Bourbaki

Proposition 27

Tag 0BIZ, permanent. Cite it as [Bourbaki, Tag 0BIZ].

General Topology, chapter III, § 2. Subgroups, quotient groups, homomorphisms homogeneous spaces, product groups

Let N\mathbf{N} and L\mathbf{L} be two groups, e′e' and e′′e'' their respective identity elements. Suppose that we are given a homomorphism y→σyy \to \sigma_y of L\mathbf{L} into the group Γ\Gamma of automorphisms of N\mathbf{N}. Then:

1) On the product set S=N×LS = \mathbf{N} \times \mathbf{L}, the internal law of composition

(3)(3)
(x,y)(x′,y′)=(xσy(x′),yy′)(x, y)(x', y') = (x\sigma_y(x'), yy')
defines a group structure, for which j1:x→(x,e′′)j_1 : x \to (x, e'') is an isomorphism of N\mathbf{N} onto a normal subgroup of SS, j2:y→(e′,y)j_2 : y \to (e', y) is an isomorphism of L\mathbf{L} onto a subgroup of SS, and pr2:S→L\mathrm{pr}_2 : S \to \mathbf{L} is a surjective homomorphism whose kernel is j1(N)j_1(\mathbf{N}) and which is such that pr2∘j2\mathrm{pr}_2 \circ j_2 is the identity automorphism of L\mathbf{L}.

2) Let f:N→Gf : \mathbf{N} \to G, g:L→Gg : \mathbf{L} \to G be two homomorphisms into a group GG, such that

(4)(4)
f(σy(x))=g(y)f(x)g(y−1)f(\sigma_y(x)) = g(y)f(x)g(y^{-1})
for all x∈Nx \in \mathbf{N} and all y∈Ly \in \mathbf{L}. Then there is a unique homomorphism h:S→Gh : S \to G such that f=h∘j1f = h \circ j_1, and g=h∘j2g = h \circ j_2.

If x,x′,x′′x, x', x'' are elements of N\mathbf{N} and y,y′,y′′y, y', y'' are elements of L\mathbf{L}, then

((x,y)(x′,y′))(x′′,y′′)=(xσy(x′),yy′)(x′′,y′′)=(xσy(x′)σy′(x′′),yy′y′′) ((x, y)(x', y'))(x'', y'') = (x\sigma_y(x'), yy')(x'', y'') = (x\sigma_y(x')\sigma_{y'}(x''), yy'y'')
and
(x,y)((x′,y′)(x′′,y′′))=(x,y)(x′σy′(x′′),y′y′′)=(xσy(x′σy′(x′′)),yy′,y′′) (x, y)((x', y')(x'', y'')) = (x, y)(x'\sigma_{y'}(x''), y'y'') = (x\sigma_y(x'\sigma_{y'}(x'')), yy', y'')
and therefore the associativity of the law (3) follows from the facts that y→σyy \to \sigma_y is a homomorphism of L\mathbf{L} into Γ\Gamma and that σy\sigma_y is an automorphism of N\mathbf{N}. Clearly (e′,e′′)(e', e'') is the identity element of (3), and finally
(x,y)(σy−1(x−1),y−1)=(σy−1(x−1),y−1)(x,y)=(e′,e′′)(x, y)(\sigma_{y^{-1}}(x^{-1}), y^{-1}) = (\sigma_{y^{-1}}(x^{-1}), y^{-1})(x, y) = (e', e'')
so that (x,y)(x, y) has an inverse in SS. The other assertions of 1) are clear. On the other hand, since (x,y)=(x,e′′)(e′,y)(x, y) = (x, e'') (e', y), a homomorphism hh which satisfies the conditions of 2) must necessarily satisfy
h(x,y)=f(x)g(y),h(x, y) = f(x)g(y),
hence is unique if it exists; moreover it is immediate from (4) that
f(xσy(x′))g(yy′)=f(x)g(y)f(x′)g(y−1)g(y)g(y′)=f(x)g(y)f(x′)g(y′)f(x\sigma_y(x'))g(yy') = f(x)g(y)f(x')g(y^{-1})g(y)g(y') = f(x)g(y)f(x')g(y')

which shows that (x,y)→f(x)g(y)(x, y) \to f(x)g(y) is indeed a homomorphism of S into G satisfying the conditions of 2).

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

Cited by

Cites

Nothing.

Transcribed from the printed volume. Not translated.