Section II.5

The Diagonalization of Compact Self-Adjoint Operators

[5/0%]
LaTeX
Problem II.5.1

Prove Corollary 5.4: If TT is a compact self-adjoint operator, then there is a sequence {μn}\left\{ \mu_{n}\right\} of real numbers and an orthonormal basis {en}\left\{ e_{n}\right\} for (ker⁡T)⊥(\ker T)^{\perp } such that for all hh,

Th=∑n=1∞μn⟨h,en⟩en. Th = \sum _{n=1}^{\infty } \mu _{n} \left\langle h, e_{n} \right\rangle e_{n}.

Note that there may be repetitions in the sequence {μn}\left\{ \mu_{n}\right\}. How many repetitions?

?
Problem II.5.2

Prove Corollary 5.5: If T∈B0(H)T \in \mathcal{B}_{0}(\mathcal{H}), T=T∗T=T^{*}, and ker⁡T=(0)\ker T = (0), then H\mathcal{H} is separable.

?
Problem II.5.3

Let KK and kk be as in Proposition 4.7 and suppose that k(x,y)=k(y,x)‾k(x,y) = \overline{k(y,x)}. Show that KK is self-adjoint and if {μn}\left\{ \mu_{n}\right\} are the eigenvalues of KK, each repeated dim⁡ker⁡(K−μn)\dim \ker (K-\mu_{n}) times, then ∑1∞∣μn∣2<∞\sum_{1}^{\infty } \left|\mu_{n}\right|^{2} < \infty.

Proposition 4.7 (context): If (X,Ω,μ)(X, \Omega , \mu ) is a measure space and k∈L2(X×X,Ω×Ω,μ×μ)k \in L^{2}(X \times X, \Omega \times \Omega , \mu \times \mu ), then

(Kf)(x)=∫k(x,y)f(y) dμ(y) (Kf)(x) = \int k(x,y) f(y) \, d\mu (y)

is a compact operator and ∥K∥≤∥k∥2\left\| K\right\| \leq \left\| k\right\|_{2}.

?
Problem II.5.4

If TT is a compact self-adjoint operator and {en}\left\{ e_{n}\right\} and {μn}\left\{ \mu_{n}\right\} are as in (5.4) and if hh is a given vector in H\mathcal{H}, show that there is a vector ff in H\mathcal{H} such that Tf=hTf=h if and only if h⊥ker⁡Th \perp \ker T and ∑nμn−2∣⟨h,en⟩∣2<∞\sum_{n} \mu_{n}^{-2} \left| \left\langle h, e_{n} \right\rangle \right|^{2} < \infty. Find the form of the general vector ff such that Tf=hTf=h.

(5.4) (context): There is a sequence {μn}\left\{ \mu_{n}\right\} of real numbers and an orthonormal basis {en}\left\{ e_{n}\right\} for (ker⁡T)⊥(\ker T)^{\perp } such that for all hh, Th=∑n=1∞μn⟨h,en⟩enTh = \sum_{n=1}^{\infty } \mu_{n} \left\langle h, e_{n} \right\rangle e_{n}.

?
Problem II.5.5

Let TT, {μn}\left\{ \mu_{n}\right\}, and {en}\left\{ e_{n}\right\} be as in (5.4). If λ≠0\lambda \neq 0 and λ≠μn\lambda \neq \mu_{n} for any μn\mu_{n}, then for every hh in H\mathcal{H} there is a unique ff in H\mathcal{H} such that (λ−T)f=h(\lambda - T)f = h. Moreover,

f=λ−1[h+∑n=1∞λn(λ−λn)−1⟨h,en⟩en]. f = \lambda ^{-1} \left[h + \sum _{n=1}^{\infty } \lambda _{n} (\lambda - \lambda _{n})^{-1} \left\langle h, e_{n} \right\rangle e_{n}\right].

Interpret this when TT is an integral operator.

(5.4) (context): There is a sequence {μn}\left\{ \mu_{n}\right\} of real numbers and an orthonormal basis {en}\left\{ e_{n}\right\} for (ker⁡T)⊥(\ker T)^{\perp } such that for all hh, Th=∑n=1∞μn⟨h,en⟩enTh = \sum_{n=1}^{\infty } \mu_{n} \left\langle h, e_{n} \right\rangle e_{n}.

?