In this example, K always denotes one of the fields R,C,H. We write δ=dimRK (thus δ=1,2 or 4). A hermitian form Φ on the right vector space Kn may be written
Φ(x,y)=Φ(x1,…,xn,y1,…,yn)=i,j=1∑nxihijyj
with hij=hji for all i and j. We denote by H the vector space over R formed by the hermitian matrices of Mn(K). The mapping (hij)↦Φ is an isomorphism of H onto the vector space of hermitian forms on Kn, by means of which we shall identify these two spaces. Let H+∗⊂H be the set of nondegenerate positive hermitian forms on Kn. The set H+∗ is convex in H; for, if Φ1,Φ2 are in H+∗ and if λ,μ are two numbers >0 such that λ+μ=1, it is clear that λΦ1+μΦ2 is a positive hermitian form; on the other hand, if (λΦ1+μΦ2)(x,x)=0, then Φ1(x,x)=Φ2(x,x)=0, therefore x=0, so that λΦ1+μΦ2 is nondegenerate. Let us now show that H+∗ is an open subset of H. Let S be the set of x=(x1,…,xn)∈Kn such that x1x1+⋯+xnxn=1; this is a compact subset of Kn; if Φ∈H+∗, the function x↦Φ(x,x) is continuous and >0 on S, hence its infimum is >0; if Φ′∈H is sufficiently near Φ, it follows that Φ′(x,x)>0 for all x∈S, so that Φ′ is positive and nondegenerate.
The general linear group GL(n,K) operates continuously on the right in H by (X,Φ)↦Φ∘X, that is, by (X,H)↦tX⋅H⋅X, where H denotes the hermitian matrix corresponding to Φ. It is clear that H+∗ is stable under GL(n,K). More precisely, by Alg., Ch. IX, §6, No. 1, Cor. 1 of Th. 1,1H+∗ is the orbit under GL(n,K) of the form ∑i=1nxiyi corresponding to the identity matrix In. The stabilizer of this form is U(n,K). By Lemma 2 of App. I, H+∗ may be identified, as a topological homogeneous space, with GL(n,K)/U(n,K).
For every X∈GL(n,K), let X be the automorphism H↦tX⋅H⋅X of the real vector space H. If μ denotes the Haar measure of the additive group H, one has X−1(μ)=∣detX∣⋅μ (§ 1, No. 10, Cor. 1 of Prop. 15). Let us show that
∣detX∣=∣N(X)∣λ,
where N denotes the norm in Mn(K)regarded as anR-algebra, and where λ=1−δnδ−2. It suffices to verify (15) for X running over a system of
(1) Cf. TVS, V, §2, No. 4, Cor. 1 of Th. 2.
generators of GL(n,K), hence (A, II, §10, No. 13, Cor. 2 of Prop. 14) for X of the following types:
a) X is the matrix of a mapping of the form
(x1,…,xn)↦(xσ(1),…,xσ(n)),
where σ∈Sn. In this case, a power of X is equal to 1, therefore ∣detX∣=∣N(X)∣=1.
b) X is the matrix of a mapping of the form
(x1,…,xn)↦(ax1,x2,…,xn).
Then, if (hij)∈H, one has X((hij))=(hij′) with h11′=ah11a=∣a∣2h11, h1i′=ah1i for i>1, hij′=hij for i>1,j>1; therefore
∣detX∣=∣a∣2∣a∣δ(n−1)=∣a∣2+δ(n−1).
On the other hand, if Y=(yij)∈Mn(K), then XY=(yij′) with y1j′=ay1j and yij′=yij for i>1; therefore ∣N(X)∣=∣a∣δn. The formula (15) is again verified.
c) X is the matrix of a mapping of the form
(x1,…,xn)↦(x1+bx2,x2,…,xn).
Then X((hij))=(hij′) with h11′=h11, h12′=h12+h11b, h1i′=h1i for i>2, h22′=h22+bh12+h12b+bh11b, h2i′=h2i+bh1i for i>2, hij′=hij for i>2,j>2. Taking into account Lemma 6, one sees that ∣detX∣=1. One verifies similarly that ∣N(X)∣=1, which completes the proof of formula (15).
This established, the measure ∣N(H)∣−λ/2dμ(H) on H is invariant under GL(n,K), since
If H∈H+∗, then H=X(In)=tXX for some X∈GL(n,K), therefore N(H)=N(X)N(X)>0. Consequently, on H+∗, the unique (up to a constant factor) measure invariant under GL(n,K) (cf. §2, No. 6, Th. 3) is the measure
dγ(H)=N(H)−λ/2dμ(H).
In particular,
dγ(H)=(detH)−(n+1)/2dμ(H)when K=R,
dγ(H)=(detH)−ndμ(H)when K=C.
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.