Chapter 12

Fundamentals of spectral theory of bounded operators in Banach spaces

[111/32%]
Show
LaTeX
Problem 12.1

Let XX be a Banach space, A∈B(X)A \in \mathcal{B}(X). Using problem 9.39, prove that ρ(A)\rho (A) is an open set (hence σ(A)\sigma (A) is a closed set) in C\mathbb {C}.

?
Problem 12.2

Using problem 9.39, prove that the spectrum of a bounded operator AA lies inside the disk {λ∈C:∣λ∣⩽∥A∥}\left\{ \lambda \in \mathbb {C}:\left|\lambda \right| \leqslant \left\| A\right\| \right\}. Prove that for ∣λ∣>∥A∥\left|\lambda \right| > \left\| A\right\| the resolvent is given by the C. Neumann series:

Rλ(A)=−1λ∑n=0∞Anλn R_{\lambda }(A) = -\frac{1}{\lambda } \sum _{n = 0}^{\infty } \frac{A^{n}}{\lambda ^{n}}
?
Problem 12.3

Prove Theorem 12.1.

Theorem 12.1: Let XX be a Banach space. If A∈B(X)A \in \mathcal{B}(X), then the limit lim⁡n→∞∥An∥1/n\lim_{n \rightarrow \infty }\left\| A^{n}\right\|^{1 / n} exists and

r(A)=lim⁡n→∞∥An∥1/n, r(A) = \lim _{n \rightarrow \infty }\left\| A^{n}\right\| ^{1 / n},

where r(A)=max⁡λ∈σ(A)∣λ∣r(A) = \max_{\lambda \in \sigma (A)}\left|\lambda \right| is the spectral radius of the operator AA (Definition 12.2).

?
Problem 12.4

© Let λ,μ∈ρ(A)\lambda , \mu \in \rho (A). Prove the Hilbert resolvent identity:

Rλ(A)−Rμ(A)=(λ−μ)Rλ(A)Rμ(A). R_{\lambda }(A)-R_{\mu }(A) = (\lambda -\mu ) R_{\lambda }(A) R_{\mu }(A).

Using this identity, prove that

Rλ(A)Rμ(A)=Rμ(A)Rλ(A) R_{\lambda }(A) R_{\mu }(A) = R_{\mu }(A) R_{\lambda }(A)
?
Problem 12.5

Let λ0,λ∈ρ(A)\lambda_{0}, \lambda \in \rho (A). Prove that if ∣λ−λ0∣<1∥Rλ0(A)∥\left|\lambda -\lambda_{0}\right| < \frac{1}{\left\| R_{\lambda_{0}}(A)\right\| }, then

∥Rλ(A)−Rλ0(A)∥⩽∣λ−λ0∣⋅∥Rλ02(A)∥1−∣λ−λ0∣⋅∥Rλ0(A)∥ \left\| R_{\lambda }(A)-R_{\lambda _{0}}(A)\right\| \leqslant \frac{\left|\lambda -\lambda _{0}\right| \cdot \left\| R_{\lambda _{0}}^{2}(A)\right\| }{1-\left|\lambda -\lambda _{0}\right| \cdot \left\| R_{\lambda _{0}}(A)\right\| }
?
Problem 12.6

Prove that Rλ(A)R_{\lambda }(A) is

?
(a)

a continuous;

(b)

an analytic function of the variable λ\lambda on the domain ρ(A)∪{∞}\rho (A) \cup \left\{ \infty \right\} with values in B(X)\mathcal{B}(X). Prove that if λ0∈ρ(A)\lambda_{0} \in \rho (A), then for ∣λ−λ0∣<1∥Rλ0(A)∥\left|\lambda -\lambda_{0}\right| < \frac{1}{\left\| R_{\lambda_{0}}(A)\right\| } the resolvent admits the representation

Rλ(A)=∑k=0∞(λ−λ0)k(Rλ0(A))k+1 R_{\lambda }(A) = \sum _{k = 0}^{\infty }\left(\lambda -\lambda _{0}\right)^{k}\left(R_{\lambda _{0}}(A)\right)^{k+1}

and for ∣λ∣>r(A)\left|\lambda \right| > r(A)

Rλ(A)=−1λ∑n=1∞Anλn R_{\lambda }(A) = -\frac{1}{\lambda } \sum _{n = 1}^{\infty } \frac{A^{n}}{\lambda ^{n}}
Problem 12.7

Let XX be a Banach space, A∈B(X)(X≠{0})A \in \mathcal{B}(X)(X \neq \left\{ 0\right\} ). Prove that the spectrum of AA is nonempty.

?
Problem 12.8

Let XX be a Banach space, A∈B(X),λ∈σ(A)A \in \mathcal{B}(X), \lambda \in \sigma (A), and suppose there exists c>0c > 0 such that ∥Ax−λx∥>c∥x∥\left\| A x-\lambda x\right\| > c\left\| x\right\| for all x∈Xx \in X. Using problem 9.49, prove that λ∈σr(A)\lambda \in \sigma_{\mathrm{r}}(A).

?
Problem 12.9

Let XX be a Banach space, A∈B(X)A \in \mathcal{B}(X). Prove that if for a point λ∈C\lambda \in \mathbb {C} there exists a Weyl sequence, then λ∈σ(A)\lambda \in \sigma (A). Prove that if λ∈σp(A)∪σc(A)\lambda \in \sigma_{\mathrm{p}}(A) \cup \sigma_{\mathrm{c}}(A), then a Weyl sequence exists for it (see also problem 12.23).

?
Problem 12.10

Let XX be a Banach space, A∈B(X)A \in \mathcal{B}(X) an invertible operator. Prove that λ∈σ(A)\lambda \in \sigma (A) if and only if 1/λ∈σ(A−1)1 / \lambda \in \sigma \left(A^{-1}\right). Prove this statement with σ\sigma replaced by σp,σc\sigma_{\mathrm{p}}, \sigma_{\mathrm{c}} and σr\sigma_{\mathrm{r}}. Prove that the eigenvectors of the operators AA and A−1A^{-1} corresponding to the eigenvalues λ\lambda and 1/λ1 / \lambda coincide.

?
Problem 12.11

Prove that the spectrum of an invertible isometric operator acting in a normed space lies on the unit circle. Prove that the spectrum of a unitary operator acting in a Hilbert space lies on the unit circle. Find the resolvent of these operators for λ∈C\lambda \in \mathbb {C}, ∣λ∣≠1\left|\lambda \right| \neq 1.

?
Problem 12.12

Let X=X0⊕X1X = X_{0} \oplus X_{1} be a Banach space, represented as a direct sum of two of its closed subspaces, and let P∈B(X)P \in \mathcal{B}(X) be the projector onto X0X_{0} along X1X_{1}. Find and classify its spectrum. Find its resolvent.

?
Problem 12.13

(R. S. Phillips, 1955). Let XX be a Banach space, A∈B(X)A \in \mathcal{B}(X). Prove that σ(A)=σ(A′)\sigma (A) = \sigma \left(A^{\prime }\right).

?
Problem 12.14

(R. S. Phillips, 1955). Let HH be a Hilbert space, A∈B(H)A \in \mathcal{B}(H). Prove that 1 σ(A)=σ(A∗)‾\sigma (A) = \overline{\sigma \left(A^{*}\right)}.

Footnotes

  1. The bar over the set here does not denote the closure of the set, but complex conjugation. ↩

?
Problem 12.15

Let XX be a Banach space, A∈B(X)A \in \mathcal{B}(X). Prove that if λ∈σr(A)\lambda \in \sigma_{\mathrm{r}}(A), then λ∈σp(A′)\lambda \in \sigma_{\mathrm{p}}\left(A^{\prime }\right); if λ∈σp(A)\lambda \in \sigma_{\mathrm{p}}(A), then λ∈σp(A′)∪σr(A′)\lambda \in \sigma_{\mathrm{p}}\left(A^{\prime }\right) \cup \sigma_{\mathrm{r}}\left(A^{\prime }\right); if λ∈σc(A)\lambda \in \sigma_{\mathrm{c}}(A), then λ∈σc(A′)∪σr(A′)\lambda \in \sigma_{\mathrm{c}}\left(A^{\prime }\right) \cup \sigma_{\mathrm{r}}\left(A^{\prime }\right).

?
Problem 12.16

Let XX be a reflexive Banach space, A∈B(X)A \in \mathcal{B}(X). Prove that σc(A)=σc(A′),σr(A′)⊂σp(A)\sigma_{\mathrm{c}}(A) = \sigma_{\mathrm{c}}\left(A^{\prime }\right), \sigma_{\mathrm{r}}\left(A^{\prime }\right) \subset \sigma_{\mathrm{p}}(A).

?
Problem 12.17

Let HH be a Hilbert space, A∈B(H)A \in \mathcal{B}(H). Prove that if λ∈σr(A)\lambda \in \sigma_{\mathrm{r}}(A), then λˉ∈σp(A∗)\bar{\lambda } \in \sigma_{\mathrm{p}}\left(A^{*}\right), if λ∈σp(A)\lambda \in \sigma_{\mathrm{p}}(A), then λˉ∈σp(A∗)∪σr(A∗)\bar{\lambda } \in \sigma_{\mathrm{p}}\left(A^{*}\right) \cup \sigma_{\mathrm{r}}\left(A^{*}\right), if λ∈σc(A)\lambda \in \sigma_{\mathrm{c}}(A), then λˉ∈σc(A∗)\bar{\lambda } \in \sigma_{\mathrm{c}}\left(A^{*}\right).

?
Problem 12.18

Find the spectra of the following operators (with classification) and compute their resolvents:

?
(a)

the multiplication operator by a bounded sequence in lp,p∈[1,∞]l_{p}, p \in [1, \infty ], and in c0c_{0}:

Ax=(λ1x1,λ2x2,…,λnxn,…),λ=(λ1,λ2,…,λn,…)∈l∞ A x = \left(\lambda _{1} x_{1}, \lambda _{2} x_{2}, \ldots , \lambda _{n} x_{n}, \ldots \right), \quad \lambda = \left(\lambda _{1}, \lambda _{2}, \ldots , \lambda _{n}, \ldots \right) \in l_{\infty }
(b)

the operators of right and left shift in the spaces l1l_{1} and l∞l_{\infty };

(c)

the multiplication operator (Ax)(t)=a(t)x(t)(A x)(t) = a(t) x(t) in C[0,1],a∈C[0,1]C[0,1], a \in C[0,1];

(d)

the multiplication operator (Ax)(t)=a(t)x(t)(A x)(t) = a(t) x(t) in Lp[0,1],p∈[1,∞]L_{p}[0,1], p \in [1, \infty ], a∈L∞[0,1]a \in L_{\infty }[0,1];

(e)

the operator (Ax)(t)=a(t)x(t)(A x)(t) = a(t) x(t) in C0(R)C_{0}(\mathbb {R}), where a∈C0(R)a \in C_{0}(\mathbb {R});

(f)

the shift operator in Lp(R),p∈[1,∞]:(Ax)(t)=x(t−a),a∈RL_{p}(\mathbb {R}), p \in [1, \infty ]:(A x)(t) = x(t-a), a \in \mathbb {R}.

Problem 12.19

Find the spectra of the operators AA and A∗A^{*}, if AA acts in l2(Z)l_{2}(\mathbb {Z}) as follows: (Ax)n=xn+1(A x)_{n} = x_{n+1} for n⩾0,(Ax)n=0n \geqslant 0,(A x)_{n} = 0 for n<0n < 0.

?
Problem 12.20

Find the spectrum, spectral radius, and resolvent of the operator (Ax)(t)=∫0tx(s)ds(A x)(t) = \int_{0}^{t} x(s) d s in the spaces:

?
(a)

Lp[0,1],p∈[1,∞]L_{p}[0,1], p \in [1, \infty ];

(b)

C[0,1]C[0,1].

Problem 12.21

Find the spectrum, spectral radius, and resolvent of the operator A:lp→lp,p∈[1,∞)A: l_{p} \rightarrow l_{p}, p \in [1, \infty ):

?
(a)

Ax=(0,x1,0,x3,…,0,x2n−1,0,…)A x = \left(0, x_{1}, 0, x_{3}, \ldots , 0, x_{2 n-1}, 0, \ldots \right);

(b)

Ax=(0,x1,0,x2,…,0,xn,0,…)A x = \left(0, x_{1}, 0, x_{2}, \ldots , 0, x_{n}, 0, \ldots \right);

(c)

Ax=(0,x1,x2,0,x4,x5,0…,0,x3n+1,x3n+2,0,…)A x = \left(0, x_{1}, x_{2}, 0, x_{4}, x_{5}, 0 \ldots , 0, x_{3 n+1}, x_{3 n+2}, 0, \ldots \right);

(d)

Ax=(x2,x1,x4,x3,…,x2n+2,x2n+1,…)A x = \left(x_{2}, x_{1}, x_{4}, x_{3}, \ldots , x_{2 n+2}, x_{2 n+1}, \ldots \right);

(e)

Ax=(x2,x3,x1,x5,x6,x4,…,x3n+2,x3n+3,x3n+1,…)A x = \left(x_{2}, x_{3}, x_{1}, x_{5}, x_{6}, x_{4}, \ldots , x_{3 n+2}, x_{3 n+3}, x_{3 n+1}, \ldots \right);

(f)

Ax=(x1,x2,…,xn,0,0,…)A x = \left(x_{1}, x_{2}, \ldots , x_{n}, 0,0, \ldots \right).

Problem 12.22

Find the spectrum and resolvent of the following operators:

?
(a)

(Ax)(t)=x(−t)(A x)(t) = x(-t) in C[−1,1]C[-1,1];

(b)

(Ax)(t)=−x(−t)(A x)(t) = -x(-t) in C[−1,1]C[-1,1];

(c)

(Ax)(t)=x(1−t)(A x)(t) = x(1-t) in C[0,1]C[0,1];

(d)

(Ax)(t)=x(1−t)(A x)(t) = x(1-t) in Lp[0,1],p∈[1,∞]L_{p}[0,1], p \in [1, \infty ];

(e)

(Ax)(t)=∫−ππcos⁡(t+s)x(s)ds(A x)(t) = \int_{-\pi }^{\pi } \cos (t+s) x(s) d s in C[−π,π]C[-\pi , \pi ];

(f)

(Ax)(t)=tx(1)−x(0)(A x)(t) = t x(1)-x(0) in C[0,1]C[0,1].

Problem 12.23

Give examples of bounded operators in Banach spaces showing that for a number λ∈C\lambda \in \mathbb {C} lying in the residual spectrum, a Weyl sequence may or may not exist (cf. problem 12.9).

?
Problem 12.24

Using problems 8.75, 8.77, prove that normal operators have no residual spectrum.

?
Problem 12.25

(H. Weyl, 1910). Let AA be a normal operator. Prove that λ∈σ(A)\lambda \in \sigma (A) if and only if a Weyl sequence exists.

?
Problem 12.26

Prove that the spectrum of a self-adjoint operator is real.

?
Problem 12.27

Prove that if the spectrum of a normal operator is real, then this operator is self-adjoint.

?
Problem 12.28

Prove that if the spectrum of a normal operator lies on the circle {z∈C:∣z∣=1}\left\{ z \in \mathbb {C}:\left|z\right| = 1\right\}, then this operator is unitary.

?
Problem 12.29

Let AA be a self-adjoint operator in a Hilbert space HH. Prove that σess (A)\σp(A)=σc(A)\sigma_{\text{ess }}(A) \backslash \sigma_{\mathrm{p}}(A) = \sigma_{\mathrm{c}}(A).

?
Problem 12.30

Let AA be a bounded self-adjoint operator in a Hilbert space HH. Prove that if there exists an orthonormal Weyl sequence for the operator A−λIA-\lambda I, then λ∈σess (A)\lambda \in \sigma_{\text{ess }}(A).

?
Problem 12.31

Find the essential spectrum for the following operators:

?
(a)

the multiplication operator by a bounded sequence in lp,p∈[1,∞]l_{p}, p \in [1, \infty ], and in c0c_{0}:

Ax=(λ1x1,λ2x2,…,λnxn,…),λ=(λ1,λ2,…,λn,…)∈l∞; A x = \left(\lambda _{1} x_{1}, \lambda _{2} x_{2}, \ldots , \lambda _{n} x_{n}, \ldots \right), \quad \lambda = \left(\lambda _{1}, \lambda _{2}, \ldots , \lambda _{n}, \ldots \right) \in l_{\infty } ;
(b)

the operators of right shift TrT_{r} and left shift TlT_{l} in the space lpl_{p}, p∈[1,∞]p \in [1, \infty ];

(c)

the multiplication operator (Ax)(t)=a(t)x(t)(A x)(t) = a(t) x(t) in C[0,1],a∈C[0,1]C[0,1], a \in C[0,1];

(d)

the multiplication operator (Ax)(t)=a(t)x(t)(A x)(t) = a(t) x(t) in Lp[0,1],p∈[1,∞]L_{p}[0,1], p \in [1, \infty ], a∈L∞[0,1]a \in L_{\infty }[0,1];

(e)

the shift operator in Lp(R),p∈[1,∞]:(Ax)(t)=x(t−a),a∈RL_{p}(\mathbb {R}), p \in [1, \infty ]:(A x)(t) = x(t-a), a \in \mathbb {R};

(f)

the integration operator (Ax)(t)=∫0tx(s)ds(A x)(t) = \int_{0}^{t} x(s) d s in the spaces

(g)

the operator (Ax)(t)=x(−t)(A x)(t) = x(-t) in C[−1,1]C[-1,1];

(h)

the operator (Ax)(t)=x(1−t)(A x)(t) = x(1-t) in C[0,1]C[0,1];

(i)

the operator (Ax)(t)=∫−ππcos⁡(t+s)x(s)ds(A x)(t) = \int_{-\pi }^{\pi } \cos (t+s) x(s) d s in C[−π,π]C[-\pi , \pi ];

(j)

the operator (Ax)(t)=tx(1)−x(0)(A x)(t) = t x(1)-x(0) in C[0,1]C[0,1].

Problem 12.32

Let HH be a Hilbert space, A∈B(H)A \in \mathcal{B}(H) and A=A∗A = A^{*}. Define the numbers m:=inf⁡∥x∥=1(Ax,x)m: = \inf_{\left\| x\right\| = 1}(A x, x) and M:=sup⁡∥x∥=1(Ax,x)M: = \sup_{\left\| x\right\| = 1}(A x, x). Prove that σ(A)⊂[m,M]\sigma (A) \subset [m, M] and both points mm and MM lie in the spectrum.

?
Problem 12.33

Using problem 8.52, prove that for any bounded operator AA in a Hilbert space σ(A∗A)⊂[0,∥A∥2]\sigma \left(A^{*} A\right) \subset \left[0,\left\| A\right\|^{2}\right], and moreover the point ∥A∥2\left\| A\right\|^{2} lies in the spectrum.

?
Problem 12.34

Prove that eigenvectors corresponding to distinct eigenvalues are orthogonal for:

?
(a)

a self-adjoint operator;

(b)

a unitary operator.

Problem 12.35

Let HH be a Hilbert space, AA a normal operator. Prove that r(A)=∥A∥r(A) = \left\| A\right\| (see problem 8.78).

?
Problem 12.36

Let X,YX, Y be Banach spaces, A∈B(X),B∈B(Y)A \in \mathcal{B}(X), B \in \mathcal{B}(Y) similar operators (see Definition 9.4). Prove that σp(A)=σp(B),σc(A)=σc(B)\sigma_{\mathrm{p}}(A) = \sigma_{\mathrm{p}}(B), \sigma_{\mathrm{c}}(A) = \sigma_{\mathrm{c}}(B) and σr(A)=σr(B)\sigma_{\mathrm{r}}(A) = \sigma_{\mathrm{r}}(B) (whence it follows that σ(A)=σ(B))\sigma (A) = \sigma (B)).

Definition 9.4: Let XX and YY be normed spaces, A∈B(X)A \in \mathcal{B}(X), B∈B(Y)B \in \mathcal{B}(Y). The operators AA and BB are called similar if there exists a bounded operator SS with bounded inverse (that is, S∈IB(Y,X)S \in \mathcal{I B}(Y, X)) such that AS=SBA S = S B (notation A∼BA \sim B).

?
Problem 12.37

Give an example of two operators AA and BB in a Banach space that are not similar but have the same spectra: σ(A)=σ(B),σp(A)=σp(B),σc(A)=σc(B),σr(A)=σr(B)\sigma (A) = \sigma (B), \sigma_{\mathrm{p}}(A) = \sigma_{\mathrm{p}}(B), \sigma_{\mathrm{c}}(A) = \sigma_{\mathrm{c}}(B), \sigma_{\mathrm{r}}(A) = \sigma_{\mathrm{r}}(B) and σess(A)=σess(B)\sigma_{\mathrm{ess}}(A) = \sigma_{\mathrm{ess}}(B).

?
Problem 12.38

Using problem 12.36, find the spectrum of the following operators:

?
(a)

(Ax)n=βxn−1+αxn+βxn+1,A∈B(l2(Z))(A x)_{n} = \beta x_{n-1}+\alpha x_{n}+\beta x_{n+1}, A \in \mathcal{B}\left(l_{2}(\mathbb {Z})\right);

(b)

(Ax)n=xn−1−xn+1,A∈B(l2(Z))(A x)_{n} = x_{n-1}-x_{n+1}, A \in \mathcal{B}\left(l_{2}(\mathbb {Z})\right);

(c)

A∈B(l2)A \in \mathcal{B}\left(l_{2}\right),

(Ax)n={αx1+βx2,n=1;βxn−1+αxn+βxn+1,n⩾2. (A x)_{n} = \begin{cases} \alpha x_{1}+\beta x_{2}, & n = 1 ; \\ \beta x_{n-1}+\alpha x_{n}+\beta x_{n+1}, & n \geqslant 2. \end{cases}

Classify the spectrum, determining σp,σc,σr\sigma_{\mathrm{p}}, \sigma_{\mathrm{c}}, \sigma_{\mathrm{r}} and σess \sigma_{\text{ess }}.

Problem 12.39

The operator AA in the space L2[0,1]L_{2}[0,1] is given by the formula

(Ax)(t)={x(2t), if t∈[0,1/2];x(2t−1), if t∈[1/2,1]. (A x)(t) = \begin{cases} x(2 t), & \text{ if } t \in [0,1 / 2] ; \\ x(2 t-1), & \text{ if } t \in [1 / 2,1]. \end{cases}

Classify the spectrum, determining σp,σc,σr\sigma_{\mathrm{p}}, \sigma_{\mathrm{c}}, \sigma_{\mathrm{r}} and σess\sigma_{\mathrm{ess}}.

?
Problem 12.40

Let XX be a Banach space, A,B∈B(X)A, B \in \mathcal{B}(X). Prove that σ(AB)∪{0}=σ(BA)∪{0}\sigma (A B) \cup \left\{ 0\right\} = \sigma (B A) \cup \left\{ 0\right\}. Prove this equality with σ\sigma replaced by σp,σc\sigma_{\mathrm{p}}, \sigma_{\mathrm{c}} and σr\sigma_{\mathrm{r}}. Prove that if at least one of the operators AA or BB is invertible, then the operators ABA B and BAB A are similar and, hence, σ(AB)=σ(BA)\sigma (A B) = \sigma (B A).

?
Problem 12.41

Let XX be a Banach space, and suppose for operators A∈L(X)A \in \mathcal{L}(X) and B∈L(X)B \in \mathcal{L}(X) the relation AB−BA=IA B-B A = I holds. Using problem 12.40, prove that one of the operators is unbounded.

?
Problem 12.42

Prove that for every compact set Ω⊂C\Omega \subset \mathbb {C} there exists a normal operator AA in a Hilbert space such that σ(A)=Ω\sigma (A) = \Omega. Prove that every compact subset of the real axis is the spectrum of some self-adjoint operator. Prove that every closed subset of the unit circle is the spectrum of some unitary operator.

?
Problem 12.43

Prove that for every compact set Ω⊂C\Omega \subset \mathbb {C} there exists an operator A in a Hilbert space such that σ(A)=σp(A)=Ω\sigma (A) = \sigma_{\mathrm{p}}(A) = \Omega.

?
Problem 12.44

Let HH be a Hilbert space, U∈B(H)U \in \mathcal{B}(H). Prove that UU is an orthogonal reflection if and only if U∗=U=U−1U^{*} = U = U^{-1}. Find σ(U)\sigma (U).

?
Problem 12.45

Let XX be a normed space, A,B∈B(X)A, B \in \mathcal{B}(X). Prove that if AA and BB commute, then Rλ(A)R_{\lambda }(A) commutes with BB for all λ∈ρ(A)\lambda \in \rho (A). Prove that if BB and Rλ(A)R_{\lambda }(A) commute for some λ∈ρ(A)\lambda \in \rho (A), then AA and BB commute.

?
Problem 12.46

Let XX be a Banach space, A,B∈B(X)A, B \in \mathcal{B}(X) and AB=BAA B = B A. Prove that

r(A+B)⩽r(A)+r(B),r(AB)⩽r(A)r(B) r(A+B) \leqslant r(A)+r(B), \quad r(A B) \leqslant r(A) r(B)
?
Problem 12.47

Give an example of a sequence of operators {An}1∞⊂B(X)\left\{ A_{n}\right\}_{1}^{\infty } \subset \mathcal{B}(X) and an operator AA in a Banach space XX such that An⇉A,σ(An)A_{n} \rightrightarrows A, \sigma \left(A_{n}\right) is the unit circle, and σ(A)\sigma (A) is the unit disk.

?
Problem 12.48

Let {An}1∞⊂B(X)\left\{ A_{n}\right\}_{1}^{\infty } \subset \mathcal{B}(X) and An⇉AA_{n} \rightrightarrows A.

?
(a)

Prove that if λ∈ρ(A)\lambda \in \rho (A), then λ∈ρ(An)\lambda \in \rho \left(A_{n}\right) for all nn starting from some index.

(b)

Let Uε(σ(A))={z∈C:dist⁡(z,σ(A))<ε}U_{\varepsilon }(\sigma (A)) = \left\{ z \in \mathbb {C}: \operatorname {dist}(z, \sigma (A)) < \varepsilon \right\} be the ε\varepsilon-neighborhood of the spectrum of the operator AA. Prove that for every ε>0\varepsilon > 0 there exists an index NN such that for all n>Nn > N the inclusion σ(An)⊂Uε(σ(A))\sigma \left(A_{n}\right) \subset U_{\varepsilon }(\sigma (A)) holds.

Problem 12.49

Find the numerical range of the following operators:

?
(a)

(1000)\left(\begin{array}{ll}1 & 0 \\ 0 & 0\end{array}\right) in l2(2)l_{2}(2);

(b)

(0010)\left(\begin{array}{ll}0 & 0 \\ 1 & 0\end{array}\right) in l2(2)l_{2}(2).

Problem 12.50

Prove the first statement of Theorem 12.2: for any bounded operator AA in a complex Hilbert space HH the numerical range W(A)W(A) is a convex bounded set.

?
Problem 12.51

Prove the second statement of Theorem 12.2: for any bounded operator AA in a complex Hilbert space HH the inclusions σ(A)⊂W(A)‾⊂{z∈C:∣z∣⩽∥A∥}\sigma (A) \subset \overline{W(A)} \subset \left\{ z \in \mathbb {C}:\left|z\right| \leqslant \left\| A\right\| \right\} hold.

?
Problem 12.52

Let HH be a complex Hilbert space, AA a self-adjoint operator. Prove that W(A)‾=[m,M]\overline{W(A)} = [m, M], where m=inf⁡{λ:λ∈σ(A)}m = \inf \left\{ \lambda : \lambda \in \sigma (A)\right\}, and M=sup⁡{λ:λ∈σ(A)}M = \sup \left\{ \lambda : \lambda \in \sigma (A)\right\}. Must the set W(A)W(A) be closed?

?
Problem 12.53

Let HH be a complex Hilbert space, AA a self-adjoint operator. Prove that W(A)‾=[m,M]\overline{W(A)} = [m, M], where either M=∥A∥M = \left\| A\right\|, or m=−∥A∥m = -\left\| A\right\|.

?
Problem 12.54

Let HH be a complex Hilbert space, A∈B(H)A \in \mathcal{B}(H), and suppose for every xx the number (Ax,x)(A x, x) is zero. Prove that A=0A = 0. Give an example of a nonzero operator AA in a real Hilbert space with the property (Ax,x)=0(A x, x) = 0 for every xx.

?
Problem 12.55

Let HH be a complex Hilbert space, A∈B(H),w(A):=sup⁡{∣z∣:z∈W(A)}A \in \mathcal{B}(H), w(A): = \sup \left\{ \left|z\right|: z \in W(A)\right\}. Prove that w(A)⩽∥A∥w(A) \leqslant \left\| A\right\|.

?
Problem 12.56

Let HH be a Hilbert space, A∈B(H)A \in \mathcal{B}(H). Prove that if ∣λ∣=∥A∥\left|\lambda \right| = \left\| A\right\| and λ∈W(A)\lambda \in W(A), then λ∈σp(A)\lambda \in \sigma_{\mathrm{p}}(A).

?
Problem 12.57

Prove that for a normal operator A∈B(H)A \in \mathcal{B}(H) in a Hilbert space, the set W(A)‾\overline{W(A)} coincides with the convex hull of the spectrum σ(A)\sigma (A).

?
Problem 12.58

Find the numerical range of the following operators:

?
(a)

the multiplication operator by a bounded sequence in l2:Ax=(λ1x1,λ2x2,…,λnxn,…)l_{2}: A x = \left(\lambda_{1} x_{1}, \lambda_{2} x_{2}, \ldots , \lambda_{n} x_{n}, \ldots \right);

(b)

the multiplication operator (Ax)(t)=a(t)x(t)(A x)(t) = a(t) x(t) in L2[0,1],a∈C[0,1]L_{2}[0,1], a \in C[0,1];

(c)

the operators of right and left shift in the space l2l_{2}.

Problem 12.59

Let XX be an infinite-dimensional Banach space, A∈K(X)A \in \mathcal{K}(X). Prove that 0∈σess (A)0 \in \sigma_{\text{ess }}(A).

?
Problem 12.60

Let XX be an infinite-dimensional Banach space, A∈K(X)A \in \mathcal{K}(X). Relying on the Fredholm alternative, prove that:

?
(a)

σess (A)\{0}=∅\sigma_{\text{ess }}(A) \backslash \left\{ 0\right\} = \varnothing;

(b)

if λ∈σ(A),λ≠0\lambda \in \sigma (A), \lambda \neq 0, then λ∈σp(A)\lambda \in \sigma_{\mathrm{p}}(A).

Problem 12.61

Prove that a nonzero eigenvalue of a compact operator has finite multiplicity.

?
Problem 12.62

Let XX be a Banach space, A∈K(X)A \in \mathcal{K}(X). Prove that for every δ>0\delta > 0 the set σp(A)∩{z:∣z∣>δ}\sigma_{\mathrm{p}}(A) \cap \left\{ z:\left|z\right| > \delta \right\} consists of a finite number of points, each of which is an eigenvalue of the operator AA of finite multiplicity.

?
Problem 12.63

Prove that the integral operator

(Ax)(t)=∫atK(t,s)x(s)ds (A x)(t) = \int _{a}^{t} K(t, s) x(s) d s

is a Volterra operator in the space

?
(a)

C[a,b]C[a, b], if the function K(t,s)K(t, s) is continuous for a⩽s⩽ta \leqslant s \leqslant t;

(b)

L2[a,b]L_{2}[a, b], if the function K(t,s)K(t, s) is measurable and bounded for a⩽s⩽ta \leqslant s \leqslant t.

Problem 12.64

Let AA be a Volterra operator and λ∈C\lambda \in \mathbb {C}. Prove that the equation (I−λA)x=y(I-\lambda A) x = y in a Banach space XX is solvable for every right-hand side and every λ\lambda. Prove that the solution is representable as the series x=∑k=0∞λkAkyx = \sum_{k = 0}^{\infty } \lambda^{k} A^{k} y.

?
Problem 12.65

Solve the equation x(t)+λ∫0tx(s)ds=y(t)x(t)+\lambda \int_{0}^{t} x(s) d s = y(t) (with respect to the function xx, with the function yy known) in the following spaces:

?
(a)

L2[0,1]L_{2}[0,1];

(b)

C[0,1]C[0,1].

Problem 12.66

Let XX be a complex infinite-dimensional Banach space, A∈B(X)A \in \mathcal{B}(X) and σ(A)={0}\sigma (A) = \left\{ 0\right\}. Is it true that AA is compact?

?
Problem 12.67

Let XX be an infinite-dimensional Banach space. Give examples of operators A∈K(X)A \in \mathcal{K}(X) for which

?
(a)

0∈σp(A)0 \in \sigma_{\mathrm{p}}(A);

(b)

0∈σc(A)0 \in \sigma_{c}(A);

(c)

0∈σr(A)0 \in \sigma_{r}(A).

Problem 12.68

By investigating the eigenvalues of the Riemann-Liouville operator in the space L2[0,1]L_{2}[0,1] with arbitrary parameter r>1/2r > 1 / 2 (see problem 6.35), prove the non-compactness of this operator.

?
Problem 12.69

Find the eigenvalues of the Hardy operator

(Ax)(t)=1t∫0tx(s)ds (A x)(t) = \frac{1}{t} \int _{0}^{t} x(s) d s

in the spaces

?
(a)

Lp[0,1],1<p<∞L_{p}[0,1], 1 < p < \infty;

(b)

L∞[0,1]L_{\infty }[0,1];

(c)

C[0,1]C[0,1] (extended by the equality (Ax)(0)=x(0))(A x)(0) = x(0)), and on the basis of the result obtained, draw a conclusion about the non-compactness of this operator in the indicated spaces (cf. problem 11.41).

Problem 12.70

Find the eigenvalues of the operator

(Ax)(t)=1tα∫0tsα−1x(s)ds,Re⁡α>1p (A x)(t) = \frac{1}{t^{\alpha }} \int _{0}^{t} s^{\alpha -1} x(s) d s, \quad \operatorname {Re} \alpha > \frac{1}{p}

in the space Lp[0,1](1⩽p<∞)L_{p}[0,1](1 \leqslant p < \infty ).

?
Problem 12.71

In the space C[0,π]C[0, \pi ], find the eigenvalues and eigenfunctions of the integral operator

(Ax)(t)=∫0πK(t,s)x(s)ds (A x)(t) = \int _{0}^{\pi } K(t, s) x(s) d s

in the case

?
(a)

K(t,s)=sin⁡(t+s)K(t, s) = \sin (t+s);

(b)

K(t,s)=cos⁡(t+s)K(t, s) = \cos (t+s).

Problem 12.72

In the space L2[−π,π]L_{2}[-\pi , \pi ], find the eigenvalues and eigenfunctions of the integral operator

(Ax)(t)=∫−ππK(t,s)x(s)ds (A x)(t) = \int _{-\pi }^{\pi } K(t, s) x(s) d s

if K(t,s)=∑n∈Z3−∣n∣e−in(t−s)K(t, s) = \sum_{n \in \mathbb {Z}} 3^{-\left|n\right|} e^{-i n(t-s)}.

?
Problem 12.73

In the space L2[0,π]L_{2}[0, \pi ], find the eigenvalues and eigenfunctions of the integral operator

(Ax)(t)=∫0πK(t,s)x(s)ds (A x)(t) = \int _{0}^{\pi } K(t, s) x(s) d s

if

K(t,s)={sin⁡tcos⁡s for 0⩽t⩽s⩽π,sin⁡scos⁡t for 0⩽s⩽t⩽π. K(t, s) = \begin{cases} \sin t \cos s & \text{ for } 0 \leqslant t \leqslant s \leqslant \pi , \\ \sin s \cos t & \text{ for } 0 \leqslant s \leqslant t \leqslant \pi . \end{cases}
?
Problem 12.74

In the space L2[0,1]L_{2}[0,1], find the eigenvalues and eigenfunctions of the integral operator

(Ax)(t)=∫01K(t,s)x(s)ds (A x)(t) = \int _{0}^{1} K(t, s) x(s) d s

if

?
(a)

K(t,s)=ts+max⁡(t,s)K(t, s) = t s+\max (t, s);

(b)

K(t,s)=ts+max⁡(t,s)−2min⁡(t,s)−K(t, s) = t s+\max (t, s)-2 \min (t, s)- -2.

Problem 12.75

(H. Weyl, 1910). Let XX be a Banach space, A∈B(X),λ∈σ(A)\σp(A)A \in \mathcal{B}(X), \lambda \in \sigma (A) \backslash \sigma_{\mathrm{p}}(A) and B=A+KB = A+K, where K∈K(X)K \in \mathcal{K}(X). Prove that λ∈σ(B)\lambda \in \sigma (B).

?
Problem 12.76

Let Tr∈B(lp(Z)),p∈[1,∞)T_{r} \in \mathcal{B}\left(l_{p}(\mathbb {Z})\right), p \in [1, \infty ), be the right shift operator, i.e., (Trx)n=xn−1,n∈Z\left(T_{r} x\right)_{n} = x_{n-1}, n \in \mathbb {Z}. Prove that σ(Tr)\sigma \left(T_{r}\right) coincides with the unit circle, and construct a one-dimensional operator K∈B(lp(Z))K \in \mathcal{B}\left(l_{p}(\mathbb {Z})\right) such that σ(Tr+K)\sigma \left(T_{r}+K\right) coincides with the unit disk. (See Example 12.1 and cf. problem 12.47).

?
Problem 12.77

Let XX be an infinite-dimensional Banach space, A,B∈B(X),K∈K(X)A, B \in \mathcal{B}(X), K \in \mathcal{K}(X). Prove that the equality AB−BA=I+KA B-B A = I+K is impossible (i.e., a compact perturbation of the identity operator cannot be a commutator, cf. problem 12.41).

?
Problem 12.78

Prove Theorem 12.3.

Theorem 12.3 (D. Hilbert, 1906; E. Schmidt, 1907): Let HH be an infinite-dimensional separable Hilbert space and A=A∗∈K(H)A = A^{*} \in \mathcal{K}(H). Then there exists an orthonormal basis {en}n=1∞\left\{ e_{n}\right\}_{n = 1}^{\infty } of HH consisting of eigenvectors of AA: Aen=λnenA e_{n} = \lambda_{n} e_{n}. The nonzero eigenvalues are real and can be numbered in nonincreasing order of absolute value: ∣λ1∣⩾∣λ2∣⩾…⩾∣λn∣⩾…\left|\lambda_{1}\right| \geqslant \left|\lambda_{2}\right| \geqslant \ldots \geqslant \left|\lambda_{n}\right| \geqslant \ldots, and moreover ∣λ1∣=∥A∥\left|\lambda_{1}\right| = \left\| A\right\| and lim⁡n→∞λn=0\lim_{n \rightarrow \infty } \lambda_{n} = 0.

?
Problem 12.79

(D. Hilbert, 1906; E. Schmidt, 1907). Prove that in a separable Hilbert space a compact operator can be represented in the form

Ax=∑i=1Nsi(A)(x,ψi)φi A x = \sum _{i = 1}^{N} s_{i}(A)\left(x, \psi _{i}\right) \varphi _{i}

where N⩽∞,{ψi}N \leqslant \infty ,\left\{ \psi_{i}\right\} is an orthonormal basis, {φi}\left\{ \varphi_{i}\right\} is an orthonormal system.

?
Problem 12.80

Let HH be a separable Hilbert space, A∈K(H)A \in \mathcal{K}(H). Prove that the numbers sk(A)s_{k}(A) are the distances from the operator AA to the set of all (k−1)(k-1)-dimensional operators, i.e.,

sk(A)=inf⁡rank⁡Pk=k−1∥A−Pk∥ s_{k}(A) = \inf _{\operatorname {rank} P_{k} = k-1}\left\| A-P_{k}\right\|

In particular, s1=∥A∥s_{1} = \left\| A\right\|.

?
Problem 12.81

(Courant-Fischer variational principle). Prove that for every compact self-adjoint nonnegative operator AA in a Hilbert space HH the equalities

λn=min⁡{max⁡{(Ax,x):x⊥X,∥x∥=1}:dim⁡X=n−1} \lambda _{n} = \min \left\{ \max \left\{ (A x, x): x \perp X,\left\| x\right\| = 1\right\} : \operatorname {dim} X = n-1\right\}

hold, where λn\lambda_{n} are the eigenvalues of the operator AA, numbered in non-increasing order counting multiplicity, and the minimum is taken over all subspaces of dimension n−1n-1.

?
Problem 12.82

Let AA and BB be compact self-adjoint operators in a Hilbert space and 0⩽A⩽B0 \leqslant A \leqslant B. Prove that λn(A)⩽λn(B)\lambda_{n}(A) \leqslant \lambda_{n}(B), where λn\lambda_{n} are the eigenvalues, numbered in non-increasing order counting multiplicity.

?
Problem 12.83

Let AA and BB be compact operators in a Hilbert space HH, with Im⁡A⊂Im⁡B\operatorname {Im} A \subset \operatorname {Im} B. Prove that there exists a number C>0C > 0 such that sn(A)⩽Csn(B)s_{n}(A) \leqslant C s_{n}(B) for all n∈Nn \in \mathbb {N}.

?
Problem 12.84

Let A∈B(L2[0,1])A \in \mathcal{B}\left(L_{2}[0,1]\right) and Im⁡A⊂W˙21[0,1]\operatorname {Im} A \subset \dot{W}_{2}^{1}[0,1]. Prove that A∈K(L2[0,1])A \in \mathcal{K}\left(L_{2}[0,1]\right) and sn(A)⩽Cn−1s_{n}(A) \leqslant C n^{-1} for some C>0C > 0 and every n∈Nn \in \mathbb {N}.

?
Problem 12.85

Given a sequence sk⩾0,sk↘0s_{k} \geqslant 0, s_{k} \searrow 0. Prove that there exists an operator A∈K(l2)A \in \mathcal{K}\left(l_{2}\right) such that sk(A)=sks_{k}(A) = s_{k} for every k⩾1k \geqslant 1.

?
Problem 12.86

Let HH be an infinite-dimensional separable Hilbert space, A∈K(H),{sk}1∞A \in \mathcal{K}(H),\left\{ s_{k}\right\}_{1}^{\infty } the ss-numbers of the operator A,{en}1∞A,\left\{ e_{n}\right\}_{1}^{\infty } an arbitrary orthonormal basis in HH. Prove that ∑k=1∞sk2=∑n=1∞∥Aen∥2\sum_{k = 1}^{\infty } s_{k}^{2} = \sum_{n = 1}^{\infty }\left\| A e_{n}\right\|^{2} (i.e., either both series diverge, or both series converge and their sums are equal).

?
Problem 12.87

Let H1H_{1} and H2H_{2} be separable Hilbert spaces, A∈K(H1,H2),{en}1∞A \in \mathcal{K}\left(H_{1}, H_{2}\right),\left\{ e_{n}\right\}_{1}^{\infty } an orthonormal basis in H1H_{1}. Prove that ∥A∥2⩽∑n=1∞∥Aen∥2\left\| A\right\|^{2} \leqslant \sum_{n = 1}^{\infty }\left\| A e_{n}\right\|^{2}.

?
Problem 12.88

Let H1H_{1} and H2H_{2} be separable Hilbert spaces, A∈B(H1,H2),{en}1∞A \in \mathcal{B}\left(H_{1}, H_{2}\right),\left\{ e_{n}\right\}_{1}^{\infty } an orthonormal basis in H1H_{1}. Prove that if the series ∑n=1∞∥Aen∥2\sum_{n = 1}^{\infty }\left\| A e_{n}\right\|^{2} converges, then AA is compact.

?
Problem 12.89

Let HH be an infinite-dimensional separable Hilbert space, A∈K(H)A \in \mathcal{K}(H). Prove the equivalence of the following statements:

  1. A∈S2(H)A \in \mathcal{S}_{2}(H), i.e., the series ∑k=1∞sk2(A)\sum_{k = 1}^{\infty } s_{k}^{2}(A) converges;

  2. for some orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } in the space HH the series ∑n=1∞∥Aen∥2\sum_{n = 1}^{\infty }\left\| A e_{n}\right\|^{2} converges;

  3. for every orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } in the space HH the series ∑n=1∞∥Aen∥2\sum_{n = 1}^{\infty }\left\| A e_{n}\right\|^{2} converges;

  4. for the matrix entries ajk=(Aej,ek)a_{j k} = \left(A e_{j}, e_{k}\right) of the operator AA in some orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } of the space HH the inequality ∑j,k=1∞∣ajk∣2<∞\sum_{j, k = 1}^{\infty }\left|a_{j k}\right|^{2} < \infty holds;

  5. for the matrix entries ajk=(Aej,ek)a_{j k} = \left(A e_{j}, e_{k}\right) of the operator AA in an arbitrary orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } of the space HH the inequality ∑j,k=1∞∣ajk∣2<∞\sum_{j, k = 1}^{\infty }\left|a_{j k}\right|^{2} < \infty holds. Prove that if the operator AA satisfies these conditions, then the sums of all the indicated series coincide.

?
Problem 12.90

Let HH be a separable infinite-dimensional Hilbert space. Prove that the set S2(H)\mathcal{S}_{2}(H) is a non-closed (in the norm of B(H)\mathcal{B}(H)) linear subspace of the space K(H)\mathcal{K}(H).

?
Problem 12.91

Let HH be a separable infinite-dimensional Hilbert space. Prove that the norm in S2(H)\mathcal{S}_{2}(H) is generated by the inner product

(A,B)S2(H)=∑j,k=1∞ajkbˉjk (A, B)_{\mathcal{S}_{2}(H)} = \sum _{j, k = 1}^{\infty } a_{j k} \bar{b}_{j k}

where ajk=(Aej,ek)a_{j k} = \left(A e_{j}, e_{k}\right) and bjk=(Bej,ek)b_{j k} = \left(B e_{j}, e_{k}\right) are the matrix entries of the operators AA and BB in an arbitrary orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } of the space HH. Prove that the same inner product can be given by the rule

(A,B)S2(H)=∑k=1∞sk(A∗B) (A, B)_{\mathcal{S}_{2}(H)} = \sum _{k = 1}^{\infty } s_{k}\left(A^{*} B\right)
?
Problem 12.92

Let HH be a separable Hilbert space. Prove that S2(H)\mathcal{S}_{2}(H) is a Hilbert space with the inner product defined in the previous problem.

?
Problem 12.93

Prove that an operator AA in the space L2[0,1]L_{2}[0,1] is a Hilbert-Schmidt operator if and only if

(Ax)(t)=∫01K(t,s)x(s)ds (A x)(t) = \int _{0}^{1} K(t, s) x(s) d s

where K∈L2([0,1]2)K \in L_{2}\left([0,1]^{2}\right). Prove that the map A↦KA \mapsto K is a unitary isomorphism of the spaces S2(L2[0,1])\mathcal{S}_{2}\left(L_{2}[0,1]\right) and L2([0,1]2)L_{2}\left([0,1]^{2}\right) (in particular, ∥A∥S2(L2[0,1])2=∫01∫01∣K(x,t)∣2dxdt\left\| A\right\|_{\mathcal{S}_{2}\left(L_{2}[0,1]\right)}^{2} = \int_{0}^{1} \int_{0}^{1}|K(x, t)|^{2} d x d t).

?
Problem 12.94

Prove that every bounded linear operator in the space L2[0,1]L_{2}[0,1] has the form

(Ax)(t)=ddt∫01K(t,s)x(s)ds (A x)(t) = \frac{d}{d t} \int _{0}^{1} K(t, s) x(s) d s

for some function K∈L2([0,1]2)K \in L_{2}\left([0,1]^{2}\right).

?
Problem 12.95

Prove that the class Sp⁡(X)\operatorname {Sp}(X) of nuclear operators in a Banach space XX (see Definition 11.2) coincides with the class S1(H)\mathcal{S}_{1}(H), if the space X=HX = H is Hilbert.

Definition 11.2: Let XX and YY be Banach spaces. An operator A∈B(X,Y)A \in \mathcal{B}(X, Y) is called nuclear, or a trace-class operator (A∈Sp⁡(X,Y)A \in \operatorname {Sp}(X, Y)), if there exist one-dimensional operators Ak∈B(X,Y)A_{k} \in \mathcal{B}(X, Y) such that A=∑k=1∞AkA = \sum_{k = 1}^{\infty } A_{k}, where the series converges absolutely in B(X,Y)\mathcal{B}(X, Y).

Here (Definition 12.7), for A∈K(H)A \in \mathcal{K}(H) the ss-numbers of AA are sn(A):=λn(A∗A)s_{n}(A) := \sqrt{\lambda_{n}\left(A^{*} A\right)}, n=1,2,…n = 1,2, \ldots (numbered in nonincreasing order, counting multiplicity), and A∈S1(H)A \in \mathcal{S}_{1}(H) if {sn}n=1∞∈l1\left\{ s_{n}\right\}_{n = 1}^{\infty } \in l_{1}.

?
Problem 12.96

Let HH be a separable Hilbert space and A∈S1(H)A \in \mathcal{S}_{1}(H). Prove that for an arbitrary orthonormal basis {ek}k=1∞\left\{ e_{k}\right\}_{k = 1}^{\infty } the series ∑k=1∞(Aek,ek)\sum_{k = 1}^{\infty }\left(A e_{k}, e_{k}\right) converges, and its sum does not depend on the choice of the basis {ek}k=1∞\left\{ e_{k}\right\}_{k = 1}^{\infty }.

?
Problem 12.97

Prove that for an operator A∈S1(H)=Sp⁡(H)A \in \mathcal{S}_{1}(H) = \operatorname {Sp}(H) the equalities

∥A∥Sp⁡(H)=∑k=1∞sk(A)=sup⁡U− unitary (∑k=1∞(UAek,ek)), \left\| A\right\| _{\operatorname {Sp}(H)} = \sum _{k = 1}^{\infty } s_{k}(A) = \sup _{U-\text{ unitary }}\left(\sum _{k = 1}^{\infty }\left(U A e_{k}, e_{k}\right)\right),

hold, where {ek}k=1∞\left\{ e_{k}\right\}_{k = 1}^{\infty } is an arbitrary orthonormal basis (for the definition of ∥A∥Sp(H)\left\| A\right\|_{\mathrm{Sp}(H)} see problem 11.27).

?
Problem 12.98

Let HH be a separable Hilbert space. Prove that Sp(H)\mathcal{S}_{p}(H) are linear subspaces of K(H)\mathcal{K}(H) and of B(H)\mathcal{B}(H). Prove that Sp(H)\mathcal{S}_{p}(H) are non-closed two-sided ideals in B(H)\mathcal{B}(H).

?
Problem 12.99

Let A,B∈S2(H)A, B \in \mathcal{S}_{2}(H). Prove that AB∈S1(H)A B \in \mathcal{S}_{1}(H).

?
Problem 12.100

Let TT be an integral operator in the space L2[0,1]L_{2}[0,1], (Tx)(t)=∫01K(t,s)x(s)ds(T x)(t) = \int_{0}^{1} K(t, s) x(s) d s, where K(⋅,⋅)∈L2([0,1]2)K(\cdot , \cdot ) \in L_{2}\left([0,1]^{2}\right) and for almost every s∈[0,1]s \in [0,1] the function t↦K(t,s)t \mapsto K(t, s) satisfies the Lipschitz condition: ∣K(t2,s)−K(t1,s)∣⩽C(s)∣t2−t1∣\left|K\left(t_{2}, s\right)-K\left(t_{1}, s\right)\right| \leqslant C(s)\left|t_{2}-t_{1}\right| for all t1,t2∈[0,1]t_{1}, t_{2} \in [0,1], where the function C(⋅)∈L2[0,1]C(\cdot ) \in L_{2}[0,1]. Prove that TT is a nuclear operator. Prove that if the function KK is moreover continuous on the square [0,1]2[0,1]^{2}, then tr⁡(T)=∫01K(t,t)dt\operatorname {tr}(T) = \int_{0}^{1} K(t, t) d t.

?
Problem 12.101

(T. Carleman's example). Give an example of a non-nuclear integral operator in the space L2[0,1]L_{2}[0,1] with continuous kernel.

?
Problem 12.102

Do there exist:

?
(a)

a nuclear operator AA and an orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } in a Hilbert space such that ∑n=1∞∥Aen∥=+∞\sum_{n = 1}^{\infty }\left\| A e_{n}\right\| = +\infty;

(b)

a bounded but non-nuclear operator AA and an orthonormal basis {en}1∞\left\{ e_{n}\right\}_{1}^{\infty } such that (Aen,en)=0\left(A e_{n}, e_{n}\right) = 0 for all n∈Nn \in \mathbb {N} and hence ∑n=1∞(Aen,en)<∞\sum_{n = 1}^{\infty }\left(A e_{n}, e_{n}\right) < \infty?

Problem 12.103

Let HH be a separable Hilbert space and A∈K(H),A⩾0A \in \mathcal{K}(H), A \geqslant 0. Let λk(A)\lambda_{k}(A) be the eigenvalues of the operator AA, numbered in non-increasing order. Prove that λk(A)=sk(A)\lambda_{k}(A) = s_{k}(A).

?
Problem 12.104

Let AA and BB be compact operators in a Hilbert space HH, with Im⁡A⊂Im⁡B\operatorname {Im} A \subset \operatorname {Im} B. Prove that if B∈Sp(H)B \in \mathcal{S}_{p}(H), then A∈Sp(H)A \in \mathcal{S}_{p}(H) as well.

?
Problem 12.105

Find the norm of the integration operator

(Ax)(t)=∫0tx(s)ds (A x)(t) = \int _{0}^{t} x(s) d s

in L2[0,1]L_{2}[0,1]:

?
(a)

using problem 12.80;

(b)

by expanding the function x(s)x(s) in a series with respect to the system {cos⁡(π(k−1/2)s)}1∞\left\{ \cos (\pi (k-1 / 2) s)\right\}_{1}^{\infty } (see problem 3.50).

Problem 12.106

Find the polar decomposition A=WSA = W S for the operator from problem 12.105.

?
Problem 12.107

Find the norms of the operators in L2[−b,b]L_{2}[-b, b]:

?
(a)

(Ax)(t)=∫−btx(s)ds(A x)(t) = \int_{-b}^{t} x(s) d s;

(b)

(Ax)(t)=∫tbx(s)ds(A x)(t) = \int_{t}^{b} x(s) d s;

(c)

(Ax)(t)=∫0tx(s)ds(A x)(t) = \int_{0}^{t} x(s) d s.

Problem 12.108

Find the spectrum of the integral operators in L2[0,π]L_{2}[0, \pi ] with kernel

?
(a)

K(t,s)=∑n=1∞2−nsin⁡ntsin⁡nsK(t, s) = \sum_{n = 1}^{\infty } 2^{-n} \sin n t \sin n s;

(b)

K(t,s)=∑n=0∞2−ncos⁡ntcos⁡nsK(t, s) = \sum_{n = 0}^{\infty } 2^{-n} \cos n t \cos n s;

(c)

K(t,s)=∑n=1∞2−ncos⁡ntcos⁡nsK(t, s) = \sum_{n = 1}^{\infty } 2^{-n} \cos n t \cos n s.

Problem 12.109

Find the spectrum of the integral operators in L2[0,1]L_{2}[0,1] with kernel

?
(a)

K(t,s)=min⁡(t,s)K(t, s) = \min (t, s);

(b)

K(t,s)=max⁡(t,s)K(t, s) = \max (t, s) (compare with problem 11.96). Determine to which Neumann-Schatten classes these operators belong.

Problem 12.110

For the operator A(x1,x2,…)=(λ1x1,λ2x2,…)A\left(x_{1}, x_{2}, \ldots \right) = \left(\lambda_{1} x_{1}, \lambda_{2} x_{2}, \ldots \right) in the space l2l_{2}, where λ:={λn}1∞∈l∞\lambda : = \left\{ \lambda_{n}\right\}_{1}^{\infty } \in l_{\infty }, find criteria for the following statements in terms of the sequence λn\lambda_{n}:

?
(a)

A∈K(l2)A \in \mathcal{K}\left(l_{2}\right);

(b)

A∈Sp(l2),p∈[1,∞)A \in \mathcal{S}_{p}\left(l_{2}\right), p \in [1, \infty );

(c)

A∈IB(l2)A \in \mathcal{I B}\left(l_{2}\right);

(d)

A∈F(l2)A \in \mathcal{F}\left(l_{2}\right);

(e)

AA is a self-adjoint operator;

(f)

AA is a unitary operator;

(g)

AA is a projector;

(h)

AA is an orthogonal projector;

(i)

AA is a partial isometry;

(j)

AA is a positive operator;

(k)

AA is a normal operator;

(l)

AA is an orthogonal reflection.

Problem 12.111

For the operator Af(x)=a(x)f(x)A f(x) = a(x) f(x) in the space L2[0,1]L_{2}[0,1], where a∈L∞[0,1]a \in L_{\infty }[0,1], find criteria for the following statements in terms of the function aa:

?
(a)

A∈K(L2[0,1])A \in \mathcal{K}\left(L_{2}[0,1]\right);

(b)

A∈S1(L2[0,1])A \in \mathcal{S}_{1}\left(L_{2}[0,1]\right);

(c)

A∈IB(L2[0,1])A \in \mathcal{I B}\left(L_{2}[0,1]\right);

(d)

A∈F(L2[0,1])A \in \mathcal{F}\left(L_{2}[0,1]\right);

(e)

AA is a self-adjoint operator;

(f)

AA is a unitary operator;

(g)

AA is a projector;

(h)

AA is an orthogonal projector;

(i)

AA is a partial isometry;

(j)

AA is a positive operator;

(k)

AA is a normal operator;

(l)

AA is an orthogonal reflection.