Let u∈L1(E;F). We have
∥u∥1= supv∈Lc(F;E)∣b(u,v)∣.
∥v∥⩽1
Since every Hilbert-Schmidt mapping is compact (cor. 2 of IV, p. 165), we have
∥u∥1⩽ supv∈∥Lv∥c(E;F)⩽1∣Tr(vu)∣= supv∈L∥vc∥(F;E)⩽1∣b(u,v)∣
by prop. 14.
Let v∈Lc(E;F) be of norm ⩽ 1. Since L2(E;F) is dense in Lc(E;F), there exists a sequence (vn)n∈N in L2(E;F) which converges to v in L(E;F). The sequence (b(u,vn))n∈N converges to b(u,v); since ∣b(u,vn)∣ = ∣Tr(vnu)∣ = ∣⟨vn∗∣u⟩∣⩽∥u∥1 (prop. 14) for every n∈N, it follows that ∣b(u,v)∣⩽∥u∥1.
Let us denote by θ the continuous linear mapping
θ:L1(E;F)→Lc(F;E)′
such that θ(u)(v)=b(u,v).
Bourbaki sets the proof directly after the statement, and the corpus does not separate them, so what is above is the statement and its proof together.
Nothing.
Nothing.