Suppose that K=R. Let u be a partial operator with dense domain from E into F.
a) The adjoint of u(C) is (u∗)(C);
b) Suppose that E = F; the partial operator u is symmetric (resp. self-adjoint) if and only if the partial operator u(C) is symmetric (resp. self-adjoint).
Let us prove a). Let y∈F(C) and write y=y1+iy2 with y1,y2∈F. For all (x1,x2)∈E×E, we have
⟨u(C)(x1+ix2)∣y⟩=⟨u(x1)∣y1⟩+i⟨u(x1)∣y2⟩
−i⟨u(x2)∣y1⟩+⟨u(x2)∣y2⟩
If y∈ dom(u∗)(C), it follows that y∈ dom((u(C))∗) and that u(C)∗(y)=(u(C))∗(y), hence u(C)∗⊂(u(C))∗.
Conversely, suppose that y belongs to dom((u(C))∗). Taking x2=0 (resp. x1=0) in the above formula, we verify that y1∈ dom(u∗) (resp. that y2∈ dom(u∗)), whence y∈ dom(u∗)(C).
Assertion a) implies that u(C) is symmetric (resp. self-adjoint) if u is.
Conversely, suppose that u(C) is symmetric. The relation ⟨u(x)∣y⟩=⟨x∣u(y)⟩ for all (x,y)∈ dom(u(C))× dom(u(C)) implies that u is symmetric on taking x and y in the subspace dom(u) of dom(u(C)). If u(C) is self-adjoint, what precedes proves that u is symmetric; since dom(u∗) = dom(u(C)∗)∩F, assertion a) implies that dom(u∗) = dom(u(C))∩F= dom(u), therefore u is self-adjoint.
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.