Suppose that there exists a real hilbertian space structure on E such that H(x)=∥x∥2 for all x∈E. For Tr(Q/H) to be finite, it is necessary and sufficient that there exists a continuous and positive endomorphism u of E with finite trace, such that Q(x)=⟨x∣u(x)⟩ for all x∈E; this endomorphism u is unique, and we have
Tr(u)=Tr(Q/H)=i∈I∑Q(ei)
for every orthonormal bases (ei)i∈I of E.
Suppose that Tr(Q/H) is finite. For every x∈E of norm 1, we have H(x)=1, hence Q(x)≤Tr(Q/H). Therefore, Q(x)≤Tr(Q/H)⋅∥x∥2 for all x∈E, and
∣⟨x∣y⟩Q∣≤Q(x)1/2Q(y)1/2≤Tr(Q/H)⋅∥x∥⋅∥y∥
by the Cauchy-Schwarz inequality. Consequently, the bilinear form (x,y)↦⟨x∣y⟩Q on E×E is continuous. There exists (V, p. 16, cor. 2) a mapping u∈L(E) such that ⟨x∣y⟩Q=⟨x∣u(y)⟩. We have ⟨x∣y⟩Q=⟨y∣x⟩Q for x,y in E, hence u is hermitian; and ⟨x∣u(x)⟩=Q(x)≥0, hence u is positive.
Conversely, let u be a continuous and positive endomorphism of E such that Q(x)=⟨x∣u(x)⟩ for all x∈E. Then
⟨x∣u(y)⟩=21(Q(x+y)−Q(x)−Q(y))=⟨x∣y⟩Q,
which gives the uniqueness of u. By formula (24') (V, p. 50), we get
Tr(u)=x1,…,xmsupi=1∑m⟨xi∣u(xi)⟩=x1,…,xmsupi=1∑mQ(xi),
where (x1,…,xm) range over the set of all finite orthonormal sequences of elements of E. By the definition of Tr(Q/H), we get Tr(u)=Tr(Q/H). Finally, for every orthonormal basis (ei)i∈I of E, we have Tr(u)=∑i∈I⟨ei∣u(ei)⟩ by formula (25) of V, p. 50, hence Tr(u)=∑i∈IQ(ei).
Bourbaki puts the proof directly after the statement, so what is above is the statement and its proof together.
Nothing.
Nothing.