Bourbaki

Proposition 14

Tag 081R, permanent. Cite it as [Bourbaki, Tag 081R].

Topological Vector Spaces, chapter V, § 4. Some classes of operators in hilbertian spaces

— Let E1,...,EnE_1, ..., E_n be hilbertian spaces, E=E1⊗^2...⊗^2EnE = E_1 \hat{\otimes}_2 ... \hat{\otimes}_2 E_n, and uiu_i a continuous endomorphism of EiE_i for 1≤i≤n1 \leq i \leq n. If u1,...,unu_1, ..., u_n are positive, then so is u=u1⊗^2...⊗^2unu = u_1 \hat{\otimes}_2 ... \hat{\otimes}_2 u_n, and

Tr⁡(u)=∏i=1nTr⁡(ui). \operatorname{Tr}(u) = \prod_{i=1}^n \operatorname{Tr}(u_i).

If ui∈L1(Ei)u_i \in \mathscr{L}^1(E_i) for all 1≤i≤n1 \leq i \leq n, then u∈L1(E)u \in \mathscr{L}^1(E) and formula (31) is again valid in this case.

Proceeding by induction on nn, we immediately reduce to the case n=2n = 2.

For i=1,2i = 1, 2, we define a sesquilinear form Φi\Phi_i on EiE_i by the formula Φi(x,y)=⟨x∣ui(y)⟩\Phi_i(x, y) = \langle x | u_i(y) \rangle for x,yx, y in EiE_i. If u1u_1 and u2u_2 are positive, the forms Φ1\Phi_1 and Φ2\Phi_2 are hermitian and positive. By prop. 1 of V, p. 25 there exists a positive hermitian form Φ\Phi on the vector space E1⊗E2E_1 \otimes E_2 such that

Φ(x1⊗x2,y1⊗y2)=Φ1(x1,y1)⋅Φ2(x2,y2) \Phi(x_1 \otimes x_2, y_1 \otimes y_2) = \Phi_1(x_1, y_1) \cdot \Phi_2(x_2, y_2)

for x1,y1x_1, y_1 in E1E_1 and x2,y2x_2, y_2 in E2E_2. We verify immediately the relation Φ(z,t)=⟨z∣u(t)⟩\Phi(z, t) = \langle z | u(t) \rangle for zz and tt in E1⊗E2E_1 \otimes E_2. Since Φ\Phi is positive, we have ⟨z∣u(z)⟩≥0\langle z | u(z) \rangle \geq 0 for all zz in E1⊗E2E_1 \otimes E_2. Since uu is continuous and E1⊗E2E_1 \otimes E_2 is dense in the hilbertian space E=E1⊗^2E2E = E_1 \hat{\otimes}_2 E_2, we conclude that uu is a continuous and positive endomorphism of E.

Let (ei)i∈I(e_i)_{i \in I} be an orthonormal basis of E1E_1 and (fj)j∈J(f_j)_{j \in J} an orthonormal basis of E2E_2; then the family (ei⊗fj)i∈I,j∈J(e_i \otimes f_j)_{i \in I, j \in J} is an orthonormal basis of EE and we have

Tr⁡(u)=∑i∈I∑j∈J⟨ei⊗fj∣u(ei⊗fj)⟩=∑i∈I∑j∈J⟨ei∣u1(ei)⟩⋅⟨fj∣u2(fj)⟩=Tr⁡(u1)⋅Tr⁡(u2). \operatorname{Tr}(u) = \sum_{i \in I} \sum_{j \in J} \langle e_i \otimes f_j | u(e_i \otimes f_j) \rangle \\ = \sum_{i \in I} \sum_{j \in J} \langle e_i | u_1(e_i) \rangle \cdot \langle f_j | u_2(f_j) \rangle \\ = \operatorname{Tr}(u_1) \cdot \operatorname{Tr}(u_2) .

In particular, if u1u_1 and u2u_2 are positive endomorphisms with finite trace, then so is uu. By linearity, we deduce that uu belongs to L1(E)\mathscr{L}^1(E) when K=CK = \mathbf{C} and that the uiu_i belong to L1(Ei)\mathscr{L}^1(E_i) for i=1,2i = 1, 2; formula (31) extends to this case by linearity. Finally, the case when K=RK = \mathbf{R} and the ui∈L1(Ei)u_i \in \mathscr{L}^1(E_i) reduces to the complex case by extension of the scalars.

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

Cited by

Nothing.

Cites

Nothing.

Transcribed from the printed volume. Not translated.