is called a 2-coboundary. The group Z2(G,F)/∂(C1(G,F))Z^2(G,F)/\partial (C^1(G,F)) is denoted by H2(G,F)H^2(G,F) and is called the second cohomology group of G with coefficients in F.

∗*The notation above agrees with that of X, §6, no8^o8, p. 112 concerning group cohomology.∗*

Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension. Let σ\sigma be a section of the surjective mapping π(\pi (Set Theory, II, §3, No. 8, p. 86), that is, a mapping from G to Γ such that π(σ(g))=g\pi (\sigma (g)) =g for every gg in G. For every g1,g2∈Gg_1, g_2\in G, σ(g1)σ(g2)σ(g1g2)−1\sigma (g_1)\sigma (g_2)\sigma (g_1g_2)^{-1} belongs to Ker(π\pi ), so that there exists a unique mapping cσ:G×G→Fc_{\sigma}: G\times G\rightarrow F such that

ι(cσ(g1,g2))=σ(g1)σ(g2)σ(g1g2)−1(9) \iota (c_{\sigma}(g_1, g_2)) =\sigma (g_1)\sigma (g_2)\sigma (g_1g_2)^{-1} \tag{9}

for all g1,g2∈Gg_1, g_2\in G. The mapping cσc_{\sigma} is constant with value 1 if and only if σ\sigma is a group homomorphism.

Proposition 6 00IQ

The mapping cσc_{\sigma} is an element of the group Z2(G,F)Z^2(G,F), and its class in the cohomology group H2(G,F)H^2(G,F) depends only on the class of the τ\tau-extension E\mathscr{E}.

We say that the mapping cσc_{\sigma} is the 2-cocycle associated with σ\sigma and that its cohomology class in H2(G,F)H^2(G,F) is the cohomology class of the τ\tau-extension E\mathscr{E}.

Let us first verify that cσc_{\sigma} satisfies the cocycle condition. Let g1,g2g_1,g_2, and g3g_3 be elements of G. Using formula (1) of VIII, p. 285 and (9), we obtain the relations

ι(g1cσ(g2,g3)cσ(g1,g2g3)) \iota (^{g_1}c_{\sigma}(g_2, g_3)c_{\sigma}(g_1, g_2g_3))
=σ(g1)σ(g2)σ(g3)σ(g2g3)−1σ(g1)−1σ(g1)σ(g2g3)σ(g1g2g3)−1 =\sigma (g_1)\sigma (g_2)\sigma (g_3)\sigma (g_2g_3)^{-1}\sigma (g_1)^{-1}\sigma (g_1)\sigma (g_2g_3)\sigma (g_1g_2g_3)^{-1}
=σ(g1)σ(g2)σ(g3)σ(g1g2g3)−1 =\sigma (g_1)\sigma (g_2)\sigma (g_3)\sigma (g_1g_2g_3)^{-1}

and

ι(cσ(g1,g2)cσ(g1g2,g3)) \iota (c_{\sigma}(g_1, g_2)c_{\sigma}(g_1g_2, g_3))
=σ(g1)σ(g2)σ(g1g2)−1σ(g1g2)σ(g3)σ(g1g2g3)−1 =\sigma (g_1)\sigma (g_2)\sigma (g_1g_2)^{-1}\sigma (g_1g_2)\sigma (g_3)\sigma (g_1g_2g_3)^{-1}
=σ(g1)σ(g2)σ(g3)σ(g1g2g3)−1 =\sigma (g_1)\sigma (g_2)\sigma (g_3)\sigma (g_1g_2g_3)^{-1}

The mapping cσc_{\sigma} is therefore an element of Z2(G,F)Z^2(G,F).

Let us now prove that the image of cσc_{\sigma} in H2(G,F)H^2(G,F) is independent of the chosen section σ\sigma. Let σ\sigma and σ′\sigma ' be such sections. For every element gg of G, there exists a unique element aga_g of F such that σ′(g)=ι(ag)σ(g)\sigma '(g) =\iota (a_g)\sigma (g). Let g1g_1 and g2g_2 be elements of G. Using the definition (9), we obtain the equalities

ι(cσ′(g1,g2))=σ′(g1)σ′(g2)σ′(g1g2)−1 \iota (c_{\sigma'}(g_1, g_2)) =\sigma '(g_1)\sigma '(g_2)\sigma '(g_1g_2)^{-1}
=ι(ag1)σ(g1)ι(ag2)σ(g2)σ(g1g2)−1ι(ag1g2)−1(10) =\iota (a_{g_1})\sigma (g_1)\iota (a_{g_2})\sigma (g_2)\sigma (g_1g_2)^{-1}\iota (a_{g_1g_2})^{-1} \tag{10}
=ι(ag1)ι(g1ag2)ι(cσ(g1,g2))ι(ag1g2)−1 =\iota (a_{g_1})\iota (^{g_1}a_{g_2})\iota (c_{\sigma}(g_1, g_2))\iota (a_{g_1g_2})^{-1}

Since the group F is commutative, we have the relations

cσ′(g1,g2)cσ(g1,g2)−1=(g1ag2)ag1g2−1ag1=(∂a)(g1,g2)(11) c_{\sigma'}(g_1, g_2)c_{\sigma}(g_1, g_2)^{-1}= (^{g_1}a_{g_2})a^{-1}_{g_1g_2}a_{g_1}= (\partial a)(g_1, g_2) \tag{11}

as follows from (8). This proves that the classes of cσ′c_{\sigma'} and cσc_{\sigma} in H2(G,F)H^2(G,F) coincide.

Finally, let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) and E′=(Γ′,ι′,π′)\mathscr{E}'= (\Gamma ', \iota ', \pi ') be isomorphic τ\tau-extensions, let u:E→E′u:\mathscr{E}\rightarrow \mathscr{E}' be a morphism of τ\tau-extensions, and let σ\sigma be a section of the mapping π\pi. The mapping u∘σu\circ \sigma is a section of the mapping π′\pi ', and we have cσ=cu∘σc_{\sigma}=c_{u\circ\sigma}. The class of cσc_{\sigma} in H2(G,F)H^2(G,F) therefore depends only on the class of E\mathscr{E} in Exτ(G,F)_{\tau}(G,F).

We denote by Θτ\Theta_{\tau} : Exτ(G,F)→H2(G,F)_{\tau}(G,F)\rightarrow H^2(G,F), or more simply Θ, the mapping that sends the class of a τ\tau-extension (Γ,ι,π)(\Gamma , \iota , \pi ) to the class of the 2-cocycle cσc_{\sigma}, where σ\sigma is a section of the surjection π\pi.

Theorem 1 00IR

The mapping Θ is a group isomorphism from Exτ(G,F)_{\tau}(G,F) to H2(G,F)H^2(G,F).

Let us first prove that Θ is a group homomorphism. Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) and E′=(Γ′,ι′,π′)\mathscr{E}'= (\Gamma ', \iota ', \pi ') be τ\tau-extensions, and let σ\sigma (resp. σ′\sigma ') be a section of the mapping π\pi (resp. π′\pi '). We denote by EE′=(Γ′′,ι′′,π′′)\mathscr{E} \mathscr{E}'= (\Gamma '', \iota '', \pi '') the product of the τ\tau-extensions E\mathscr{E} and E′\mathscr{E}'. We denote by [γ,γ′][\gamma , \gamma '] the image in Γ′′\Gamma '' of an element (γ,γ′)(\gamma , \gamma ') of Γ×GΓ′\Gamma \times_G\Gamma ' by the surjective homomorphism from the remark of VIII, p. 293. The mapping from G to Γ′′\Gamma '' that sends an element gg to [σ(g),σ′(g)][\sigma (g), \sigma '(g)] is a section σ′′\sigma '' of the mapping π′′\pi ''. Let g1g_1 and g2g_2 be elements of G. We have the relations

ι′′(cσ′′(g1,g2))=σ′′(g1)σ′′(g2)σ′′(g1g2)−1 \iota ''(c_{\sigma''}(g_1, g_2)) =\sigma ''(g_1)\sigma ''(g_2)\sigma ''(g_1g_2)^{-1}
=[ι(cσ(g1,g2)),ι′(cσ′(g1,g2)] = [\iota (c_{\sigma}(g_1, g_2)), \iota '(c_{\sigma'}(g_1, g_2)]
=ι′′(cσ(g1,g2)cσ′(g1,g2)) =\iota ''(c_{\sigma}(g_1, g_2)c_{\sigma'}(g_1, g_2))

We therefore have cσ′′(g1,g2)=cσ(g1,g2)cσ′(g1,g2)c_{\sigma''}(g_1, g_2) =c_{\sigma}(g_1, g_2)c_{\sigma'}(g_1, g_2).

Let us prove that the mapping Θ is injective. Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension, and let σ\sigma be a section of the mapping π\pi such that the image of cσc_{\sigma} in H2(G,F)H^2(G,F) is trivial. There exists a mapping a:G→Fa: G\rightarrow F such that

cσ(g1,g2)=(g1a(g2))a(g1g2)−1a(g1) c_{\sigma}(g_1, g_2) = (^{g_1}a(g_2))a(g_1g_2)^{-1}a(g_1)

for all g1,g2∈Gg_1, g_2\in G. We then define σ′:G→Γ\sigma ': G\rightarrow \Gamma by

σ′(g)=ι(a(g)−1)σ(g) \sigma '(g) =\iota (a(g)^{-1})\sigma (g)

for g∈Gg\in G. Using (11), we obtain that cσ′c_{\sigma'} is constant with value 1; consequently, σ′\sigma ' is a group homomorphism, which proves that the τ\tau-extension E\mathscr{E} is semitrivial (I, §6, No. 1, p. 67, Proposition 4).

Let us prove that the mapping Θ is surjective. Let cc be an element of Z2(G,F)Z^2(G,F). We endow the set Γ=F×G\Gamma = F\times G with the following law of composition: (12) (f1,g1)(f2,g2)=(f1(g1f2)c(g1,g2),g1g2)(f_1, g_1)(f_2, g_2) =(f_1(^{g_1}f_2)c(g_1, g_2), g_1g_2)

for all f1,f2∈Ff_1, f_2\in F and g1,g2∈Gg_1, g_2\in G. We have the relations

(f1,g1)((f2,g2)(f3,g3))=(f1,g1)(f2(g2f3)c(g2,g3),g2g3) (f_1, g_1) ((f_2, g_2)(f_3, g_3)) = (f_1, g_1)(f_2(^{g_2}f_3)c(g_2, g_3), g_2g_3)
=(f1(g1f2)(g1g2f3)(g1c(g2,g3))c(g1,g2g3),g1g2g3) =(f_1(^{g_1}f_2)(^{g_1g_2}f_3)(^{g_1}c(g_2, g_3))c(g_1, g_2g_3), g_1g_2g_3)

and

((f1,g1)(f2,g2))(f3,g3)=(f1(g1f2)c(g1,g2),g1g2)(f3,g3) ((f_1, g_1)(f_2, g_2)) (f_3, g_3) =(f_1(^{g_1}f_2)c(g_1, g_2), g_1g_2)(f_3, g_3)
=(f1(g1f2)(g1g2f3)c(g1,g2)c(g1g2,g3),g1g2g3) =(f_1(^{g_1}f_2)(^{g_1g_2}f_3)c(g_1, g_2)c(g_1g_2, g_3), g_1g_2g_3)

for all f1,f2,f3∈Ff_1, f_2, f_3\in F and g1,g2,g3∈Gg_1,g_2, g_3\in G. It follows that this law is associative because cc is a 2-cocycle. For every (f,g)∈Γ(f, g)\in \Gamma, we have

(f,g)(c(e,e)−1,e)=(f(gc(e,e)−1)c(g,e),g) (f, g)(c(e, e)^{-1}, e) = (f(^gc(e, e)^{-1})c(g, e), g)

Now, it follows from the definition of a 2-cocycle that c(g,e)=gc(e,e)c(g, e) =^gc(e, e), from which we deduce that (f,g)(c(e,e)−1,e)=(f,g)(f, g)(c(e, e)^{-1}, e) = (f, g). We prove likewise that

(c(e,e)−1,e)(f,g)=(fc(e,e)−1c(e,g),g)=(f,g) (c(e, e)^{-1}, e)(f, g) = (f c(e, e)^{-1}c(e, g), g) = (f, g)

The law of composition defined by (12) therefore has identity element (c(e,e)−1,e)(c(e, e)^{-1}, e), and every element (f,g)(f, g) of Γ is invertible, with inverse

(g−1(f−1c(e,e)−1c(g,g−1)−1),g−1) (^{g^{-1}}(f^{-1}c(e, e)^{-1}c(g, g^{-1})^{-1}), g^{-1})

So we have endowed Γ with a group structure. Let us denote by ι:F→G\iota : F\rightarrow G the mapping that sends ff to (c(e,e)−1f,e)(c(e, e)^{-1}f, e), by π:Γ→G\pi : \Gamma \rightarrow G the second projection, and by σ\sigma the mapping g↦(1,g)g\mapsto (1, g). The mappings ι\iota and π\pi are group homomorphisms, so the triple E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) is a τ\tau-extension, σ\sigma is a section of the mapping π\pi, and the associated cocycle cσc_{\sigma} is equal to cc because

σ(g1)σ(g2)σ(g1g2)−1=(1,g1)(1,g2)(1,g1g2)−1=(c(g1,g2),g1g2)(1,g1g2)−1 \sigma (g_1)\sigma (g_2)\sigma (g_1g_2)^{-1}= (1, g_1)(1, g_2)(1, g_1g_2)^{-1}= (c(g_1, g_2), g_1g_2)(1, g_1g_2)^{-1}
=(c(g1,g2)c(e,g1g2)−1,e)=(c(e,e)−1c(g1,g2),e) = (c(g_1, g_2)c(e, g_1g_2)^{-1}, e) = (c(e, e)^{-1}c(g_1, g_2), e)

for g1,g2∈Gg_1, g_2\in G.

Remark 00RK

Let G be a group, let F and F′F' be abelian groups, let τ\tau (resp. τ′\tau ') be a group homomorphism from G to the automorphism group of F (resp. F′F'), and let v:F→F′v: F\rightarrow F' be a group morphism such that

v(τ(g)⋅f)=τ′(g)⋅v(f)(13) v(\tau (g)\cdot f) =\tau '(g)\cdot v(f) \tag{13}

for g∈Gg\in G and f∈Ff\in F. Let α\alpha be an element of Exτ(G,F)_{\tau}(G,F). If the cocycle cc represents Θ(α)\Theta (\alpha ), then v∘cv\circ c represents Θ(v∗(α))∈H2(G,F′)\Theta (v_*(\alpha ))\in H^2(G,F').

6. Restriction and Corestriction

Let G and G′G' be groups, F be an abelian group, and τ\tau be a homomorphism from G to the automorphism group of F. Let u:G′→Gu: G'\rightarrow G be a group homomorphism. We write τ′=τ∘u\tau '=\tau \circ u.

If ψ:G×G→F\psi : G\times G\rightarrow F is a 2-cocycle of G with values in F, then the mapping ψ∘(u×u)\psi \circ (u\times u) from G′×G′G'\times G' to F given by (g1,g2)↦ψ(u(g1),u(g2))(g_1, g_2)\mapsto \psi (u(g_1), u(g_2)) is a 2-cocycle of G′G' with values in F, and the mapping ψ↦ψ∘(u×u)\psi \mapsto \psi \circ (u\times u) from Z2(G,F)Z^2(G,F) to Z2(G′,F)Z^2(G',F) induces a group homomorphism u∗:H2(G,F)→H2(G′,F)u^*: H^2(G,F)\rightarrow H^2(G',F). For any λ∈H2(G,F)\lambda \in H^2(G,F), the element u∗(λ)u^*(\lambda ) is called the inverse image of λ\lambda by uu. When G′G' is a subgroup of G and u:G′→Gu: G'\rightarrow G is the canonical injection, the homomorphism u∗u^* is called the restriction homomorphism from G to G′G'; we denote it by ResG′G^G_{G'}. When G is a quotient of G′G' and u:G′→Gu: G'\rightarrow G is the canonical surjection, the homomorphism u∗u^* is called the inflation homomorphism from G to G′G'.

Proposition 7 00IS

In the notation from above, the following diagram commutes:

Exτ(G,F)u∗_{\tau}(G,F)^{u^*} // Exτ(G′,F)_{\tau}(G',F)

ΘτΘτ′\Theta_{\tau}\Theta_{\tau'}

H2(G,F)u∗H_2(G,F)^{u^*} // H2(G′,F)H_2(G',F).

Let H be a subgroup of G of finite index. Let ss be a section of the canonical surjection from G to H\GH\backslash G. We denote by (g,x)↦x⋅g(g, x)\mapsto x\cdot g the right action of G on H\GH\backslash G induced by the right action of G on itself by multiplication. Note that for every x∈H\Gx\in H\backslash G and g∈Gg\in G, the element s(x)gs(x⋅g)−1s(x)gs(x\cdot g)^{-1} belongs to H. For any mapping c:H×H→Fc: H\times H\rightarrow F, we can therefore define a mapping c~s:G×G→F\widetilde{c}_s: G\times G\rightarrow F by the relation

c~s(g1,g2)=∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1) \widetilde{c}_s(g_1, g_2) =\prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1})

for g1,g2∈Gg_1,g_2\in G. The mapping c↦c~sc\mapsto \widetilde{c}_s is a group homomorphism from FH×HF^{H\times H} to FG×GF^{G\times G}.

Lemma 4 00IT

If c:H×H→Fc: H\times H\rightarrow F is a 2-cocycle of H with values in F, then c~s\widetilde{c}_s is a 2-cocycle of G with values in F.

Let g1,g2g_1,g_2, and g3g_3 be elements of G. For every i∈{1,2,3}i\in \{1,2,3\}, we define a mapping hi:H\G→Hh_i: H\backslash G\rightarrow H by the relation

hi(x)=s(x⋅g1⋯gi−1)gis(x⋅g1⋯gi)−1 h_i(x) =s(x\cdot g_1\cdots g_{i-1})g_is(x\cdot g_1\cdots g_i)^{-1}

for x∈H\Gx\in H\backslash G. Note that

h1(x)h2(x)=s(x)g1g2s(x⋅g1g2)−1 h_1(x)h_2(x) =s(x)g_1g_2s(x\cdot g_1g_2)^{-1}
h2(x)h3(x)=s(x⋅g1)g2g3s(x⋅g1g2g3)−1 h_2(x)h_3(x) =s(x\cdot g_1)g_2g_3s(x\cdot g_1g_2g_3)^{-1}

for x∈H\Gx\in H\backslash G. We then have the relations

g1c~s(g2,g3)c~s(g1g2,g3)−1c~s(g1,g2g3)c~s(g1,g2)−1 ^{g_1}\widetilde{c}_s(g_2, g_3)\widetilde{c}_s(g_1g_2, g_3)^{-1}\widetilde{c}_s(g_1, g_2g_3)\widetilde{c}_s(g_1, g_2)^{-1}
=∏x∈H\Gg1s(x)−1c(s(x)g2s(x⋅g2)−1,s(x⋅g2)g3s(x⋅g2g3)−1) =\prod_{x\in H\backslash G}^{g_1s(x)^{-1}}c(s(x)g_2s(x\cdot g_2)^{-1}, s(x\cdot g_2)g_3s(x\cdot g_2g_3)^{-1})
×∏x∈H\Gs(x)−1(c(h1(x)h2(x),h3(x))−1 \times \prod_{x\in H\backslash G}^{s(x)^{-1}}(c(h_1(x)h_2(x), h_3(x))^{-1}
×c(h1(x),h2(x)h3(x))c(h1(x),h2(x))−1) \times c(h_1(x), h_2(x)h_3(x))c(h_1(x), h_2(x))^{-1})
=∏x∈H\Gs(x⋅g1−1)−1h1(x⋅g1−1)c(h2(x⋅g1−1),h3(x⋅g1−1)) =\prod_{x\in H\backslash G}^{s(x\cdot g_1^{-1})^{-1}h_1(x\cdot g^{-1}_1)}c(h_2(x\cdot g_1^{-1}), h_3(x\cdot g_1^{-1}))
×∏x∈H\Gs(x)−1h1(x)c(h2(x),h3(x))−1 \times \prod_{x\in H\backslash G}^{s(x)^{-1}h_1(x)}c(h_2(x), h_3(x))^{-1}
=1 = 1

where the second equality follows from the fact that cc is a 2-cocycle.

Lemma 5 00IU

If cc is a 2-coboundary, then so is c~s\widetilde{c}_s.

Let t:H→Ft: H\rightarrow F be a mapping such that c=∂tc=\partial t. Let t~s:G→F\widetilde{t}_s: G\rightarrow F be the mapping defined by

t~s(g)=∏x∈H\Gs(x)−1t(s(x)gs(x⋅g)−1) \widetilde{t}_s(g) =\prod_{x\in H\backslash G}^{s(x)^{-1}}t(s(x)gs(x\cdot g)^{-1})

for g∈Gg\in G. It suffices to prove that c~s=∂t~s\widetilde{c}_s=\partial \widetilde{t}_s, which follows from the relations c~s(g1,g2)=∏x∈H\Gs(x)−1s(x)g1s(x⋅g1)−1t(s(x⋅g1)g2s(x⋅g1g2)−1)\widetilde{c}_s(g_1, g_2) =\prod_{x\in H\backslash G}^{s(x)^{-1}s(x)g_1s(x\cdot g_1)^{-1}}t(s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1})

×∏x∈H\Gs(x)−1(t(s(x)g1g2s(x⋅g1g2)−1)−1t(s(x)g1s(x⋅g1)−1)) \times \prod_{x\in H\backslash G}^{s(x)^{-1}}(t(s(x)g_1g_2s(x\cdot g_1g_2)^{-1})^{-1}t(s(x)g_1s(x\cdot g_1)^{-1}))
=∂t~s(g1,g2) =\partial \widetilde{t}_s(g_1, g_2)

for g1,g2∈Gg_1, g_2\in G.

Lemma 6 00IV

Let cc be a 2-cocycle of H with values in F. The image of c~s\widetilde{c}_s in the group H2(G,F)H^2(G,F) does not depend on the choice of the section ss.

Let s′s' be a section of the canonical surjection from G to H\GH\backslash G. Let h:H\G→Hh: H\backslash G\rightarrow H be the mapping characterized by the relation

s′(x)=h(x)s(x) s'(x) =h(x)s(x)

for x∈H\Gx\in H\backslash G. The 2-cocycle c~s′\widetilde{c}_{s'} is then given by the relations

c~s′(g1,g2) \widetilde{c}_{s'}(g_1, g_2)
=∏x∈H\Gs(x)−1h(x)−1c(h(x)s(x)g1s′(x⋅g1)−1,s′(x⋅g1)g2s′(x⋅g1g2)−1) =\prod_{x\in H\backslash G}^{s(x)^{-1}h(x)^{-1}}c(h(x)s(x)g_1s'(x\cdot g_1)^{-1}, s'(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1})
=∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1h(x⋅g1)−1,h(x⋅g1)s(x⋅g1)g2s′(x⋅g1g2)−1) =\prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}h(x\cdot g_1)^{-1}, h(x\cdot g_1)s(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1})
s(x)−1(−1′′−1−1s(x)^{-1}(-1''-1-1
×∏x∈H\Gc(h(x),s(x)g1g2s(x⋅g1g2)) \times \prod_{x\in H\backslash G}c(h(x), s(x)g_1g_2s(x\cdot g_1g_2))
×c(h(x)−1,s′(x)g1s′(x⋅g1)−1)) \times c(h(x)^{-1}, s'(x)g_1s'(x\cdot g_1)^{-1}))
=∏x∈H\Gg1s(x⋅g1)−1c(h(x⋅g1)−1,h(x⋅g1)s(x⋅g1)g2s′(x⋅g1g2)−1)=\prod_{x\in H\backslash G}^{g_1s(x\cdot g_1)^{-1}}c(h(x\cdot g_1)^{-1}, h(x\cdot g_1)s(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1})
×∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s′(x⋅g1g2)−1) \times \prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1})
s(x)−1−1−1−1s(x)^{-1}-1-1-1 ×∏x∈H\Gc(s(x)g1s(x⋅g1),h(x⋅g1))\times \prod_{x\in H\backslash G}c(s(x)g_1s(x\cdot g_1), h(x\cdot g_1))
×∏x∈H\Gs(x)−1(c(h(x)−1,s′(x)g1g2s′(x⋅g1g2)−1)−1 \times \prod_{x\in H\backslash G}^{s(x)^{-1}}(c(h(x)^{-1}, s'(x)g_1g_2s'(x\cdot g_1g_2)^{-1})^{-1}
×c(h(x)−1,s′(x)g1s′(x⋅g1)−1)) \times c(h(x)^{-1}, s'(x)g_1s'(x\cdot g_1)^{-1}))
=∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s′(x⋅g1g2)−1)=\prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1}) s(x)−1−1−1−1s(x)^{-1}-1-1-1 ×∏x∈H\Gc(s(x)g1s(x⋅g1),h(x⋅g1))\times \prod_{x\in H\backslash G}c(s(x)g_1s(x\cdot g_1), h(x\cdot g_1)) ×∏x∈H\Gg1s(x)−1c(h(x)−1,h(x)s(x)g2s′(x⋅g2)−1)\times \prod_{x\in H\backslash G}^{g_1s(x)^{-1}}c(h(x)^{-1}, h(x)s(x)g_2s'(x\cdot g_2)^{-1}) s(x)−1(−1′′−1−1s(x)^{-1}(-1''-1-1
×∏x∈H\Gc(h(x),s(x)g1g2s(x⋅g1g2)) \times \prod_{x\in H\backslash G}c(h(x), s(x)g_1g_2s(x\cdot g_1g_2))
×c(h(x)−1,s′(x)g1s′(x⋅g1)−1)) \times c(h(x)^{-1}, s'(x)g_1s'(x\cdot g_1)^{-1}))

for all g1,g2∈Gg_1, g_2\in G. The first equality comes from the cocycle relation (VIII, p. 295, formula (7)) applied to the elements

h(x)−1,h(x)s(x)g1s′(x⋅g1)−1h(x)^{-1},h(x)s(x)g_1s'(x\cdot g_1)^{-1}, and s′(x⋅g1)g2s′(x⋅g1g2)−1s'(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1};

the second is obtained by applying the cocycle relation to the elements

s(x)g1s(x⋅g1)−1,h(x⋅g1)−1s(x)g_1s(x\cdot g_1)^{-1},h(x\cdot g_1)^{-1}, and h(x⋅g1)s(x⋅g1)g2s′(x⋅g1g2)−1h(x\cdot g_1)s(x\cdot g_1)g_2s'(x\cdot g_1g_2)^{-1}; and the last simply uses the fact that the mapping x↦x⋅g1x\mapsto x\cdot g_1 is a permutation of H\GH\backslash G.

The last two lines of the obtained expression correspond to a 2-coboundary. We find that c~s′\widetilde{c}_{s'} has the same class in H2(G,F)H^2(G,F) as the cocycle whose value in (g1,g2)∈G2(g_1, g_2)\in G^2 is given by the expression

∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1h(x⋅g1g2)−1) \prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1}h(x\cdot g_1g_2)^{-1})
s(x)−1−1−1−1s(x)^{-1}-1-1-1
×∏x∈H\Gc(s(x)g1s(x⋅g1),h(x⋅g1)) \times \prod_{x\in H\backslash G}c(s(x)g_1s(x\cdot g_1), h(x\cdot g_1))
=∏x∈H\Gg1s(x⋅g1)−1c(s(x⋅g1)g2s(x⋅g1g2)−1,h(x⋅g1g2)−1)−1=\prod_{x\in H\backslash G}^{g_1s(x\cdot g_1)^{-1}}c(s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1}, h(x\cdot g_1g_2)^{-1})^{-1}
×∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1) \times \prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1})
×∏x∈H\Gs(x)−1(c(s(x)g1g2s(x⋅g1g2)−1,h(x⋅g1g2)−1) \times \prod_{x\in H\backslash G}^{s(x)^{-1}}(c(s(x)g_1g_2s(x\cdot g_1g_2)^{-1}, h(x\cdot g_1g_2)^{-1})
−1−1−1)-1-1-1) ×c(s(x)g1s(x⋅g1),h(x⋅g1))\times c(s(x)g_1s(x\cdot g_1), h(x\cdot g_1)) =∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1)=\prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1}) gs(x)−1−1−1−1gs(x)^{-1}-1-1-1
×∏x∈H\G1c(s(x)g2s(x⋅g2),h(x⋅g2)) \times \prod_{x\in H\backslash G^1}c(s(x)g_2s(x\cdot g_2), h(x\cdot g_2))
×∏x∈H\Gs(x)−1(c(s(x)g1g2s(x⋅g1g2)−1,h(x⋅g1g2)−1) \times \prod_{x\in H\backslash G}^{s(x)^{-1}}(c(s(x)g_1g_2s(x\cdot g_1g_2)^{-1}, h(x\cdot g_1g_2)^{-1})
×c(s(x)g1s(x⋅g1)−1,h(x⋅g1)−1)−1) \times c(s(x)g_1s(x\cdot g_1)^{-1}, h(x\cdot g_1)^{-1})^{-1})

where the first equality follows from the cocycle relation applied to the elements

s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1s(x)g_1s(x\cdot g_1)^{-1},s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1}, and h(x⋅g1g2)−1h(x\cdot g_1g_2)^{-1}.

The last two lines of the obtained expression correspond to a 2-coboundary. We find that the class of c~s′\widetilde{c}_{s'} coincides with that of c~s\widetilde{c}_s.

We have constructed a homomorphism from the group H2(H,F)H^2(H,F) to the group H2(G,F)H^2(G,F), which we call the corestriction homomorphism from H to G and denote by CorHG^G_H.

Proposition 8 00IW

Let H be a subgroup of G of finite index. The endomorphism CorHG∘^G_H\circ ResHG^G_H of the group H2(G,F)H^2(G,F) coincides with multiplication by the index (G : H).

Let α\alpha be an element of H2(G,F)H^2(G,F), and let cc be an element of Z2(G,F)Z^2(G,F) representing α\alpha. The element CorHG∘^G_H\circ ResHG(α)^G_H(\alpha ) is the class of the cocycle whose value in (g1,g2)∈G2(g_1, g_2)\in G^2 is given by the expression

∏x∈H\Gs(x)−1c(s(x)g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1) \prod_{x\in H\backslash G}^{s(x)^{-1}}c(s(x)g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1})
=∏x∈H\Gc(g1s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1) =\prod_{x\in H\backslash G}c(g_1s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1})
×∏x∈H\G(c(s(x)−1,s(x)g1g2s(x⋅g1g2)−1)−1c(s(x)−1,s(x)g1s(x⋅g1)−1)) \times \prod_{x\in H\backslash G}(c(s(x)^{-1}, s(x)g_1g_2s(x\cdot g_1g_2)^{-1})^{-1}c(s(x)^{-1}, s(x)g_1s(x\cdot g_1)^{-1}))
=∏x∈H\Gg1c(s(x⋅g1)−1,s(x⋅g1)g2s(x⋅g1g2)−1) =\prod_{x\in H\backslash G}^{g_1}c(s(x\cdot g_1)^{-1}, s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1})
×∏x∈H\Gc(g1,g2s(x⋅g1g2)−1)c(g1,s(x⋅g1)−1)−1 \times \prod_{x\in H\backslash G}c(g_1, g_2s(x\cdot g_1g_2)^{-1})c(g_1, s(x\cdot g_1)^{-1})^{-1}
×∏x∈H\G(c(s(x)−1,s(x)g1g2s(x⋅g1g2)−1)−1c(s(x)−1,s(x)g1s(x⋅g1)−1)) \times \prod_{x\in H\backslash G}(c(s(x)^{-1}, s(x)g_1g_2s(x\cdot g_1g_2)^{-1})^{-1}c(s(x)^{-1}, s(x)g_1s(x\cdot g_1)^{-1}))
=∏x∈H\Gc(g1,g2s(x⋅g1g2)−1)c(g1,s(x⋅g1)−1)−1=\prod_{x\in H\backslash G}c(g_1, g_2s(x\cdot g_1g_2)^{-1})c(g_1, s(x\cdot g_1)^{-1})^{-1}
×∏x∈H\Gg1c(s(x)−1,s(x)g2s(xg2)−1) \times \prod_{x\in H\backslash G}^{g_1}c(s(x)^{-1}, s(x)g_2s(xg_2)^{-1})
×∏x∈H\G(c(s(x)−1,s(x)g1g2s(x⋅g1g2)−1)−1c(s(x)−1,s(x)g1s(x⋅g1)−1)) \times \prod_{x\in H\backslash G}(c(s(x)^{-1}, s(x)g_1g_2s(x\cdot g_1g_2)^{-1})^{-1}c(s(x)^{-1}, s(x)g_1s(x\cdot g_1)^{-1}))

The first equality comes from the cocycle relation (VIII, p. 295, formula (7)) applied to the elements

s(x)−1,s(x)g1s(x⋅g1)−1s(x)^{-1},s(x)g_1s(x\cdot g_1)^{-1}, and s(x⋅g1)g2s(x⋅g1g2)−1s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1};

the second is obtained by applying the cocycle relation to the elements

g1,s(x⋅g1)−1g_1,s(x\cdot g_1)^{-1}, and s(x⋅g1)g2s(x⋅g1g2)−1s(x\cdot g_1)g_2s(x\cdot g_1g_2)^{-1}.

By removing a coboundary, we obtain that CorHG∘^G_H\circ ResHG(α)^G_H(\alpha ) is the class of the cocycle whose value in (g1,g2)∈G2(g_1, g_2)\in G^2 is given by the expression

∏x∈H\Gc(g1,g2s(x⋅g1g2)−1)c(g1,s(x⋅g1)−1)−1 \prod_{x\in H\backslash G}c(g_1, g_2s(x\cdot g_1g_2)^{-1})c(g_1, s(x\cdot g_1)^{-1})^{-1}
=∏x∈H\Gg1c(g2,s(x⋅g1g2)−1)−1c(g1g2,s(x⋅g1g2)−1)c(g1,g2) =\prod_{x\in H\backslash G}^{g_1}c(g_2, s(x\cdot g_1g_2)^{-1})^{-1}c(g_1g_2, s(x\cdot g_1g_2)^{-1})c(g_1, g_2)
×∏x∈H\Gc(g1,s(x⋅g1)−1)−1 \times \prod_{x\in H\backslash G}c(g_1, s(x\cdot g_1)^{-1})^{-1}
=∏x∈H\Gg1c(g2,s(x⋅g2)−1)−1c(g1g2,s(x⋅g1g2)−1)c(g1,s(x⋅g1)−1)−1 =\prod_{x\in H\backslash G}^{g_1}c(g_2, s(x\cdot g_2)^{-1})^{-1}c(g_1g_2, s(x\cdot g_1g_2)^{-1})c(g_1, s(x\cdot g_1)^{-1})^{-1}
×c(g1,g2)(G:H) \times c(g_1, g_2)^{(G:H)}

which proves the result.

7. Galois Algebras

Let K be a commutative field. If E is a K-algebra, then we denote its automorphism group by AutK(E)_K(E). If E is a Galois extension of the field K, then the group AutK(E)_K(E) is simply the Galois group Gal(E/K/K) (V, §10, No. 2, p. 58).

Let G be a group. A(K,G)A (K,G)-algebra is a K-algebra E endowed with a group homomorphism λ:G→\lambda : G\rightarrow AutK(E)_K(E). The homomorphism λ\lambda then endows E with the structure of a group with operators in G as well as the structure of a left K[G]-module with external law given by

((∑g∈Gμgg)x=∑g∈Gμgλ(g).x(14) ((\sum_{g\in G}\mu_gg)x=\sum_{g\in G}\mu_g\lambda (g).x \tag{14}

for every x∈Lx\in L and every element (μg)g∈G(\mu_g)_{g\in G} of K[G]. A morphism of (K,G)(K,G)-algebras is a morphism of algebras that is also a morphism of groups with operators.

For every family (Ei)i∈I(E_i)_{i\in I} of (K(K, G)-algebras, the product K-algebra ∏i∈IEi\prod_{i\in I}E_i endowed with its structure of a product group with operators is a (K(K, G)-algebra.

If E is a (K(K, G)-algebra, then the set EGE^G of elements of E invariant under G is a subalgebra of E.

Let E be a (K(K, G)-algebra, where G acts by λ:G→\lambda : G\rightarrow AutK(E)_K(E). If K′K' is an extension of K, then for any g∈Gg\in G, let λ′(g)\lambda '(g) be the automorphism IdK′⊗λ(g)_{K'}\otimes \lambda (g) of the K′K'-algebra L(K′)L_{(K')}. Then λ′:g↦λ′(g)\lambda ':g\mapsto \lambda '(g) endows L(K′)L_{(K')} with the structure of a (K′(K', G)-algebra.

Given a group H and H-sets X and Y, we denote by FH(X,Y)\mathscr{F}_H(X,Y) the set of morphisms of H-sets from X to Y. It is the set of mappings f:X→Yf: X\rightarrow Y such that f(hx)=hf(x)f(hx) =hf(x) for every h∈Hh\in H and x∈Xx\in X.

Let G be a group with identity element ee, let H be a subgroup of G, and let E be a (K(K, H)-algebra. The (K,G)(K,G)-algebra deduced from E by coinduction from H to G, denoted by CoindHG^G_H(E), is the K-algebra FH(G,E)\mathscr{F}_H(G,E) endowed with the action of G given by the homomorphism λ\lambda from G to AutK_K(CoindHG(E)^G_H(E)) defined by

(λ(g)⋅f)(g′)=f(g′g)(15) (\lambda (g)\cdot f)(g') =f(g'g) \tag{15}

for f∈f\in CoindHG(E)^G_H(E) and g,g′∈Gg, g'\in G.

Lemma 7 00IX

a) Let S be a subset of G such that every element of G can be written uniquely as hshs with h∈Hh\in H and s∈Ss\in S. Then the mapping that sends ff to (f(s))s∈S(f(s))_{s\in S} is an isomorphism from the K-algebra CoindHG(E)^G_H(E) to the K-algebra F(S,E)\mathscr{F}(S,E) of mappings from S to E.

b) The algebra CoindHG(E)^G_H(E) has finite degree over K if and only if E has finite degree over K and the index of H in G is finite. In this case, we have the formula

[CoindHG(E):K]=(G:H)[E:K]^G_H(E) : K] = (G : H)[E : K].

c) Let EHE^H be the algebra of invariants of the group H in E and CoindHG(E)G^G_H(E)^G that of the group G in CoindHG(E)^G_H(E). The mapping f↦f(e)f\mapsto f(e) from CoindHG(E)^G_H(E) to E restricts to an algebra isomorphism from CoindHG(E)G^G_H(E)^G to EHE^H.

Assertion a) follows from the definitions and implies b). By (15), a mapping from G to E is an element of CoindHG(E)^G_H(E) invariant under G if and only if it is constant with value an element of E invariant under H.

Remark 1 00IY

Let G be a group, H a subgroup of G, and N a subgroup of H. Let E be a (K(K, N)-algebra. Let α\alpha be an element of FH(G,FN(H,E))\mathscr{F}_H(G,\mathscr{F}_N(H,E)). We have the relation

α(g)(nh)=n(α(g)(h))(16) \alpha (g)(nh) =n(\alpha (g)(h)) \tag{16}

for every g∈G,h∈Hg\in G,h\in H, and n∈Nn\in N and the relations

α(hg)(h′)=(hα(g))(h′)=α(g)(h′h)(17) \alpha (hg)(h') = (h\alpha (g))(h') =\alpha (g)(h'h) \tag{17}

for all h,h′∈Hh, h'\in H and every g∈Gg\in G. Consequently, α(ng)(e)=α(g)(n)=\alpha (ng)(e) =\alpha (g)(n) = n(α(g)(e))n(\alpha (g)(e)) for every g∈Gg\in G and n∈Nn\in N. We can therefore consider the mapping

ψ:FH(G,FN(H,E))→FN(G,E) \psi :\mathscr{F}_H(G,\mathscr{F}_N(H,E))\rightarrow \mathscr{F}_N(G,E)

defined by the relation ψ(α)(g)=α(g)(e)\psi (\alpha )(g) =\alpha (g)(e) for α\alpha in FH(G,FN(H,E))\mathscr{F}_H(G,\mathscr{F}_N(H,E)) and gg in G. The mapping ψ\psi is an algebra isomorphism from CoindHG^G_H(CoindNH(E)^H_N(E)) to CoindNG(E)^G_N(E) whose inverse sends an element β\beta from FN(G,E)\mathscr{F}_N(G,E) to the mapping α\alpha defined by the relation α(g)(h)=β(hg)\alpha (g)(h) =\beta (hg) for g∈Gg\in G and h∈Hh\in H.

We now suppose given a finite group G and a reduced (V, §6, No. 6, p. 32) commutative K-algebra L of finite degree endowed with an action of G given by a homomorphism λ\lambda from G to AutK(L)_K(L). For x∈Lx\in L and g∈Gg\in G, we denote by g⋅xg\cdot x the transform of xx under the automorphism λ(g)\lambda (g) of L. Let S\mathscr{S} be the set of maximal ideals of L; we denote by g⋅mg\cdot \mathfrak{m} the transform of an element m\mathfrak{m} of S\mathscr{S} under the automorphism λ(g)\lambda (g) of L. It is an element of S\mathscr{S}. For every m\mathfrak{m} in S\mathscr{S}, the field L/mL/\mathfrak{m} is a finite extension of K. We write πm:L→L/m\pi_{\mathfrak{m}}: L\rightarrow L/\mathfrak{m} for the projection and denote by GmG_{\mathfrak{m}} the stabilizer of m\mathfrak{m} in G, that is, the set of g∈Gg\in G such that g⋅m=mg\cdot \mathfrak{m}=\mathfrak{m}. The K-algebra L/mL/\mathfrak{m} is endowed with an action of GmG_{\mathfrak{m}} through the homomorphism λm\lambda_{\mathfrak{m}} from GmG_{\mathfrak{m}} to AutK(L/m)_K(L/\mathfrak{m}) that sends an element hh of GmG_{\mathfrak{m}} to the automorphism of L/mL/\mathfrak{m} deduced from λ(h)\lambda (h) when passing to the quotients.

Let O\mathscr{O} be the set of orbits of G in S\mathscr{S}. Given an orbit σ∈O\sigma \in \mathscr{O}, set aσ=⋂m∈σm\mathfrak{a}_{\sigma}=\bigcap_{\mathfrak{m}\in\sigma}\mathfrak{m} and Lσ=L/aσL_{\sigma}= L/\mathfrak{a}_{\sigma}. Since aσ\mathfrak{a}_{\sigma} is invariant under G, by passing to the quotients, the action of G on L defines a homomorphism λσ\lambda_{\sigma} from G to AutK(Lσ)_K(L_{\sigma}). Finally, denote by πσ\pi_{\sigma} the canonical mapping from L to LσL_{\sigma}.

Lemma 8 00RL

a) For every g∈G,σ∈Og\in G,\sigma \in \mathscr{O}, and m∈σ\mathfrak{m}\in \sigma, we have

[Lσ:K]=[L_{\sigma}: K] = Card(σ\sigma )[L/m:K][L/\mathfrak{m}: K].

b) The mapping π:x↦(πσ(x))σ∈O\pi :x\mapsto (\pi_{\sigma}(x))_{\sigma\in\mathscr{O}} is an isomorphism of (K,G)(K,G)-algebras from L to ∏σ∈OLσ\prod_{\sigma\in\mathscr{O}}L_{\sigma}.

c) Denote by LG(L^G(resp. LσG)L^G_{\sigma}) the subalgebra of L (resp. LσL_{\sigma}) of elements invariant under the action of G. Then π\pi induces an isomorphism from LGL^G to ∏σ∈OLσG\prod_{\sigma\in\mathscr{O}}L^G_{\sigma}.

Since the algebra L is reduced and of finite degree, the intersection of the maximal ideals of L is reduced to 0 (VIII, p. 173, Corollary 2). Moreover, if m\mathfrak{m} and m′\mathfrak{m}' are two distinct maximal ideals of L, then we have m+m′=L\mathfrak{m}+\mathfrak{m}'= L. By Proposition 10 of I, §8, No. 11, p. 110, the canonical mapping from L to ∏m∈SL/m\prod_{\mathfrak{m}\in\mathscr{S}}L/\mathfrak{m} is an isomorphism, as is the canonical mapping from L/aσL/\mathfrak{a}_{\sigma} to ∏m∈σL/m\prod_{\mathfrak{m}\in\sigma}L/\mathfrak{m} for every σ∈O\sigma \in \mathscr{O}. Assertion a) follows. Since O\mathscr{O} is a partition of S\mathscr{S}, assertion b) follows; assertion c) is an immediate consequence of b).

Let us now fix an orbit σ∈O\sigma \in \mathscr{O} and an element m\mathfrak{m} of σ\sigma. Set FmF_{\mathfrak{m}} = CoindGmG(L/m)^G_{G_{\mathfrak{m}}}(L/\mathfrak{m}), and denote by λFm\lambda_{F_{\mathfrak{m}}} the action of G on FmF_{\mathfrak{m}}. For any x∈Lx\in L, denote by x‾\overline{x} the mapping from G to L/mL/\mathfrak{m} that sends gg to πm(gx)\pi_{\mathfrak{m}}(gx). By formula (15) of VIII, p. 305 and the definition of the action of GmG_{\mathfrak{m}} on L/mL/\mathfrak{m}, it is immediate that x‾\overline{x} belongs to FmF_{\mathfrak{m}}, and the mapping u:x↦x‾u:x\mapsto \overline{x} from L to FmF_{\mathfrak{m}} satisfies λFm(g)∘u=u∘λ(g)\lambda_{F_{\mathfrak{m}}}(g)\circ u=u\circ \lambda (g) for g∈Gg\in G. In other words, uu is a morphism of (K(K, G)-algebras.

Lemma 9 00IZ

The morphism uu is surjective with kernel aσ\mathfrak{a}_{\sigma}. The mapping deduced from uu by passing to the quotient is an isomorphism of (K,G)(K,G)-algebras from LσL_{\sigma} to FmF_{\mathfrak{m}}.

Since the kernel of πm\pi_{\mathfrak{m}} is equal to m\mathfrak{m}, that of uu is equal to ⋂g∈Gg−1.m=aσ\bigcap_{g\in G}g^{-1}.\mathfrak{m}=\mathfrak{a}_{\sigma}. To prove that uu is surjective, it suffices to prove that the vector spaces Lσ=L_{\sigma}= L/aσL/\mathfrak{a}_{\sigma} and FmF_{\mathfrak{m}} have the same dimension over K. Now, all ideals g⋅mg\cdot \mathfrak{m} have the same codimension in L, and, by Lemma 8, a),

[Lσ:K]=[L_{\sigma}: K] = Card(σ\sigma ).[L/m:K].[L/\mathfrak{m}: K].

Moreover, by Lemma 7, b), we have

[Fm:K]=(G:Gm)[L/m:K] [F_{\mathfrak{m}}: K] = (G : G_{\mathfrak{m}})[L/\mathfrak{m}: K]

Since we have Card(σ\sigma ) =(G:Gm)= (G : G_{\mathfrak{m}}), we have proved the equality [Lσ:K]=[L_{\sigma}: K] = [Fm:K][F_{\mathfrak{m}}: K].

We now suppose, furthermore, that the homomorphism λ\lambda from G to AutK(L)_K(L) is injective. We identify K with its image in L through the mapping ξ↦ξ⋅1\xi \mapsto \xi \cdot 1. Let Ω be an algebraically closed extension of K. The set F(G,Ω)\mathscr{F}(G,\Omega ) of mappings from G to Ω coincides with the coinduced (Ω(\Omega, G)-algebra Coind{e}G(Ω)^G_{\{e\}}(\Omega ). It is a free Ω[G]-module of rank 1. Let H\mathscr{H} be the set of K-algebra homomorphisms from L to Ω. We define a right action of G on H\mathscr{H} by (g,χ)↦χ∘λ(g)(g, \chi )\mapsto \chi \circ \lambda (g). The Ω-algebra F(H,Ω)\mathscr{F}(\mathscr{H},\Omega ) is then endowed with the structure of an (Ω(\Omega, G)-algebra deduced from the right action of G on H\mathscr{H}.

Lemma 10 00J0

Let L be a (K,G)(K,G)-algebra that is étale over K. The mapping ψ\psi from L(Ω)L_{(\Omega )} to F(H,Ω)\mathscr{F}(\mathscr{H},\Omega ) characterized by the relation

ψ(ξ⊗x)=(ξχ(x))χ∈H \psi (\xi \otimes x) = (\xi \chi (x))_{\chi\in\mathscr{H}}

is an isomorphism of (K,G)(K,G)-algebras.

Since L is étale, the mapping ψ\psi is an isomorphism of Ω-algebras (V, §6, No. 3, p. 30, Proposition 2 and V, §6, No. 3, p. 29, Proposition 1, c)). We have the relations

ψ\psi((Id ⊗λ(g)\otimes \lambda (g))(ξ⊗x)(\xi \otimes x)) =(ξ(χ∘λ(g))(x))χ∈H= (\xi (\chi \circ \lambda (g))(x))_{\chi\in\mathscr{H}}

for ξ∈Ω,x∈L\xi \in \Omega ,x\in L, and g∈Gg\in G. So ψ\psi is a morphism of (Ω(\Omega, G)-algebras.

Theorem 2 00J1

Let G be a finite group, and let L be a commutative K-algebra of finite degree endowed with an action of G given by an injective homomorphism λ\lambda from G to AutK(L)_K(L). The following properties are then equivalent:

(i) There exist a subgroup H of G, a Galois extension E of K of finite degree, an isomorphism from H to Gal(E/K/K), and an isomorphism of (K,G)(K,G)-algebras from L to CoindHG(E)^G_H(E).

(ii) The algebra L is étale, and H\mathscr{H} is a homogeneous principal G-set (I, §5, No. 6, p. 60, Definition 7).

(iii) There exists an isomorphism of (Ω,G)(\Omega ,G)-algebras ψ:L(Ω)→F(G,Ω)\psi : L_{(\Omega )}\rightarrow \mathscr{F}(G,\Omega ); in other words, for every g∈Gg\in G, the automorphism ψ∘λ(g)(Ω)∘ψ−1\psi \circ \lambda (g)_{(\Omega )}\circ \psi^{-1} of ΩG\Omega^G is equal to the automorphism

(xh)h∈G⟼(xhg)h∈G (x_h)_{h\in G}\longmapsto (x_{hg})_{h\in G}

(iv) The algebra L is reduced, and L is a free K[G]-module of rank 1.

(v) The algebra L is reduced, we have Card(G) = [L : K], and K is the subring of L of elements invariant under the action of G.

(vi) The algebra L is reduced, the group G acts transitively on the set of maximal ideals of L, and, for every maximal ideal m\mathfrak{m} of L, the stabilizer GmG_{\mathfrak{m}} of m\mathfrak{m} in G acts faithfully on L/mL/\mathfrak{m} and admits K as subfield of invariants.

(i)⇒\Rightarrow(ii): Let E be a Galois extension of K of finite degree and τ\tau an isomorphism from H to AutK(E)_K(E). Let S be a system of representatives of the right cosets of G modulo H. The K-algebra F = CoindHG(E)^G_H(E) is isomorphic to F(S,E)\mathscr{F}(S,E) (Lemma 7, a)); it is therefore étale. We denote by λF\lambda_F the action of G on F. Furthermore, let ψ\psi be a K-algebra homomorphism from E to Ω, and let χ0\chi_0 be the homomorphism f↦ψ(f(e))f\mapsto \psi (f(e)) from F to Ω. Let g∈Gg\in G be such that we have χ0∘λF(g)=χ0\chi_0\circ \lambda_F(g) =\chi_0; since ψ\psi is injective, we then have f(g)=f(e)f(g) =f(e) for every f∈Ff\in F. In view of Lemma 7, a), we have g∈Hg\in H, and by the formula

f(h)=τ(h)⋅f(e)(18) f(h) =\tau (h)\cdot f(e) \tag{18}

which holds for every h∈Hh\in H, this can only happen if g=eg=e. On the other hand, by Lemma 7, b) and Theorem 3 of V, §10, No. 6, p. 66, we have

[F : K] = (G : H)[E : K] = (G : H) Card(H) = Card(G).

The set K\mathscr{K} of K-homomorphisms from F to Ω has cardinal [F : K] because F is étale (V, §6, No. 5, p. 32, Proposition 4), so Card(K\mathscr{K}) == Card(G). Since the stabilizer of χ0\chi_0 in G is equal to {e}\{e\} by the above, K\mathscr{K} is a homogenous principal G-set.

(ii)⇒\Rightarrow(iii): Suppose that L is étale and that H\mathscr{H} is a homogenous principal G-set. By Lemma 10, the (Ω(\Omega, G)-algebras L(Ω)L_{(\Omega )} and F(H,Ω)\mathscr{F}(\mathscr{H},\Omega ) are isomorphic. Since H\mathscr{H} is a homogenous principal G-set, the (Ω(\Omega, G)-algebras F(H,Ω)\mathscr{F}(\mathscr{H},\Omega ) and F(G,Ω)\mathscr{F}(G,\Omega ) are isomorphic.

(iii)⇒\Rightarrow(iv): Suppose that property (iii) holds. Then L(Ω)L_{(\Omega )} is a free module of rank 1 over the algebra Ω[G]; the latter can be canonically identified with K[G](Ω)K[G]_{(\Omega )}. We then apply Theorem 3 of VIII, p. 37.

The implication (iv)⇒\Rightarrow(v) is immediate.

Let us prove the implication (v)⇒\Rightarrow(vi). The algebra L is reduced. By Lemma 8, c), the group G acts transitively on the set S\mathscr{S} of maximal ideals of L. Let m\mathfrak{m} be an element of S\mathscr{S}. By Lemma 9, since ⋂n∈Sn={0}\bigcap_{\mathfrak{n}\in\mathscr{S}}\mathfrak{n}=\{0\}, the algebra L is isomorphic to the algebra CoindGmG(L/m)^G_{G_{\mathfrak{m}}}(L/\mathfrak{m}). The algebra of invariants of GmG_{\mathfrak{m}} in L/mL/\mathfrak{m} therefore coincides with K by Lemma 7, c). Hence the homomorphism λm\lambda_{\mathfrak{m}} from GmG_{\mathfrak{m}} to Gal((L/m/\mathfrak{m})/K/K) is surjective. By Lemma 7, we moreover have

Card(G) =[L:K]=(G:Gm)[L/m:K]= [L : K] = (G : G_{\mathfrak{m}})[L/\mathfrak{m}: K]. So Card(Gm_{\mathfrak{m}}) =[L/m:K]= [L/\mathfrak{m}: K], and the homomorphism λm\lambda_{\mathfrak{m}} is injective.

It remains to prove the implication (vi)⇒\Rightarrow(i). Let m\mathfrak{m} be a maximal ideal of L. By Lemma 9, the algebra L is isomorphic to the algebra CoindGmG(L/m)^G_{G_{\mathfrak{m}}}(L/\mathfrak{m}) as a (K(K, G)-algebra. Since GmG_{\mathfrak{m}} acts faithfully in L/mL/\mathfrak{m} and admits K as subfield of invariants, the group homomorphism λm\lambda_{\mathfrak{m}} defines an isomorphism from GmG_{\mathfrak{m}} to Gal((L/m/\mathfrak{m})/K/K).

Remark 2 00J2

In Theorem 2, we can replace the assumption that Ω is algebraically closed with the assumption that Ω is separably closed. Indeed, if L is étale, then the image of every K-algebra homomorphism from L to Ω is a separable extension of K.

Remark 3 00J3

The normal basis theorem (V, §10, No. 9, p. 73, Theorem 6) is a specific case of the implication (i)⇒\Rightarrow(iv) in Theorem 2.

Definition 1 00J4

Let G be a finite group, and let L be a nonzero commutative algebra of finite degree over K endowed with the structure of a (K,G)(K,G)-algebra, where the action of G is given by an injective homomorphism from G to the group AutK(L)_K(L). We say that L is a Galois algebra with group G if it has the equivalent properties (i) through (vi) of Theorem 2.

Remark 4 00J5

Suppose that L is an extension of K endowed with an action λ\lambda of G. Then L is a Galois algebra over K if and only if L is a Galois extension of K and λ\lambda is an isomorphism from G to the Galois group of L over K.

Remark 5 00J6

Suppose that the group G is abelian. If G acts faithfully and transitively on a set X, then the stabilizer of any point of X is reduced to the identity (because the stabilizers of the points of X are all equal and their intersection is reduced to the identity element of G). Consequently, properties (i) through (vi) of Theorem 2 are also equivalent to the following:

(vii) The algebra L is étale, and G acts faithfully and transitively on H\mathscr{H}.

Remark 6 00J7

By V, §10, No. 1, p. 75, Lemma 5; V, §10, No. 1, p. 76, Proposition 12; and VIII, p. 137, Proposition 3, a Galois algebra is commutative and semisimple.

Example 1 00J8

Let nn be a strictly positive integer, prime to the characteristic exponent of K. Suppose that the group μn\mu_n of nn-th roots of unity in K has order nn. Then for every divisor dd of nn, the group μd\mu_d of dd-th roots of unity has order dd. Let aa be a nonzero element of K, let L be the algebra K[X]/(Xn−a)K[X]/(X^n-a), and let xx be the class of X in L. The sequence (1,x,...,xn−1)(1, x, . . . , x^{n-1}) is a basis of the vector space L over the field K, and we have xn=ax^n=a. Moreover, the polynomial Xn−aX^n-a is prime to its derivative nXn−1nX^{n-1}, so the algebra L is étale (V, §7, No. 2, p. 37, Proposition 3). For every ζ\zeta in μn\mu_n, the automorphism P(X)↦P(ζX)P(X)\mapsto P(\zeta X) of the ring K[X] defines, by passing to the quotients, an endomorphism λ(ζ)\lambda (\zeta ) of L because we have (ζX)n−a=Xn−a(\zeta X)^n-a= X^n-a; it is an automorphism. We have

λ(ζ)xi=ζixi(19) \lambda (\zeta )x^i=\zeta^ix^i \tag{19}

for 0⩽i<n0\leqslant i < n. The mapping λ:ζ↦λ(ζ)\lambda :\zeta \mapsto \lambda (\zeta ) is an injective homomorphism from μn\mu_n to AutK_K(L), and the ring of invariants of the group λ(μn)\lambda (\mu_n) in L is equal to K⋅1K\cdot 1. Since the cardinal of μn\mu_n is equal to n=[L:K]n= [L : K], the algebra L endowed with the action of λ\lambda is a Galois algebra (VIII, p. 308, Theorem 2, (v)).

Let rr be the least strictly positive integer such that ara^r belongs to K∗nK^{*n}; it divides nn, and there exists an element bb of K∗K^* such that a=bn/ra=b^{n/r}. Then (V, §11, No. 8, p. 91, Remark) the polynomial Xr−bX^r-b is irreducible, and we have Xn−a=∏ζ∈μn/r(Xr−ζb)X^n-a=\prod_{\zeta\in\mu_{n/r}}(X^r-\zeta b). Let E be the field K[Y]/(Yr−b)K[Y]/(Y^r-b), and let yy be the class of Y in E. There exists an isomorphism θ\theta from μn/r\mu_{n/r} to Gal(E/K/K), characterized by the relation θ(ξ)(y)=ξy\theta (\xi )(y) =\xi y (V, §11, No. 8, p. 91, Example 3). We then verify that the Galois algebra L is isomorphic to the (K,μn(K, \mu_n)-algebra Coindμn/rμn(E)^{\mu_n}_{\mu_{n/r}}(E).

Example 2 00J9

Now suppose that the field K has characteristic p≠0p\not= 0. Let cc be an element of K. The polynomial f=Xp−X−cf= X^p-X-c is prime to its derivative f′=−1f'=-1, so the algebra L=K[X]/(f)L = K[X]/(f) is étale (V, §7, No. 2, p. 37, Proposition 3). We denote the image of X in L by xx; we have the relation xp=x+cx^p=x+c. The sequence (1,x,...,xp−1)(1, x, . . . , x^{p-1}) is a basis of L viewed as a vector space over K.

Let P be the additive group of the prime subfield of K; it is a cyclic group of order pp, generated by the unit element 1 of K. For every jj in P, we have jp=jj^p=j (V, §1, No. 3, p. 4, formula (4)) and therefore f(X+j)=f(X)f(X +j) =f(X). Hence there exists an automorphism γ(j)\gamma (j) of the algebra L characterized by the relation γ(j)(x)=x+j\gamma (j)(x) =x+j; moreover, the resulting mapping γ\gamma is an injective homomorphism from P to AutK(L)_K(L).

Let Ω be an algebraically closed extension of K, and let ξ\xi be a root of the polynomial ff in Ω. We have ξp=ξ+c\xi^p=\xi +c, hence

Xp−X−c=(Xp−ξp)−(X−ξ)=(X−ξ)p−(X−ξ)=∏j∈P(X−ξ−j) X^p-X-c= (X^p-\xi^p)-(X-\xi ) = (X-\xi )^p-(X-\xi ) =\prod_{j\in P}(X-\xi -j)

by V, §12, No. 1, p. 94, formula (1). For every jj in P, there exists a unique algebra homomorphism χj:L→Ω\chi_j: L\rightarrow \Omega that sends xx to ξ+j\xi +j; moreover, every homomorphism from L to Ω is one of the χj\chi_j, and we have the relation χj=χ0∘γ(j)\chi_j=\chi_0\circ \gamma (j). The algebra L endowed with γ\gamma has property (ii) of Theorem 2 of VIII, p. 308 and is therefore a Galois algebra over K.

To describe the structure of L, we must distinguish between two cases:

a) We have ξ∉K\xi \notin K. Then the polynomial f(X)f(X) is irreducible in K[X] (V, §11, No. 9, p. 93, Example 3). In this case, L is a cyclic extension of K of degree pp, and γ\gamma is an isomorphism from P to Gal(L/K/K).

b) We have ξ∈K\xi \in K. Then the mapping ψ:y↦(χj(y))j∈P\psi :y\mapsto (\chi_j(y))_{j\in P} is an isomorphism from the algebra L to the product algebra KPK^P; moreover, ψ∘γ(k)∘ψ−1\psi \circ \gamma (k)\circ \psi^{-1} is the automorphism (xj)j∈P↦(xj+k)j∈P(x_j)_{j\in P}\mapsto (x_{j+k})_{j\in P} of KPK^P for every k∈Pk\in P.

8. Actions on Galois Algebras

Proposition 9 00JA

Let G be a finite group, H be a subgroup of G, and E be a Galois algebra over the field K with group H. Then the (K,G)(K,G)-algebra CoindHG(E)^G_H(E) deduced from E by coinduction from H to G is a Galois algebra over the field K with group G.

Since E is a Galois algebra over K, by property (v) of Theorem 2 of VIII, p. 308, it is reduced, we have Card(H) = [E : K], and K is the ring of invariants of H in E. But by Lemma 7 of VIII, p. 305, the algebra F = CoindHG(E)^G_H(E) is reduced, and we have

[F : K] = (G : H)[E : K] = (G : H) Card(H) = Card(G),

and K is the ring of invariants of G in F. So F is a Galois algebra by the criterion given in Theorem 2, (v).

Proposition 10 00JB

Let G be a finite group. Let L be a Galois algebra over the field K with group G, and let K′K' be an extension of K. Then the (K′,G)(K',G)-algebra L(K′)L_{(K')} is a Galois algebra over K′K'.

We use property (v) in Theorem 2 by observing that if the K-algebra L is étale, then the K′K'-algebra L(K′)L_{(K')} is too (V, §6, No. 5, p. 32, Corollary 2), that we have the equality [L(K′):K′]=[L:K][L_{(K')}: K'] = [L : K], and that the ring of invariants of G in L(K′)L_{(K')} is (LG)(K′)(L^G)_{(K')}, where LGL^G is the ring of invariants of G in L.

Proposition 11 00JC

Let G1G_1 and G2G_2 be groups. Let L1L_1 and L2L_2 be Galois algebras over K with respective actions λ1:G1→\lambda_1: G_1\rightarrow AutK(L1)_K(L_1) and λ2:G2→\lambda_2: G_2\rightarrow AutK(L2)_K(L_2). Set L=L1⊗KL2,G=G1×G2L = L_1\otimes_KL_2,G = G_1\times G_2, and λ(g1,g2)=λ1(g1)⊗λ2(g2)\lambda (g_1, g_2) =\lambda_1(g_1)\otimes \lambda_2(g_2) for (g1,g2)∈G(g_1, g_2)\in G. Then the K-algebra L endowed with the action λ\lambda is a Galois algebra over K.

We reason as before, taking the following into account: If L1L_1 and L2L_2 are étale, then so is the algebra L=L1⊗KL2L = L_1\otimes_KL_2 (V, §6, No. 5, p. 32, Corollary 1), and we have the equalities

[L:K]=[L1:K][L2:K][L : K] = [L_1: K][L_2: K] and Card(G) = Card(G1_1) Card(G2_2).

Moreover, if LiGiL^{G_i}_i denotes the ring of invariants of GiG_i in LiL_i, then it follows from Lemma 11 below that L1G1⊗KL2G2L^{G_1}_1\otimes_KL^{G_2}_2 is the ring of invariants of G1×G2G_1\times G_2 in L1⊗KL2L_1\otimes_KL_2.

Lemma 11 00JD

Let G1G_1 and G2G_2 be groups, and let W1W_1 and W2W_2 be K-vector spaces. We endow W1(W_1(resp. W2)W_2) with an action of G1(G_1(resp. G2)G_2) given by a group homomorphism ρ1:G1→\rho_1: G_1\rightarrow AutK(W1)(_K(W_1) (resp. ρ2:G2→\rho_2: G_2\rightarrow AutK(W2))_K(W_2)). We consider the group homomorphism

ρ1⊗ρ2:G1×G2⟶\rho_1\otimes \rho_2: G_1\times G_2\longrightarrow AutK(W1⊗KW2)_K(W_1\otimes_KW_2)

defined by the relation (ρ1⊗ρ2)(g1,g2)(w1⊗w2)=ρ1(g1)(w1)⊗ρ2(g2)(w2)(\rho_1\otimes \rho_2)(g_1, g_2)(w_1\otimes w_2) =\rho_1(g_1)(w_1)\otimes \rho_2(g_2)(w_2) for g1∈G1,g2∈G2,w1∈W1g_1\in G_1,g_2\in G_2,w_1\in W_1, and w2∈W2w_2\in W_2. Then the linear mapping from W1G1⊗KW2G2W^{G_1}_1\otimes_KW^{G_2}_2 to W1⊗KW2W_1\otimes_KW_2 given by the tensor product of the canonical injections induces an isomorphism of K-vector spaces from W1G1⊗KW2G2W_1^{G_1}\otimes_KW^{G_2}_2 to (W1⊗KW2)G1×G2(W_1\otimes_KW_2)^{G_1\times G_2}.

This follows from Lemma 1 of VIII, p. 213 applied to the K[G]-modules M1=M2=KM_1= M_2= K endowed with the trivial action of G,N1=W1G, N_1= W_1, and N2=W2N_2= W_2.

Remark 00JE

Let L be a Galois extension of the field K of finite degree, and let G be its Galois group; we denote the identity mapping on G by λ\lambda. Then L endowed with λ\lambda is a Galois algebra over K with group G. Let K′K' be an extension of K. By Proposition 10, the algebra L(K′)L_{(K')} is Galois over K′K', but it is not in general an extension of K′K'. Analogously, by Proposition 11, the tensor product of Galois extensions E and F of K of finite degree can be viewed as a Galois algebra; in general, it is not a Galois extension of K. It is one, however, if E and F are moreover linearly disjoint subextensions of an extension of K (V, §2, No. 5, p. 13 and V, §10, No. 1, p. 57, Proposition 1).

9. Cross Products

Let K be a field, and let G be a group whose identity element we denote by ee. Let L be a commutative K-algebra, and let λ\lambda be a homomorphism from G to the automorphism group of the K-algebra L. For any gg in G, let τ(g)\tau (g) be the automorphism of the multiplicative group L∗L^* of L induced by λ(g)\lambda (g).

Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension of the group G by L∗L^*. Define a right action of the group L∗L^* on the set L×ΓL\times \Gamma by

(β,γ).α=(βα,ι(α)−1γ)(20) (\beta , \gamma ).\alpha = (\beta \alpha , \iota (\alpha )^{-1}\gamma ) \tag{20}

for α∈L∗,β∈L\alpha \in L^*,\beta \in L, and γ∈Γ\gamma \in \Gamma. We denote by E the set of orbits of L∗L^* in L×ΓL\times \Gamma and by [β;γ][\beta ;\gamma ] the orbit of the pair (β,γ)(\beta , \gamma ). We therefore have, by construction, the relation

[βα;γ]=[β;ι(α)γ](21) [\beta \alpha ;\gamma ] = [\beta ;\iota (\alpha )\gamma ] \tag{21}

for α∈L∗,β∈L\alpha \in L^*,\beta \in L, and γ∈Γ\gamma \in \Gamma.

Given β\beta in L and γ\gamma in Γ, we denote by γβ^{\gamma}\beta the transform of β\beta under the automorphism λ∘π(γ)\lambda \circ \pi (\gamma ) of L; we have the relations

(22) γ(γ′β)=γγ′β,γ(β+β′)=γβ+γβ′^{\gamma}(^{\gamma'}\beta ) =^{\gamma \gamma'}\beta ,^{\gamma}(\beta +\beta ') =^{\gamma}\beta +^{\gamma}\beta ', and γ(ββ′)=γβγβ′^{\gamma}(\beta \beta ') =^{\gamma}\beta^{\gamma}\beta '

for γ,γ′\gamma , \gamma ' in Γ and β,β′\beta , \beta ' in L. There exists a law of composition on E characterized by the relation

[β;γ][β′;γ′]=[βγβ′;γγ′](23) [\beta ;\gamma ][\beta ';\gamma '] = [\beta^{\gamma}\beta ';\gamma \gamma '] \tag{23}

Indeed, it suffices to verify that the right-hand side does not change if we replace, respectively, β,γ,β′\beta ,\gamma ,\beta ', and γ′\gamma ' with βα,ι(α)−1γ,β′α′\beta \alpha ,\iota (\alpha )^{-1}\gamma ,\beta '\alpha ', and ι(α′)−1γ′\iota (\alpha ')^{-1}\gamma ' for all α,α′\alpha ,\alpha ' in L∗L^*. This follows immediately by applying to E\mathscr{E} formula (1) of VIII, p. 285, which can also be written as γι(α)=ι(γα)γ\gamma \iota (\alpha ) =\iota (^{\gamma}\alpha )\gamma for γ∈Γ\gamma \in \Gamma and α∈L∗\alpha \in L^*. By the formulas in (22), the set E endowed with this law is a monoid with identity element [1;e][1;e].

Since π∘ι\pi \circ \iota is constant with value ee, there exists a mapping π~\widetilde{\pi} from E to G such that we have

π~([β;γ])=π(γ)(24) \widetilde{\pi}([\beta ;\gamma ]) =\pi (\gamma ) \tag{24}

for β∈L\beta \in L and γ∈Γ\gamma \in \Gamma.

Let gg be an element of G, and let Eg=π~−1(g)E_g=\widetilde{\pi}^{-1}(g). If γ0\gamma_0 is a fixed element of π−1(g)\pi^{-1}(g), then the mapping β↦[β;γ0]\beta \mapsto [\beta ;\gamma_0] is a bijection from L to EgE_g, thanks to which we will transfer to EgE_g the K-vector space structure on L. Since π−1(g)\pi^{-1}(g) is composed of elements of the form ι(α)γ0\iota (\alpha )\gamma_0, where α\alpha runs through L∗L^*, and we have

[β;ι(α)γ0]=[βα;γ0] [\beta ;\iota (\alpha )\gamma_0] = [\beta \alpha ;\gamma_0]

the vector space structure on EgE_g does not depend on the choice of γ0\gamma_0.

Let gg and g′g' be elements of G; by formulas (22) and (23), the law of composition on E induces by restriction a K-bilinear mapping from Eg×Eg′E_g\times E_{g'} to Egg′E_{gg'}. Consequently, the vector space P=⨁g∈GEgP =\bigoplus_{g\in G}E_g is endowed with the structure of an associative and unital algebra, whose multiplication induces the previous bilinear mapping from Eg×Eg′E_g\times E_{g'} to Egg′E_{gg'} for all gg and g′g' in G. The algebra P is called the cross product of L and E\mathscr{E} and is denoted by A[E;L]\mathbf{A}[\mathscr{E}; L]; its unit element is the element [1;e][1;e] of EeE_e.

Set

u(β)=[β;e](25) u(\beta ) = [\beta ;e] \tag{25}

for β\beta in L. Then u:L→A[E;L]u: L\rightarrow \mathbf{A}[\mathscr{E}; L] is an injective homomorphism of K-algebras. By (23), for every γ∈Γ\gamma \in \Gamma, the element [1;γ][1;\gamma ] is invertible in A[E,L]\mathbf{A}[\mathscr{E},L], and the mapping v:Γ→A[E,L]∗v: \Gamma \rightarrow \mathbf{A}[\mathscr{E},L]^* that sends γ\gamma to [1;γ][1;\gamma ] is an injective group homomorphism. The homomorphisms uu and vv are called canonical. We have the relations

u(α)=v(ι(α))(26) u(\alpha ) =v(\iota (\alpha )) \tag{26}
u(γβ)=v(γ)u(β)v(γ)−1(27) u(^{\gamma}\beta ) =v(\gamma )u(\beta )v(\gamma )^{-1} \tag{27}
[β;γ]=u(β)v(γ)(28) [\beta ;\gamma ] =u(\beta )v(\gamma ) \tag{28}

for α∈L∗,β∈L\alpha \in L^*,\beta \in L, and γ∈Γ\gamma \in \Gamma.

Conversely, we have the following universal property of the algebra A[E;L]\mathbf{A}[\mathscr{E}; L].

Proposition 12 00JF

Let B be a K-algebra, u′:L→Bu': L\rightarrow B a K-algebra homomorphism, and v′:Γ→B∗v': \Gamma \rightarrow B^* a group homomorphism. We assume satisfied the relations

u′(α)=v′(ι(α))(29) u'(\alpha ) =v'(\iota (\alpha )) \tag{29}
u′(γβ)=v′(γ)u′(β)v′(γ)−1(30) u'(^{\gamma}\beta ) =v'(\gamma )u'(\beta )v'(\gamma )^{-1} \tag{30}

for α∈L∗,β∈L\alpha \in L^*,\beta \in L, and γ∈Γ\gamma \in \Gamma. Then there exists a unique algebra homomorphism ff from A[E;L]\mathbf{A}[\mathscr{E}; L] to B such that we have u′=f∘uu'=f\circ u and v′=f∘vv'=f\circ v.

To prove the uniqueness of ff, observe that the vector space A[E;L]\mathbf{A}[\mathscr{E}; L] over the field K is generated by the set of elements of the form [β;γ]=u(β)v(γ)[\beta ;\gamma ] =u(\beta )v(\gamma ). If the homomorphism f′:A[E;L]→Bf':\mathbf{A}[\mathscr{E}; L]\rightarrow B satisfies f′∘u=u′f'\circ u=u' and f′∘v=v′f'\circ v=v', then it sends [β;γ][\beta ;\gamma ] to u′(β)v′(γ)u'(\beta )v'(\gamma ) and therefore coincides with ff.

By formula (29), we have

u′(βα)v′(ι(α)−1γ)=u′(β)v′(γ)(31) u'(\beta \alpha )v'(\iota (\alpha )^{-1}\gamma ) =u'(\beta )v'(\gamma ) \tag{31}

for α\alpha in L∗,βL^*,\beta in L, and γ\gamma in Γ. By the definition of E, there consequently exists a mapping f0:E→Bf_0: E\rightarrow B such that f0([β;γ])=u′(β)v′(γ)f_0([\beta ;\gamma ]) =u'(\beta )v'(\gamma ). By formulas (23) and (30), we have f0(xx′)=f0(x)f0(x′)f_0(xx') =f_0(x)f_0(x') for xx and x′x' in E. The restriction of f0f_0 to EgE_g is K-linear for every element gg of G; consequently, there exists a unique K-linear mapping ff from A[E;L]\mathbf{A}[\mathscr{E}; L] to B that coincides with f0f_0 on E. The mapping ff is an algebra morphism that satisfies u′=f∘uu'=f\circ u and v′=f∘vv'=f\circ v.

Remark 00JG

Let σ:G→Γ\sigma : G\rightarrow \Gamma be a section of the mapping π\pi. Set εg=v(σ(g))\varepsilon_g=v(\sigma (g)) for every gg in G, and denote by cσc_{\sigma} the 2-cocycle associated with σ\sigma (VIII, p. 296). In particular, we have

εgεg′=u(cσ(g,g′))εgg′(32) \varepsilon_g\varepsilon_{g'}=u(c_{\sigma}(g, g'))\varepsilon_{gg'} \tag{32}

for all g,g′∈Gg, g'\in G. Let us, moreover, identify L with a subalgebra of A[E;L]\mathbf{A}[\mathscr{E}; L] via the homomorphism uu. Then every element of A[E;L]\mathbf{A}[\mathscr{E}; L] can be written uniquely as ∑g∈Gagεg\sum_{g\in G}a_g\varepsilon_g, where (ag)(a_g) is a family of elements of L with finite support. The multiplication in A[E;L]\mathbf{A}[\mathscr{E}; L] can be expressed by the formula

(∑gagεg)(∑gbgεg)=∑gdgεg(33) (\sum_ga_g\varepsilon_g)(\sum_gb_g\varepsilon_g)=\sum_gd_g\varepsilon_g \tag{33}

with

dg=∑hh′=gah(λ(h)⋅bh′)cσ(h,h′)(34) d_g=\sum_{hh'=g}a_h(\lambda (h)\cdot b_{h'})c_{\sigma}(h, h') \tag{34}

If the extension E\mathscr{E} is semitrivial, then we can choose a section σ\sigma of π\pi that is a group morphism from G to Γ; the cocycle cσc_{\sigma} is therefore constant with value 1, and formula (34) simplifies to

dg=∑hh′=gahλ(h)⋅bh′(35) d_g=\sum_{hh'=g}a_h\lambda (h)\cdot b_{h'} \tag{35}

Let K′K' be an extension of the field K. Denote by L′L' the K′K'-algebra L(K′)L_{(K')}, and for any gg in G, denote by λ′(g)\lambda '(g) the automorphism λ(g)(K′)\lambda (g)_{(K')} induced by λ(g)\lambda (g) on L(K′)L_{(K')}. Moreover, let us denote by τ′(g)\tau '(g) the automorphism of L′∗L^{'*} induced by λ′(g)\lambda '(g). Finally, let hh be the homomorphism from L∗L^* to L′∗L^{'*} that sends xx to x⊗1x\otimes 1. Let E′=(Γ′,ι′,π′)\mathscr{E}'= (\Gamma ', \iota ', \pi ') be the direct image of the extension of E\mathscr{E} by hh (VIII, p. 289). Let A[E′,L′]\mathbf{A}[\mathscr{E}',L'] be the cross product K′K'-algebra of E′\mathscr{E}' and L′L', and let u′:L′→A[E′,L′]u': L'\rightarrow \mathbf{A}[\mathscr{E}',L'] and v′:Γ′→A[E′,L′]∗v': \Gamma '\rightarrow \mathbf{A}[\mathscr{E}',L']^* be the canonical homomorphisms.

Proposition 13 00JH

There exists a unique isomorphism of K′K'-algebras φ\varphi from A[E;L](K′)\mathbf{A}[\mathscr{E}; L]_{(K')} to A[E′;L′]\mathbf{A}[\mathscr{E}'; L'] that satisfies the relations

u′(h(β))=φ(1⊗u(β))(36) u'(h(\beta )) =\varphi (1\otimes u(\beta )) \tag{36}

for β∈L\beta \in L and

v′(h(γ))=φ(1⊗v(γ))(37) v'(h(\gamma )) =\varphi (1\otimes v(\gamma )) \tag{37}

for γ∈Γ\gamma \in \Gamma.

The proof follows immediately from the constructions.

The isomorphism φ\varphi is called canonical.

10. Application to the Brauer Group

In this subsection, K is a field, and L is a Galois algebra over K with action λ:G→\lambda : G\rightarrow AutK(L)_K(L). We denote the degree of L over K by nn. We have n=n= Card(G).

Theorem 3 00JI

Let E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) be a τ\tau-extension of G by L∗L^*. The algebra A[E;L]\mathbf{A}[\mathscr{E}; L] is central simple and of degree n2n^2 over K. Moreover, the canonical homomorphism uu from L to A[E;L]\mathbf{A}[\mathscr{E}; L] is an isomorphism from L to a maximal commutative subalgebra of A[E;L]\mathbf{A}[\mathscr{E}; L].

Lemma 12 00JJ

a) There does not exist any ideal of L, other than {0}\{0\} and L that is invariant under G.

b) Let gg be an element of G distinct from the identity element, and let ag\mathfrak{a}_g be the ideal of L generated by the elements of the form x−λ(g)xx-\lambda (g)x, where xx runs through L. We have ag=L\mathfrak{a}_g= L.

Let S\mathscr{S} be the set of maximal ideals of L; for any subset S of S\mathscr{S}, set a(S)=⋂m∈Sm\mathfrak{a}(S) =\bigcap_{\mathfrak{m}\in S}\mathfrak{m}. Since the ring L is commutative and semisimple (VIII, p. 310, Remark 6), the map W↦a(W)W\mapsto \mathfrak{a}(W) is a bijection from the set of subsets of S\mathscr{S} to the set of ideals of L (VIII, p. 142, Proposition 9). The ideal a(S)\mathfrak{a}(S) is invariant under G if and only if S is. Since G acts transitively on S\mathscr{S} (VIII, p. 308, Theorem 2, vi)), the only subsets of S\mathscr{S} invariant under G are ∅\emptyset and S\mathscr{S}. Since a(∅)=L\mathfrak{a}(\emptyset ) = L and a(S)={0}\mathfrak{a}(\mathscr{S}) =\{0\}, we have proved assertion a).

Let us prove b) by contradiction; suppose that we have ag≠L\mathfrak{a}_g\not= L. Let m\mathfrak{m} be a maximal ideal of L containing ag\mathfrak{a}_g. We then have λ(g)x≡x\lambda (g)x\equiv x mod m\mathfrak{m} for every xx in L. In particular, m\mathfrak{m} is invariant under gg, and λ(g)\lambda (g) is induced by the identity on the field L/mL/\mathfrak{m}. By property (vi) in Theorem 2 of VIII, p. 308, the element gg of GmG_{\mathfrak{m}} is trivial, which contradicts the assumption of b).

Let us now prove Theorem 3. The vector space A[E;L]\mathbf{A}[\mathscr{E}; L] has finite dimension Card(G)[L:K]=n2: K] =n^2 over K, by VIII, p. 316 and the assumed equality Card(G) =[L:K]=n= [L : K] =n.

Let a\mathfrak{a} be a nonzero two-sided ideal of the algebra A[E;L]\mathbf{A}[\mathscr{E}; L]. We use the notation of VIII, p. 316. Every element aa of A[E;L]\mathbf{A}[\mathscr{E}; L] can be written uniquely as a=∑g∈Gagεga=\sum_{g\in G}a_g\varepsilon_g with ag∈La_g\in L for every g∈Gg\in G; we denote by Φ(a)\Phi (a) the set of elements gg of G such that ag≠0a_g\not= 0. By formula (32) of VIII, p. 316, we have the relation

εgεg′=cσ(g,g′)εgg′(38) \varepsilon_g\varepsilon_{g'}=c_{\sigma}(g, g')\varepsilon_{gg'} \tag{38}

for all g,g′∈Gg, g'\in G and consequently

Φ(aεg)=Φ(a).g(39) \Phi (a\varepsilon_g) = \Phi (a).g \tag{39}

for every g∈Gg\in G and a∈A[E;L]a\in \mathbf{A}[\mathscr{E}; L].

Let aa be a nonzero element of a\mathfrak{a} for which Φ(a)\Phi (a) is minimal for the inclusion; by formula (39), if necessary replacing aa with an element of the form aεg−1a\varepsilon_{g^{-1}} with g∈Φ(a)g\in \Phi (a), we may assume that we have e∈Φ(a)e\in \Phi (a). Let ss be an element of Φ(a)\Phi (a) distinct from ee. By Lemma 12, b), there exists an element xx of L such that as(x−λ(s)⋅x)≠0a_s(x-\lambda (s)\cdot x)\not= 0. But we have the relation

xa−ax=∑gag(x−λ(g)⋅x)εg(40) xa-ax=\sum_ga_g(x-\lambda (g)\cdot x)\varepsilon_g \tag{40}

where the sum is taken over the elements gg of Φ(a)\Phi (a) distinct from ee. Because of the minimality of Φ(a)\Phi (a), we then have xa−ax=0xa-ax= 0, but this contradicts the assumption on xx. We have, therefore, proved by contradiction that Φ(a)\Phi (a) contains only the identity element ee of G, so we have a∈La\in L.

Hence L∩aL\cap \mathfrak{a} is an ideal of L not reduced to {0}\{0\}. Moreover, for every xx in L, we have εgxεg−1=λ(g)⋅x\varepsilon_gx\varepsilon^{-1}_g=\lambda (g)\cdot x, so L∩aL\cap \mathfrak{a} is invariant under G. By Lemma 12, we therefore have L∩a=LL\cap \mathfrak{a}= L, that is, L⊂aL\subset \mathfrak{a}. Since L contains the identity element of A[E;L]\mathbf{A}[\mathscr{E}; L], we have a=A[E;L]\mathfrak{a}=\mathbf{A}[\mathscr{E}; L].

Since the algebra A[E;L]\mathbf{A}[\mathscr{E}; L] has nonzero finite dimension over K and its only two-sided ideals are {0}\{0\} and A[E;L]\mathbf{A}[\mathscr{E}; L], it is simple.

Let us prove that every element aa of A[E;L]\mathbf{A}[\mathscr{E}; L] that commutes with L belongs to L. We write aa as ∑g∈Gagεg\sum_{g\in G}a_g\varepsilon_g with coefficients aga_g in L. For every xx in L, we have xa−ax=0xa-ax= 0, and relation (40) shows that we have

ag(x−λ(g)⋅x)=0(41) a_g(x-\lambda (g)\cdot x) = 0 \tag{41}

for every g∈Gg\in G and x∈Lx\in L. By Lemma 12, we therefore have ag=0a_g= 0 for g≠eg\not=e; consequently, a∈La\in L.

Let us, finally, determine the center of A[E;L]\mathbf{A}[\mathscr{E}; L]. If zz belongs to the center, then it commutes with L, so belongs to L. But we then have

0=εgz−zεg=(λ(g)⋅z−z)εg 0 =\varepsilon_gz-z\varepsilon_g= (\lambda (g)\cdot z-z)\varepsilon_g

for every gg of G. Hence zz is invariant under the automorphism group λ(G)\lambda (G) of L. We therefore have z∈Kz\in K by assumption.

Theorem 4 00JK

Let A be a central simple algebra of finite degree over K, and let L be a maximal commutative subalgebra of A. Then there exists a τ\tau-extension E\mathscr{E} of G by L∗L^* such that A is isomorphic to A[E;L]\mathbf{A}[\mathscr{E}; L].

Without loss of generality, we may assume that L is a maximal commutative subalgebra of A. Let Γ be the multiplicative group consisting of the invertible elements γ\gamma of A such that there exists a gg in G with

γxγ−1=λ(g).x(42) \gamma x\gamma^{-1}=\lambda (g).x \tag{42}

for every element xx of L. If γ∈Γ\gamma \in \Gamma is given, then the element gg satisfying this relation is unique; we denote it by π(γ)\pi (\gamma ). It is immediate that π\pi is a homomorphism from Γ to G and that its kernel is equal to L∗L^*. The surjectivity of π\pi follows from the Skolem–Noether theorem (VIII, p. 263, Corollary).

If we denote the canonical injection of L∗L^* into Γ by ι\iota, then it follows from the constructions that E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ) is a τ\tau-extension of G by L∗L^*. Let

u:L→A[E;L]u: L\rightarrow \mathbf{A}[\mathscr{E}; L] and v:Γ→A[E;L]v: \Gamma \rightarrow \mathbf{A}[\mathscr{E}; L]

be the canonical homomorphisms. By the universal property of A[E;L]\mathbf{A}[\mathscr{E}; L] (VIII, p. 315, Proposition 12), there exists a unique algebra homomorphism f:A[E;L]→Af:\mathbf{A}[\mathscr{E}; L]\rightarrow A such that f∘u(x)=xf\circ u(x) =x and f∘v(γ)=γf\circ v(\gamma ) =\gamma for x∈Lx\in L and γ∈Γ\gamma \in \Gamma. Since the algebra A[E;L]\mathbf{A}[\mathscr{E}; L] is simple, the homomorphism ff is injective. Now, the algebra A[E;L]\mathbf{A}[\mathscr{E}; L] has degree n2n^2 over K and so does A because L is a maximal commutative semisimple subalgebra of A and we have n=[L:K]n= [L : K] (VIII, p. 262, Proposition 3, (ii)). Hence ff is bijective.

Definition 2 00JL

Let S\mathscr{S} be the set of maximal ideals of L. We define

Br(L/K/K) =⋂m∈S=\bigcap_{\mathfrak{m}\in\mathscr{S}} Ker(r(L/m)/Kr_{(L/\mathfrak{m})/K}), where r(L/m)/K:r_{(L/\mathfrak{m})/K}: Br(K) →\rightarrow Br(L/m/\mathfrak{m}) is the extension of scalars homomorphism (VIII, p. 281).

Theorem 5 00JM

There exists a group isomorphism

Ψ : Exτ(G,L∗)⟶_{\tau}(G,L^*)\longrightarrow Br(L/K/K)

that sends the class of a τ\tau-extension E\mathscr{E} of G by L∗L^* to the class in Br(L/K/K) of the algebra A[E;L]\mathbf{A}[\mathscr{E}; L].

To define Ψ and verify that it is a bijection, we must establish the following points:

a) If E\mathscr{E} and E′\mathscr{E}' are isomorphic τ\tau-extensions of G by L∗L^*, then the algebras A[E;L]\mathbf{A}[\mathscr{E}; L] and A[E′;L]\mathbf{A}[\mathscr{E}'; L] are isomorphic.

b) Conversely, if the algebras A[E;L]\mathbf{A}[\mathscr{E}; L] and A[E′;L]\mathbf{A}[\mathscr{E}'; L] are isomorphic, then the τ\tau-extensions E\mathscr{E} and E′\mathscr{E}' of G by L∗L^* are isomorphic.

c) In every class in Br(L/K/K), there is an algebra E containing L as a maximal commutative subalgebra.

d) If E is a central simple algebra of finite degree over K containing L as a maximal commutative subalgebra, then there exists a τ\tau-extension E\mathscr{E} of G by L∗L^* such that E is isomorphic to A[E;L]\mathbf{A}[\mathscr{E}; L].

Assertion a) follows from the construction of the cross product; b) follows from VIII, p. 263, Corollary. Assertion c) follows from Proposition 5 of VIII, p. 281, and d) is simply Theorem 4 above.

It remains to verify that Ψ is a group homomorphism; for this, it suffices to prove that if E1=(Γ1,ι1,π1)\mathscr{E}_1= (\Gamma_1, \iota_1, \pi_1) and E2=(Γ2,ι2,π2)\mathscr{E}_2= (\Gamma_2, \iota_2, \pi_2) are τ\tau-extensions, then the algebra A[E1E2;L]\mathbf{A}[\mathscr{E}_1\mathscr{E}_2; L] is equivalent to the algebra A[E1;L]⊗A[E2;L]\mathbf{A}[\mathscr{E}_1; L]\otimes \mathbf{A}[\mathscr{E}_2; L]. We denote the product extension E1E2\mathscr{E}_1\mathscr{E}_2 by E=(Γ,ι,π)\mathscr{E}= (\Gamma , \iota , \pi ). The group Γ is isomorphic to the cokernel of the homomorphism ρ\rho from L∗L^* to the fiber product Γ1×GΓ2\Gamma_1\times_G\Gamma_2 that sends μ\mu to (ι1(μ),ι2(μ)−1)(\iota_1(\mu), \iota_2(\mu)^{-1}). For i∈{1,2}i\in \{1,2\}, write Ai=A[Ei;L]A_i=\mathbf{A}[\mathscr{E}_i; L]. Let ui:L→Aiu_i: L\rightarrow A_i and vi:Γi→Ai∗v_i: \Gamma_i\rightarrow A^*_i be the canonical homomorphisms. We identify L with its images by the homomorphisms uiu_i, which make L into maximal commutative subalgebras of the AiA_i. We denote by ViV_i the L-vector space defined by left multiplication in AiA_i. The ring L⊗KAioL\otimes_KA^o_i acts on ViV_i. Since it is simple and has dimension n2n^2 over L, we obtain an isomorphism from L⊗KAioL\otimes_KA^o_i to EndL(Vi)_L(V_i). Hence the ring L⊗KA1o⊗KA2oL\otimes_KA^o_1\otimes_KA^o_2, which we can identify with (L⊗KA1o)⊗L(L⊗KA2o)(L\otimes_KA^o_1)\otimes_L(L\otimes_KA^o_2), is isomorphic to EndL(V1⊗LV2)_L(V_1\otimes_LV_2). Set C = EndA1o⊗KA2o(V1⊗LV2)_{A^o_1\otimes_KA^o_2}(V_1\otimes_LV_2). By Lemma 3 of VIII, p. 282, the ring C is similar to A1⊗KA2A_1\otimes_KA_2, and L⊗1⊗1L\otimes 1\otimes 1 is a maximal commutative subalgebra of C. For every pair (γ1,γ2)∈Γ1×Γ2(\gamma_1, \gamma_2)\in \Gamma_1\times \Gamma_2 satisfying π1(γ1)=π2(γ2)\pi_1(\gamma_1) =\pi_2(\gamma_2), we denote by w(γ1,γ2)w(\gamma_1, \gamma_2) the unique λ(π1(γ1\lambda (\pi_1(\gamma_1))-semilinear (II, §1, No. 13, p. 223) endomorphism such that

w(γ1,γ2)(x1⊗x2)=v1(γ1)x1⊗v2(γ2)x2 w(\gamma_1, \gamma_2)(x_1\otimes x_2) =v_1(\gamma_1)x_1\otimes v_2(\gamma_2)x_2

for x1∈V1x_1\in V_1 and x2∈V2x_2\in V_2. We have w(γ1,γ2)∈C∗w(\gamma_1, \gamma_2)\in C^*, and ww is a group homomorphism from the fiber product Γ1×GΓ2\Gamma_1\times_G\Gamma_2 to C∗C^*. This homomorphism is trivial on the image of ρ\rho and induces a homomorphism vv from Γ to C∗C^*. Denote by u:L→Cu: L\rightarrow C the morphism given by u:ℓ↦ℓ⊗1⊗1u:\ell \mapsto \ell \otimes 1\otimes 1. We can verify the relations

u(α)=v(ι(α))u(\alpha ) =v(\iota (\alpha )) and u(γβ)=v(γ)u(β)v(γ)−1u(^{\gamma}\beta ) =v(\gamma )u(\beta )v(\gamma )^{-1}

for α∈L∗,β∈L\alpha \in L^*,\beta \in L, and γ∈Γ\gamma \in \Gamma. Proposition 12 of VIII, p. 315 gives a homomorphism ff from the algebra A[E;L]\mathbf{A}[\mathscr{E}; L] to C. Since the algebra A[E;L]\mathbf{A}[\mathscr{E}; L] is simple, the homomorphism ff is injective. The algebras C and A[E;L]\mathbf{A}[\mathscr{E}; L] have the same dimension over K, so ff is an isomorphism.

Remark 1 00JN

If L is an étale algebra over K and G is the automorphism group of L, then it is not always true that the algebra A[E;L]\mathbf{A}[\mathscr{E}; L] is central simple (for example, we can take L=KnL = K^n and G=SnG =\mathfrak{S}_n).

Remark 2 00JO

We can calculate a 2-cocycle cc associated with an algebra A split by a finite Galois extension L with group G as follows. First, there exists a K-homomorphism φ:A→Mm\varphi : A\rightarrow \mathbf{M}_m(L), where [A:K]=m2[A : K] =m^2. For g∈Gg\in G, let φg\varphi^g be the homomorphism from A to Mm(L)\mathbf{M}_m(L) given by a↦φ(g−1ag)a\mapsto \varphi (g^{-1}ag). By the Skolem–Noether theorem (VIII, p. 256, Theorem 3), for every g∈Gg\in G, there exists an element ugu_g of GLm(L)\mathbf{G}\mathbf{L}_m(L) such that

φg(a)=ugφ(a)ug−1 \varphi^g(a) =u_g\varphi (a)u^{-1}_g

for a∈Aa\in A. We then set

c(g,g′)=ugug′ugg′−1 c(g, g') =u_gu_{g'}u^{-1}_{gg'}

We can also define an extension of G by L∗L^* using φ:\varphi : we consider the group Γ⊂GLm(L)\Gamma \subset \mathbf{G}\mathbf{L}_m(L) consisting of the γ\gamma for which there exists a g∈Gg\in G with

φg(a)=γφ(a)γ−1 \varphi^g(a) =\gamma \varphi (a)\gamma^{-1}

for every a∈Aa\in A. The class of this extension is the inverse image by Ψ of the class of A in Br(L/K/K).

Corollary 00JP

The mapping ΦL/K=Θ∘Ψ−1\Phi_{L/K}= \Theta \circ \Psi^{-1} defines a group isomorphism from Br(L/K/K) to H2(G,L∗)H^2(G,L^*).

Let K′K' be an extension of K and φ:K′→L\varphi : K'\rightarrow L a morphism of K-algebras. The set H of elements hh of G such that λ(h)∘φ=φ\lambda (h)\circ \varphi =\varphi is a subgroup of G, and the K′K'-algebra L endowed with the restriction of λ\lambda to H is a Galois algebra over K′K'.

Proposition 14 00JQ

The following diagram commutes:

Br(L/K/K)ΦL/K/^{\Phi_{L/K}}// H2(G,L∗)H_2(G,L_*)

rK′/Kr_{K'/K} ResHG^G_H

Br(L/K′/K')ΦL/K′/^{\Phi_{L/K'}}// H2(H,L∗)H_2(H,L_*) .

This follows from Propositions 7 of VIII, p. 299 and 13 of VIII, p. 317.

11. Index and Exponent

Theorem 6 00JR

Let K be a commutative field, and let A be a central simple A-algebra of finite degree over K. Let L be a separable extension of K of finite degree that is a splitting field for the algebra A. Then [L : K][A] is zero in Br(K).

There exists an extension M of L that is a Galois extension of K of finite degree (V, §10, No. 1, p. 57, Proposition 2). The class [A] of A in the Brauer group of K belongs to the subgroup Br(M/K/K). Let G be the Galois group of M over K, and let α\alpha be the image of [A] in H2(G,M∗)H^2(G,M^*) (VIII, p. 321, Corollary). Let H be the Galois group of M over L. Then H is a subgroup of index [L : K] in G (V, §10, No. 7, p. 68, Corollary 5). Since ResHG(α)=ΦM/L(A(L))^G_H(\alpha ) = \Phi_{M/L}(A_{(L)}) (Proposition 14), we have ResHG(α)=0^G_H(\alpha ) = 0. By Proposition 8 of VIII, p. 303, it follows that [L:K]α=0[L : K]\alpha = 0, and consequently [L : K][A] = 0.

Let K be a commutative field, and let A be a central simple algebra of finite degree over K. Then A is isomorphic to an algebra of the form Mn\mathbf{M}_n(D), where D is a field with center K, and [A] = [D] in Br(K). The reduced degree of D depends only on A. We call this reduced degree the index of A. The index of A divides the reduced degree of A. The exponent of A is the order of the class of A in the Brauer group of K.

Corollary 1 00JS

The exponent of a central simple algebra of finite degree over a commutative field divides its index.

It suffices to prove this for a field D of finite degree over its center K. Let L be a maximal commutative subfield of D that is a separable extension of K; we have [D:K]=[L:K]2[D : K] = [L : K]^2 (VIII, p. 265, Corollary 2, b) and c)). Then the extension L of K is a splitting field for the algebra D (VIII, p. 281, Proposition 5), and [L : K] coincides with the reduced degree of D. We then apply Theorem 6.

Corollary 2 00JT

Let K be a commutative field, and let A be a central simple algebra of finite degree over K. Let pp be a prime number. If pp divides the index of A, then it divides the exponent of A.

Let us suppose that the prime number pp does not divide the exponent of A and prove that it does not divide its index. It suffices to prove this result in the case when A is a field. Let L be a Galois extension of K of finite degree that splits A. Let G be the Galois group of L over K, and let H be a Sylow pp-subgroup of G (I, §6, No. 6, p. 78). We denote by L′=LHL'= L^H the subfield of L of invariants of H. The exponent of A(L′)A_{(L')} divides that of A and is therefore prime to pp. By Theorem 6, we have [L:L′][A(L′)]=0[L : L'][A_{(L')}] = 0 in the Brauer group of L′L'. It follows that [A(L′)]=0[A_{(L')}] = 0 and that the field L′L' splits A. We then apply Corollary 2 of VIII, p. 283; it follows that the index of A divides [L′:K][L': K] and is therefore not divisible by pp.

Corollary 3 00JU

Let pp be a prime number and let K be a perfect field of characteristic pp. Let A be a central simple algebra of finite degree over K. Then pp does not divide the index of A.

Let us prove that the Brauer group of K does not contain any element of order pp. Every Galois extension M of K of finite degree is a perfect field (V, §7, No. 1, p. 36, Proposition 2). Hence taking the pp-th power is an automorphism of the group M∗M^*. It follows that multiplication by pp is an automorphism of the group H2H^2(Gal(M/K/K),M∗,M^*) that is isomorphic to Br(M/K/K).

Consequently, the order of [A] is prime to pp and, by Corollary 2, its index is not divisible by pp.

Remark 00JV

By considering tensor products of quaternion algebras, it is possible to construct fields with center K, exponent 2, and arbitrarily large index (cf. VIII, p. 371, Exercises 7 and 8).

∗*Conversely, if K is a finite Galois extension of a pp-adic field or a field of formal power series over a finite field, then the exponent of a central simple algebra of finite degree over K is equal to its index (VIII, p. 332, Exercise 17, e)).∗*

Exercises

See the exercises for § 16.

Transcribed from the printed edition, 2023, Springer Nature. Not translated.