Section 9.2

Problems to Solve

[71/82%]
Show
LaTeX
Problem 9.6

Let HH be a Hilbert space, A,B∈L(H)A, B \in \mathcal{L}(H). Prove the following statements:

?
(1)

I∗=I,0∗=0I^{*} = I, 0^{*} = 0;

(2)

(αA+βB)∗=αˉA∗+βˉB∗,α,β∈C(\alpha A+\beta B)^{*} = \bar{\alpha } A^{*}+\bar{\beta } B^{*}, \alpha , \beta \in \mathbb {C};

(3)

(AB)∗=B∗A∗(A B)^{*} = B^{*} A^{*};

(4)

if A=A∗,B∗=BA = A^{*}, B^{*} = B, then (AB)∗=AB⇔AB=BA(A B)^{*} = A B \Leftrightarrow A B = B A;

(5)

if A−1∈L(H)A^{-1} \in \mathcal{L}(H) exists, then (A∗)−1∈L(H)\left(A^{*}\right)^{-1} \in \mathcal{L}(H) also exists, and (A∗)−1=(A−1)∗\left(A^{*}\right)^{-1} = \left(A^{-1}\right)^{*}.

Problem 9.7

Let HH be a Hilbert space over the field K,b:H×H→K\mathbb {K}, b: H \times H \rightarrow \mathbb {K} be a bilinear form, b[x]=b(x,x),x∈Hb\left[x\right] = b(x, x), x \in H, be a corresponding quadratic form. Prove that

?
(1)

if K=C\mathbb {K} = \mathbb {C}, then

4b(x,y)=b[x+y]−b[x−y]+ib[x+iy]−ib[x−iy],x,y∈H 4 b(x, y) = b\left[x+y\right]-b\left[x-y\right]+i b\left[x+i y\right]-i b\left[x-i y\right], x, y \in H

(polarization identity);

(2)

if K=R\mathbb {K} = \mathbb {R}, then the form bb is symmetric (i.e., b(x,y)=b(y,x),x,y∈Hb(x, y) = b(y, x), x, y \in H) if and only if 4b(x,y)=b[x+y]−b[x−y],x,y∈H4 b(x, y) = b\left[x+y\right]-b\left[x-y\right], x, y \in H.

Problem 9.8

Prove that the continuity of a bilinear form bb is equivalent to each of the following conditions:

?
(1)

the bilinear form bb is bounded;

(2)

there exists an operator A∈L(H)A \in \mathcal{L}(H) such that b(x,y)=(Ax,y),x,y∈Hb(x, y) = (A x, y), x, y \in H; in this case ∥b∥=∥A∥\left\| b\right\| = \left\| A\right\|;

(3)

there exists an operator C∈L(H)C \in \mathcal{L}(H) such that b(x,y)=(x,Cy),x,y∈Hb(x, y) = (x, C y), x, y \in H; in this case ∥b∥=∥C∥\left\| b\right\| = \left\| C\right\|.

Problem 9.9

Let b:H×H→Cb: H \times H \rightarrow \mathbb {C} be a bilinear form on a complex Hilbert space H,b[x]=b(x,x),x∈HH, b\left[x\right] = b(x, x), x \in H. Prove that

?
(1)

b(x,y)=1N∑k=1Nb[x+e2πikNy]e2πikN,x,y∈H,N∈N,N≥3b(x, y) = \frac{1}{N} \sum_{k = 1}^{N} b\left[x+e^{\frac{2 \pi i k}{N}} y\right] e^{\frac{2 \pi i k}{N}}, x, y \in H, N \in \mathbb {N}, N \geq 3;

(2)

b(x,y)=12π∫02πb[x+eiφy]eiφdφ,x,y∈Hb(x, y) = \frac{1}{2 \pi } \int_{0}^{2 \pi } b\left[x+e^{i \varphi } y\right] e^{i \varphi } d \varphi , x, y \in H.

Problem 9.10

Prove that in a complex Hilbert space the following statements are equivalent:

?
(1)

operator A∈L(H)A \in \mathcal{L}(H) is self-adjoint;

(2)

the bilinear form bAb_{A}, induced by the operator AA, is Hermitian;

(3)

the quadratic form bA[x]=bA(x,x),x∈Hb_{A}\left[x\right] = b_{A}(x, x), x \in H, attains only real values.

Prove that in a real Hilbert space the operator A∈L(H)A \in \mathcal{L}(H) is self-adjoint if and only if the bilinear form bAb_{A} is symmetric.

Problem 9.11
?
(1)

Let HH be a complex Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that A=0A = 0 if and only if (Ax,x)=0(A x, x) = 0 for all x∈Hx \in H.

(2)

Provide an example of a real Hilbert space HH and a nonzero operator A∈L(H)A \in \mathcal{L}(H) such that (Ax,x)=0(A x, x) = 0 for all x∈Hx \in H.

Problem 9.12

Let HH be a Hilbert space over the field K\mathbb {K} and A,B,C,D∈L(H)A, B, C, D \in \mathcal{L}(H). Prove that

?
(1)

if A∗=AA^{*} = A, then A2≥0A^{2} \geq 0;

(2)

the condition A∗=AA^{*} = A in item (1) is essential;

(3)

if A≤B,B≤AA \leq B, B \leq A, then A=BA = B;

(4)

if A≤B,B≤CA \leq B, B \leq C, then A≤CA \leq C;

(5)

if A≤C,B≤DA \leq C, B \leq D, then A+B≤C+DA+B \leq C+D;

(6)

if A≤B,λ≥0A \leq B, \lambda \geq 0, then λA≤λB\lambda A \leq \lambda B;

(7)

if A≤BA \leq B, then −B≤−A-B \leq -A;

(8)

if A∗=A,B∗=BA^{*} = A, B^{*} = B and AB=BAA B = B A, then A2+B2≥2ABA^{2}+B^{2} \geq 2 A B;

(9)

if A∗=AA^{*} = A, then there exist m,M∈Rm, M \in \mathbb {R} such that mI≤A≤MIm I \leq A \leq M I;

(10)

if A∗=A,B∗=BA^{*} = A, B^{*} = B, then it is not necessarily that A≤BA \leq B or B≤AB \leq A;

(11)

if A∗=A,A≥0A^{*} = A, A \geq 0 and A−1∈L(H)A^{-1} \in \mathcal{L}(H), then A−1≥0A^{-1} \geq 0;

(12)

if AA and BB are unitary operators, then ABA B is also a unitary operator;

(13)

if A∗=A,B∗=B,A≥BA^{*} = A, B^{*} = B, A \geq B, then the operators C∗AC,C∗BCC^{*} A C, C^{*} B C are selfadjoint and C∗AC≥C∗BCC^{*} A C \geq C^{*} B C;

(14)

if A2=AA^{2} = A, then R(A)=Ker⁡(I−A)R(A) = \operatorname {Ker}(I-A), in particular, the set R(A)R(A) is a subspace of HH;

(15)

if K=C,α,β∈C,∣α∣=∣β∣=1\mathbb {K} = \mathbb {C}, \alpha , \beta \in \mathbb {C},\left|\alpha \right| = \left|\beta \right| = 1, then the operator αA+βA∗\alpha A+\beta A^{*} is normal;

(16)

if AA is an orthoprojector, then A≥0A \geq 0;

(17)

if AA is an orthoprojector, then I−AI-A is also an orthoprojector;

(18)

if K=C,A=A∗,B=B∗\mathbb {K} = \mathbb {C}, A = A^{*}, B = B^{*}, then the operator C=i(AB−BA)C = i(A B-B A) is selfadjoint.

Problem 9.13

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that

?
(1)

AA is isometric ⇔∥Ax∥=∥x∥,x∈H\Leftrightarrow \left\| A x\right\| = \left\| x\right\| , x \in H;

(2)

if HH is a finite-dimensional space, then AA is unitary if and only if AA is isometric;

(3)

AA is isometric ⇔A∗A=I\Leftrightarrow A^{*} A = I;

(4)

AA is unitary ⇔A∗A=I\Leftrightarrow A^{*} A = I and AA∗=IA A^{*} = I;

(5)

if HH is an infinite-dimensional space, there exists AA which is isometric but not unitary.

Problem 9.14

Let a linear operator AA in Hilbert space Cm\mathbb {C}^{m} be represented by a matrix A=(ajk)j,k=1m\mathcal{A} = \left(a_{j k}\right)_{j, k = 1}^{m}, let A∗=A‾T\mathcal{A}^{*} = \overline{\mathcal{A}}^{T} be the transposed complex conjugate matrix, and J\mathcal{J} be the identity matrix. Prove that

?
(1)

AA is self-adjoint ⇔A∗=A\Leftrightarrow \mathcal{A}^{*} = \mathcal{A};

(2)

AA is unitary ⇔A\Leftrightarrow A is isometric ⇔A∗A=J\Leftrightarrow \mathcal{A}^{*} \mathcal{A} = \mathcal{J};

(3)

AA is nonnegative ⇔A∗=A\Leftrightarrow \mathcal{A}^{*} = \mathcal{A} and all roots of the equation det⁡(A−λJ)=0\operatorname {det}\left(\mathcal{A}-\lambda \mathcal{J}\right) = 0 are nonnegative;

(4)

if operator AA in Hilbert space Rm\mathbb {R}^{m} is represented by matrix A\mathcal{A} and A∗=AT\mathcal{A}^{*} = \mathcal{A}^{T} is the transposed matrix, then statements (1)-(3) are true, but the condition (Ax,x)≥0,x∈Rm(A x, x) \geq 0, x \in \mathbb {R}^{m}, does not imply that operator AA is self-adjoint, because it is not necessarily A∗=A\mathcal{A}^{*} = \mathcal{A}.

Problem 9.15

Find the adjoint operator of an operator A:l2→l2A: l_{2} \rightarrow l_{2}.

?
(1)

Ax=(α1x1,α2x2,…)A x = \left(\alpha_{1} x_{1}, \alpha_{2} x_{2}, \ldots \right);

(2)

Ax=(x2,x3,…)A x = \left(x_{2}, x_{3}, \ldots \right);

(3)

Ax=(x1,…,xj,0,0,…)A x = \left(x_{1}, \ldots , x_{j}, 0,0, \ldots \right);

(4)

Ax=(0,…,0⏟j−1,x1,0,0,…)A x = (\underbrace{0, \ldots , 0}_{j-1}, x_{1}, 0,0, \ldots );

(5)

Ax=(x1,…,x1⏟j,0,0,…)A x = (\underbrace{x_{1}, \ldots , x_{1}}_{j}, 0,0, \ldots );

(6)

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

(7)

Ax=(αjxj,αj+1xj+1,…)A x = \left(\alpha_{j} x_{j}, \alpha_{j+1} x_{j+1}, \ldots \right);

(8)

Ax=(0,0,α1x1,α2x2,…)A x = \left(0,0, \alpha_{1} x_{1}, \alpha_{2} x_{2}, \ldots \right);

(9)

Ax=(2x1+3x2,3x1+5x2,x3,0,0,…)A x = \left(2 x_{1}+3 x_{2}, 3 x_{1}+5 x_{2}, x_{3}, 0,0, \ldots \right);

(10)

Ax=(x1,4x2+7x3,7x2+2x3,0,x5,0,0,…)A x = \left(x_{1}, 4 x_{2}+7 x_{3}, 7 x_{2}+2 x_{3}, 0, x_{5}, 0,0, \ldots \right);

(11)

Ax=(3x1−2x2,x2,x3,x4,…)A x = \left(3 x_{1}-2 x_{2}, x_{2}, x_{3}, x_{4}, \ldots \right), where j∈Nj \in \mathbb {N} is fixed, and {αn:n≥1}⊂C\left\{ \alpha_{n}: n \geq 1\right\} \subset \mathbb {C} is a fixed bounded sequence.

Problem 9.16

Find the adjoint operator of an operator A:H→HA: H \rightarrow H.

?
(1)

H=L2(R),(Ax)(t)=χ[−α,α](t)x(t)H = L_{2}(\mathbb {R}),(A x)(t) = \chi_{\left[-\alpha , \alpha \right]}(t) x(t), where α>0\alpha > 0;

(2)

H=L2([0,1]),(Ax)(t)=tx(t)H = L_{2}(\left[0,1\right]),(A x)(t) = t x(t);

(3)

H=L2(R),(Ax)(t)=a(t)x(t+τ)H = L_{2}(\mathbb {R}),(A x)(t) = a(t) x(t+\tau ), where a∈L∞(R)a \in L_{\infty }(\mathbb {R}) is a given function, τ∈R;\tau \in \mathbb {R} ;

(4)

H=L2(R),(Ax)(t)=12(x(t)+x(−t))H = L_{2}(\mathbb {R}),(A x)(t) = \frac{1}{2}(x(t)+x(-t));

(5)

H=L2([0,1]),(Ax)(t)=∫01et+sx(s)dsH = L_{2}(\left[0,1\right]),(A x)(t) = \int_{0}^{1} e^{t+s} x(s) d s;

(6)

H=L2([0,1]),(Ax)(t)=t3∫0t2x(s)dsH = L_{2}(\left[0,1\right]),(A x)(t) = t^{3} \int_{0}^{t^{2}} x(s) d s;

(7)

H=L2([0,1]),(Ax)(t)=∫01t2s3x(s)dsH = L_{2}(\left[0,1\right]),(A x)(t) = \int_{0}^{1} t^{2} s^{3} x(s) d s;

(8)

H=L2([0,1]),(Ax)(t)=∫t5ttsx(s)dsH = L_{2}(\left[0,1\right]),(A x)(t) = \int_{t^{5}}^{t} t s x(s) d s;

(9)

H=L2([−1,1]),(Ax)(t)=∫t21(t−3s2)x(s)dsH = L_{2}(\left[-1,1\right]),(A x)(t) = \int_{t^{2}}^{1}\left(t-3 s^{2}\right) x(s) d s;

(10)

H=L2([0,1]),(Ax)(t)=x(tα)H = L_{2}(\left[0,1\right]),(A x)(t) = x\left(t^{\alpha }\right), where 0<α≤10 < \alpha \leq 1;

(11)

Ax=(x,y)zA x = (x, y) z, where y,z∈Hy, z \in H are fixed elements.

Problem 9.17

Find the adjoint operator of an operator A∈L(H)A \in \mathcal{L}(H). Prove that AA is a unitary operator.

?
(1)

H=L2(R),(Ax)(t)=7t3x(t7−1)H = L_{2}(\mathbb {R}),(A x)(t) = \sqrt{7} t^{3} x\left(t^{7}-1\right);

(2)

H=L2([0,∞)),(Ax)(t)=2ttx(t4)H = L_{2}([0, \infty )),(A x)(t) = 2 t \sqrt{t} x\left(t^{4}\right).

Problem 9.18

Let {αn:n≥1}⊂C\left\{ \alpha_{n}: n \geq 1\right\} \subset \mathbb {C} be a fixed bounded sequence,

Ax=(α1x1,α2x2,…),x=(x1,x2,…)∈l2. A x = \left(\alpha _{1} x_{1}, \alpha _{2} x_{2}, \ldots \right), x = \left(x_{1}, x_{2}, \ldots \right) \in l_{2}.

Express the following statements in terms of the numbers αn,n≥1\alpha_{n}, n \geq 1 :

?
(1)

AA is self-adjoint;

(2)

AA is normal;

(3)

AA is unitary;

(4)

AA is isometric;

(5)

AA is an orthogonal projector;

(6)

Ker⁡A={0}\operatorname {Ker} A = \left\{ 0\right\}.

Problem 9.19

Let (T,F,μ)(T, \mathcal{F}, \mu ) be a space with a σ\sigma-finite measure, AA be an operator of multiplication by the function p∈L∞(T,μ)p \in L_{\infty }(T, \mu ) in the space L2(T,μ)L_{2}(T, \mu ). Prove that

?
(1)

A=A∗⇔A = A^{*} \Leftrightarrow the function pp is almost everywhere real;

(2)

AA is unitary ⇔∣p∣=1\Leftrightarrow \left|p\right| = 1 almost everywhere;

(3)

A≥0⇔p≥0A \geq 0 \Leftrightarrow p \geq 0 almost everywhere;

(4)

AA is an orthogonal projector ⇔p2=p\Leftrightarrow p^{2} = p almost everywhere, i.e. p=χC( mod μ)p = \chi_{C}(\bmod \mu ), where C∈FC \in \mathcal{F} is a measurable set.

Problem 9.20

Let s∈R,(Asx)(t)=x(t+s),t∈R,x∈L2(R)s \in \mathbb {R},\left(A_{s} x\right)(t) = x(t+s), t \in \mathbb {R}, x \in L_{2}(\mathbb {R}) be a shift operator.

?
(1)

Find As∗A_{s}^{*} and As−1A_{s}^{-1} and prove that AsA_{s} is a unitary operator.

(2)

Let A=12(As0+As1),s0,s1∈RA = \frac{1}{2}\left(A_{s_{0}}+A_{s_{1}}\right), s_{0}, s_{1} \in \mathbb {R}. Prove that A∗=AA^{*} = A if and only if s0+s1=0s_{0}+s_{1} = 0.

Problem 9.21

Under what conditions on a subspace of a Hilbert space does an orthoprojector onto this subspace have a continuous inverse?

?
Problem 9.22

Let HH be a separable Hilbert space and {en:n≥1}\left\{ e_{n}: n \geq 1\right\} be an orthonormal basis in HH.

?
(1)

Find the orthogonal projector onto the linear span of the set {e1,…,ek},k∈N\left\{ e_{1}, \ldots , e_{k}\right\} , k \in \mathbb {N}.

(2)

Prove that the operator Ax=∑n=1∞xσ(n)en,x=∑n=1∞xnen∈HA x = \sum_{n = 1}^{\infty } x_{\sigma (n)} e_{n}, x = \sum_{n = 1}^{\infty } x_{n} e_{n} \in H, where σ:N→N\sigma : \mathbb {N} \rightarrow \mathbb {N} is a fixed bijection, is a unitary operator.

Problem 9.23

Provide an example of a Hilbert space HH and operators A,B∈L(H)A, B \in \mathcal{L}(H) such that

?
(1)

AA is normal, but not self-adjoint;

(2)

A2=AA^{2} = A, but AA is not an orthogonal projector;

(3)

AA is a unitary operator, but not an orthogonal projector;

(4)

A∗=AA^{*} = A, but does not satisfy either of the inequalities A≥0,A≤0A \geq 0, A \leq 0;

(5)

A2=IA^{2} = I, but does not satisfy either of the inequalities A≥0,A≤0A \geq 0, A \leq 0;

(6)

A∗=AA^{*} = A, but the range R(A)R(A) is not closed in HH;

(7)

A∗=AA^{*} = A, but the set {(Ax,x)∣x∈H,∥x∥=1}\left\{ (A x, x) \mid x \in H,\left\| x\right\| = 1\right\} is not closed in R\mathbb {R};

(8)

A≥0,B≥0A \geq 0, B \geq 0, but ABA B is not positive;

(9)

AA is not normal;

(10)

A≠0A \neq 0, but A2=0A^{2} = 0;

(11)

A=A∗,B=B∗,A2=B2A = A^{*}, B = B^{*}, A^{2} = B^{2}, but AB≠BAA B \neq B A.

Problem 9.24

Let HH be a Hilbert space over the field K,A∈L(H),A≥0\mathbb {K}, A \in \mathcal{L}(H), A \geq 0. Prove that

?
(1)

∣(Ax,y)∣2≤(Ax,x)(Ay,y),x,y∈H\left|(A x, y)\right|^{2} \leq (A x, x)(A y, y), x, y \in H;

(2)

∥Ax∥2≤∥A∥(Ax,x),x∈H\left\| A x\right\|^{2} \leq \left\| A\right\| (A x, x), x \in H.

Problem 9.25

Let A∈L(H)A \in \mathcal{L}(H). Is the set G={x∈H∣(Ax,x)=0}G = \left\{ x \in H \mid (A x, x) = 0\right\} necessarily a subspace in the Hilbert space HH, if

?
(1)

A∗=AA^{*} = A;

(2)

A≥0A \geq 0?

Problem 9.26

Let HH be a Hilbert space, A∈L(H),A∗=A,x∈H,x∉Ker⁡AA \in \mathcal{L}(H), A^{*} = A, x \in H, x \notin \operatorname {Ker} A. Prove that

?
(1)

Anx≠0A^{n} x \neq 0 for all n∈Nn \in \mathbb {N};

(2)

the sequence {αn=∥An+1x∥∥Anx∥:n≥1}\left\{ \alpha_{n} = \frac{\left\| A^{n+1} x\right\| }{\left\| A^{n} x\right\| }: n \geq 1\right\} converges.

Problem 9.27

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that

?
(1)

∥A∥=sup⁡x,y∈S(0,1)∣(Ax,y)∣\left\| A\right\| = \sup_{x, y \in S(0,1)}\left|(A x, y)\right|;

(2)

if A∗=AA^{*} = A, then ∥A∥=sup⁡x∈S(0,1)∣(Ax,x)∣\left\| A\right\| = \sup_{x \in S(0,1)}\left|(A x, x)\right|;

(3)

if A∗=AA^{*} = A, then A=0A = 0 if and only if (Ax,x)(A x, x) = 0 for all x∈Hx \in H.

Problem 9.28

Let HH be a complex Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that

?
(1)

the operators Re(A)=12(A+A∗),Im(A)=12i(A−A∗)\mathrm{Re}\left(A\right) = \frac{1}{2}\left(A+A^{*}\right), \mathrm{Im}\left(A\right) = \frac{1}{2 i}\left(A-A^{*}\right) are self-adjoint;

(2)

the operator AA can be uniquely represented in the form A=B+iCA = B+i C, where BB and CC are self-adjoint operators;

(3)

the operator AA is normal ⇔Re(A)⋅Im(A)=Im(A)⋅Re(A)\Leftrightarrow \mathrm{Re}\left(A\right) \cdot \mathrm{Im}\left(A\right) = \mathrm{Im}\left(A\right) \cdot \mathrm{Re}\left(A\right);

(4)

the operator AA is unitary if and only if the following conditions are fulfilled:

(a) AA is normal;

(b) (Re(A))2+(Im(A))2=I(\mathrm{Re}\left(A\right))^{2}+(\mathrm{Im}\left(A\right))^{2} = I.

Problem 9.29

Let A,BA, B be linear operators in Hilbert space HH such that (Ax,y)=(x,By),x,y∈H(A x, y) = (x, B y), x, y \in H. Prove that A∈L(H)A \in \mathcal{L}(H) and B=A∗B = A^{*}.

?
Problem 9.30

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that (R(A))⊥=Ker⁡A∗,(Ker⁡A)⊥=R(A∗)‾(R(A))^{\perp } = \operatorname {Ker} A^{*},(\operatorname {Ker} A)^{\perp } = \overline{R\left(A^{*}\right)} (here the bar denotes closure).

?
Problem 9.31

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that AA∗=A∗AA A^{*} = A^{*} A if and only if ∥Ax∥=∥A∗x∥\left\| A x\right\| = \left\| A^{*} x\right\| for all x∈Hx \in H.

?
Problem 9.32

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H) is a normal operator. Prove that

?
(1)

Ker⁡A=Ker⁡A∗=(R(A))⊥\operatorname {Ker} A = \operatorname {Ker} A^{*} = (R(A))^{\perp };

(2)

∥A2∥=∥A∥2\left\| A^{2}\right\| = \left\| A\right\|^{2};

(3)

if A2=0A^{2} = 0, then A=0A = 0;

(4)

if A−1∈L(H)A^{-1} \in \mathcal{L}(H) exists, then A−1A^{-1} is a normal operator.

Problem 9.33

Let HH be a Hilbert space, P∈L(H),P2=PP \in \mathcal{L}(H), P^{2} = P. Prove that the following statements are equivalent:

?
(1)

PP is an orthoprojector;

(2)

P=P∗P = P^{*};

(3)

PP∗=P∗PP P^{*} = P^{*} P;

(4)

Ker⁡P=(R(P))⊥\operatorname {Ker} P = (R(P))^{\perp };

(5)

(Px,x)=∥Px∥2,x∈H(P x, x) = \left\| P x\right\|^{2}, x \in H.

(6)

(Px,x)≥0,x∈H(P x, x) \geq 0, x \in H.

Problem 9.34

Let H1H_{1} and H2H_{2} be subspaces of a Hilbert space H,H0=H1∩H2,PjH, H_{0} = H_{1} \cap H_{2}, P_{j} be the orthogonal projector onto Hj,j=0,1,2H_{j}, j = 0,1,2. Denote Hi⊖H0={x∈Hi∣x⊥H0},i=1,2H_{i} \ominus H_{0} = \left\{ x \in H_{i} \mid x \perp H_{0}\right\} , i = 1,2. Prove the following statements:

?
(1)

P1+P2P_{1}+P_{2} is an orthogonal projector ⇔P1P2=0⇔H1⊥H2\Leftrightarrow P_{1} P_{2} = 0 \Leftrightarrow H_{1} \perp H_{2}; in this case P1+P2P_{1}+P_{2} is an orthogonal projector onto H1⊕H2H_{1} \oplus H_{2};

(2)

P1P2P_{1} P_{2} is an orthogonal projector ⇔P1P2=P2P1⇔\Leftrightarrow P_{1} P_{2} = P_{2} P_{1} \Leftrightarrow ⇔(H1⊖H0)⊥(H2⊖H0)\Leftrightarrow \left(H_{1} \ominus H_{0}\right) \perp \left(H_{2} \ominus H_{0}\right); in this case P1P2P_{1} P_{2} is an orthogonal projector onto H0H_{0};

(3)

P1−P2P_{1}-P_{2} is an orthogonal projector ⇔P1≥P2⇔H1⊃H2\Leftrightarrow P_{1} \geq P_{2} \Leftrightarrow H_{1} \supset H_{2}; in this case P1−P2P_{1}-P_{2} is an orthogonal projector onto H1⊖H0H_{1} \ominus H_{0};

(4)

P1P2=P2⇔P2P1=P2⇔P1≥P2⇔H1⊃H2P_{1} P_{2} = P_{2} \Leftrightarrow P_{2} P_{1} = P_{2} \Leftrightarrow P_{1} \geq P_{2} \Leftrightarrow H_{1} \supset H_{2}.

For an operator A∈L(H)A \in \mathcal{L}(H) express in terms of algebraic relations between AA and P1P_{1} the following statements:

(5)

H1H_{1} is invariant with respect to AA (i.e., A(H1)⊂H1A\left(H_{1}\right) \subset H_{1});

(6)

both H1H_{1} and H1⊥H_{1}^{\perp } are invariant with respect to AA.

Problem 9.35

Let HH be a Hilbert space and {A,An:n≥1}⊂L(H)\left\{ A, A_{n}: n \geq 1\right\} \subset \mathcal{L}(H). Prove that

?
(1)

An⇉A⇔An∗⇉A∗A_{n} \rightrightarrows A \Leftrightarrow A_{n}^{*} \rightrightarrows A^{*};

(2)

An→wA⇔An∗→wA∗A_{n} \xrightarrow {w} A \Leftrightarrow A_{n}^{*} \xrightarrow {w} A^{*};

(3)

An→sA⇔An∗→sA∗A_{n} \xrightarrow {s} A \Leftrightarrow A_{n}^{*} \xrightarrow {s} A^{*}.

Problem 9.36

Let HH be a Hilbert space and {A,An:n≥1}⊂L(H)\left\{ A, A_{n}: n \geq 1\right\} \subset \mathcal{L}(H). Prove that

?
(1)

if An=An∗,n≥1,An→wA,n→∞A_{n} = A_{n}^{*}, n \geq 1, A_{n} \xrightarrow {w} A, n \rightarrow \infty, then A=A∗A = A^{*};

(2)

if An≥0,n≥1,An→wA,n→∞A_{n} \geq 0, n \geq 1, A_{n} \xrightarrow {w} A, n \rightarrow \infty, then A≥0A \geq 0;

(3)

if An→wAA_{n} \xrightarrow {w} A and ∥Anx∥→∥Ax∥\left\| A_{n} x\right\| \rightarrow \left\| A x\right\| for all x∈Hx \in H, then An→sAA_{n} \xrightarrow {s} A; in particular, if A,An,n≥1A, A_{n}, n \geq 1, are isometric, then An→wA⇔An→sAA_{n} \xrightarrow {w} A \Leftrightarrow A_{n} \xrightarrow {s} A;

(4)

if An,n≥1A_{n}, n \geq 1, and AA are normal, An→sAA_{n} \xrightarrow {s} A, then An∗→sA∗A_{n}^{*} \xrightarrow {s} A^{*};

(5)

if AnA_{n} are isometric for each n≥1n \geq 1, and An→wAA_{n} \xrightarrow {w} A, then AA is not necessarily isometric;

(6)

if AnA_{n} are unitary for each n≥1n \geq 1, and An→wAA_{n} \xrightarrow {w} A, then AA is not necessarily unitary.

Problem 9.37

Let HH be a Hilbert space, {An:n≥1}⊂L(H)\left\{ A_{n}: n \geq 1\right\} \subset \mathcal{L}(H) are such that An≥0,n≥1A_{n} \geq 0, n \geq 1, and also there exists C>0C > 0 such that ∥An∥≤C,n≥1\left\| A_{n}\right\| \leq C, n \geq 1. Prove that

An→s0⇔An→w0⇔(Anx,x)→0 for all x∈H. A_{n} \xrightarrow {s} 0 \Leftrightarrow A_{n} \xrightarrow {w} 0 \Leftrightarrow \left(A_{n} x, x\right) \rightarrow 0 \text{ for all } x \in H.
?
Problem 9.38

Let HH be a Hilbert space, {An:n≥1}⊂L(H)\left\{ A_{n}: n \geq 1\right\} \subset \mathcal{L}(H) be a bounded sequence of self-adjoint operators such that An+1≥AnA_{n+1} \geq A_{n} for all n≥1n \geq 1. Prove that the sequence {An:n≥1}\left\{ A_{n}: n \geq 1\right\} converges strongly.

?
Problem 9.39

Let HH be a Hilbert space, PnP_{n} be orthogonal projectors onto subspaces Hn,n≥H_{n}, n \geq 1. Prove that

?
(1)

if Pn+1≥Pn,n≥1P_{n+1} \geq P_{n}, n \geq 1, then Pn→sPP_{n} \xrightarrow {s} P, where PP is an orthogonal projector onto span⁡(∪n=1∞Hn);\operatorname {span}\left(\cup_{n = 1}^{\infty } H_{n}\right) ;

(2)

if Pn+1≤Pn,n≥1P_{n+1} \leq P_{n}, n \geq 1, then Pn→sPP_{n} \xrightarrow {s} P, where PP is an orthogonal projector onto ⋂n=1∞Hn;\bigcap_{n = 1}^{\infty } H_{n} ;

(3)

if Hk⊥Hj,k≠j,k,j≥1H_{k} \perp H_{j}, k \neq j, k, j \geq 1, then the series ∑n=1∞Pn\sum_{n = 1}^{\infty } P_{n} converges strongly to the orthogonal projector onto span⁡(⋃n=1∞Hn)‾\overline{\operatorname {span}\left(\bigcup_{n = 1}^{\infty } H_{n}\right)};

(4)

if Pn→sPP_{n} \xrightarrow {s} P, then PP is an orthogonal projector;

(5)

if the sequence {Pn:n≥1}\left\{ P_{n}: n \geq 1\right\} is monotone and Pn⇉P,n→∞P_{n} \rightrightarrows P, n \rightarrow \infty, then there exists n0∈Nn_{0} \in \mathbb {N} such that Pn=P,n≥n0P_{n} = P, n \geq n_{0}. Is the monotonicity essential in this statement?

(6)

if Pn→wPP_{n} \xrightarrow {w} P, then PP is not necessarily an orthoprojector;

(7)

if Pn→wPP_{n} \xrightarrow {w} P and PP is an orthoprojector, then Pn→sPP_{n} \xrightarrow {s} P.

Problem 9.40

Let HH be a Hilbert space over the field K\mathbb {K}, and A∈L(H)A \in \mathcal{L}(H) be such that AB=BAA B = B A for all B∈L(H)B \in \mathcal{L}(H). Prove that A=γIA = \gamma I for some γ∈K\gamma \in \mathbb {K}.

?
Problem 9.41

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H) be a self-adjoint operator, λ∈C\lambda \in \mathbb {C}. Prove that the operator A−λIA-\lambda I is continuously invertible if and only if there exists c>0c > 0 such that ∥(A−λI)x∥≥c∥x∥,x∈H\left\| (A-\lambda I) x\right\| \geq c\left\| x\right\| , x \in H.

?
Problem 9.42

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that

?
(1)

if A≥0A \geq 0, then A−1∈L(H)A^{-1} \in \mathcal{L}(H) exists if and only if there exists λ>0\lambda > 0 such that A≥λI;A \geq \lambda I ;

(2)

there exists (I+AA∗)−1∈L(H)\left(I+A A^{*}\right)^{-1} \in \mathcal{L}(H);

(3)

if A∗=A,m=inf⁡x∈S(0,1)(Ax,x),M=sup⁡x∈S(0,1)(Ax,x)A^{*} = A, m = \inf_{x \in S(0,1)}(A x, x), M = \sup_{x \in S(0,1)}(A x, x), then the operator A−λIA-\lambda I is continuously invertible for all λ∈R\[m,M]\lambda \in \mathbb {R} \backslash [m, M] and is not continuously invertible for λ∈{m,M}\lambda \in \left\{ m, M\right\};

(4)

if A∗=A,λ=α+βi∈CA^{*} = A, \lambda = \alpha +\beta i \in \mathbb {C}, then

∥(A−λI)x∥≥∣β∣⋅∥x∥,x∈H \left\| (A-\lambda I) x\right\| \geq \left|\beta \right| \cdot \left\| x\right\| , x \in H
(5)

if A∗=A,λ∈C\RA^{*} = A, \lambda \in \mathbb {C} \backslash \mathbb {R}, then there exists (A−λI)−1∈L(H)(A-\lambda I)^{-1} \in \mathcal{L}(H).

Problem 9.43

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that operator AA is continuously invertible if and only if there exists c>0c > 0 such that AA∗≥cIA A^{*} \geq c I and A∗A≥cIA^{*} A \geq c I.

?
Problem 9.44

Let HH be a Hilbert space and A∈L(H)A \in \mathcal{L}(H) be a self-adjoint operator. Prove that

?
(1)

AnA^{n} is a self-adjoint operator for every n≥1n \geq 1;

(2)

∥An∥=∥A∥n,n≥1\left\| A^{n}\right\| = \left\| A\right\|^{n}, n \geq 1;

(3)

if A≥0A \geq 0, then An≥0A^{n} \geq 0 for every n≥1n \geq 1.

Problem 9.45

Let HH be a Hilbert space and A∈L(H)A \in \mathcal{L}(H) be a self-adjoint operator. Prove that the series ∑n=0∞An\sum_{n = 0}^{\infty } A^{n} converges in L(H)\mathcal{L}(H) if and only if ∥A∥<1\left\| A\right\| < 1.

?
Problem 9.46

Let AA be an operator in l2l_{2}, which is represented by a Jacobi matrix, i.e., a matrix of the form

A=[α1β1000…γ1α2β200…0γ2α3β30………………], where {αj,βj,γj:j≥1}⊂K. \mathcal{A} =\left[\begin{smallmatrix} \alpha _{1} & \beta _{1} & 0 & 0 & 0 & \ldots \\ \gamma _{1} & \alpha _{2} & \beta _{2} & 0 & 0 & \ldots \\ 0 & \gamma _{2} & \alpha _{3} & \beta _{3} & 0 & \ldots \\ \ldots & \ldots & \ldots & \ldots & \ldots \end{smallmatrix}\right], \text{ where }\left\{ \alpha _{j}, \beta _{j}, \gamma _{j}: j \geq 1\right\} \subset \mathbb {K}.

Prove that the operator AA is bounded if and only if

sup⁡k≥1(∣αk∣2+∣βk∣2+∣γk∣2)<+∞ \sup _{k \geq 1}\left(\left|\alpha _{k}\right|^{2}+\left|\beta _{k}\right|^{2}+\left|\gamma _{k}\right|^{2}\right) < +\infty

Under what conditions on A\mathcal{A} is the operator AA self-adjoint? Consider the cases K=R\mathbb {K} = \mathbb {R} and K=C\mathbb {K} = \mathbb {C}.

?
Problem 9.47

Prove that

?
(1)

if {Pn:n≥0}\left\{ P_{n}: n \geq 0\right\} are Legendre polynomials (see item (2) of problem (2.38)), then

(n+1)Pn+1(t)−(2n+1)tPn(t)+nPn−1(t)=0,t∈[−1,1],n≥1 (n+1) P_{n+1}(t)-(2 n+1) t P_{n}(t)+n P_{n-1}(t) = 0, t \in [-1,1], n \geq 1
(2)

the operator (Ax)(t)=tx(t),t∈[−1,1],x∈L2([−1,1])(A x)(t) = t x(t), t \in [-1,1], x \in L_{2}(\left[-1,1\right]), is represented by a Jacobi matrix in the basis which consists of Legendre polynomials.

Problem 9.48

Let {en:n≥1}\left\{ e_{n}: n \geq 1\right\} be an orthonormal basis in a Hilbert space HH. Prove that a matrix (ajk)j,k=1∞,ajk∈C,j≥1,k≥1\left(a_{j k}\right)_{j, k = 1}^{\infty }, a_{j k} \in \mathbb {C}, j \geq 1, k \geq 1, represents a continuous linear operator in HH if and only if the following conditions are fulfilled:

(a) the series ∑k=1∞ajkxk\sum_{k = 1}^{\infty } a_{j k} x_{k} converges for any j∈Nj \in \mathbb {N} and x=∑k=1∞xkek∈Hx = \sum_{k = 1}^{\infty } x_{k} e_{k} \in H;

(b) ∃C>0∀x∈H:∑j=1∞∣∑k=1∞ajkxk∣2≤C∥x∥2\exists C > 0 \forall x \in H: \sum_{j = 1}^{\infty }\left|\sum_{k = 1}^{\infty } a_{j k} x_{k}\right|^{2} \leq C\left\| x\right\|^{2}.

?
Problem 9.49

Prove that any of the following two conditions is necessary and sufficient for the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty } to represent an operator A∈L(H)A \in \mathcal{L}(H) in a given orthonormal basis:

?
(1)

∃C>0∀m,n∈N∀x1,…,xn,y1,…,ym∈C\exists C > 0 \forall m, n \in \mathbb {N} \forall x_{1}, \ldots , x_{n}, y_{1}, \ldots , y_{m} \in \mathbb {C} :

∣∑j=1n∑k=1majkxjyk‾∣≤C(∑j=1n∣xj∣2)12(∑k=1m∣yk∣2)12 \left|\sum _{j = 1}^{n} \sum _{k = 1}^{m} a_{j k} x_{j} \overline{y_{k}}\right| \leq C\left(\sum _{j = 1}^{n}\left|x_{j}\right|^{2}\right)^{\frac{1}{2}}\left(\sum _{k = 1}^{m}\left|y_{k}\right|^{2}\right)^{\frac{1}{2}}
(2)

∃C≥0∀m,n∈N∀x1,…,xm∈C\exists C \geq 0 \forall m, n \in \mathbb {N} \forall x_{1}, \ldots , x_{m} \in \mathbb {C} :

∑j=1n∣∑k=1majkxk∣2≤C∑k=1m∣xk∣2. \sum _{j = 1}^{n}\left|\sum _{k = 1}^{m} a_{j k} x_{k}\right|^{2} \leq C \sum _{k = 1}^{m}\left|x_{k}\right|^{2}.

Prove that condition (2) is fulfilled in each of the following cases:

(a) ∑j=1∞∑k=1∞∣ajk∣2<+∞\sum_{j = 1}^{\infty } \sum_{k = 1}^{\infty }\left|a_{j k}\right|^{2} < +\infty;

(b) ajk=δjkaj,sup⁡j≥1∣aj∣<+∞a_{j k} = \delta_{j k} a_{j}, \sup_{j \geq 1}\left|a_{j}\right| < +\infty.

Problem 9.50

Let (ajk)j,k=1∞,(bjk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty },\left(b_{j k}\right)_{j, k = 1}^{\infty } be the matrices which represent operators A,B∈L(H)A, B \in \mathcal{L}(H) in an orthonormal basis {en:n≥1}\left\{ e_{n}: n \geq 1\right\}. Find the matrices that represent the operators A+B,ABA+B, A B.

?
Problem 9.51

Let the operator A∈L(H)A \in \mathcal{L}(H) be represented by the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty } in some orthonormal basis of the complex Hilbert space HH. Prove that

?
(1)

AA is self-adjoint ⇔\Leftrightarrow the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty } is Hermitian;

(2)

A≥0⇔A \geq 0 \Leftrightarrow the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty } is Hermitian and for every n∈Nn \in \mathbb {N} the matrix (ajk)j,k=1n\left(a_{j k}\right)_{j, k = 1}^{n} is nonnegatively definite;

(3)

AA is an orthogonal projector ⇔\Leftrightarrow the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty } is Hermitian, and ajk=∑i=1∞ajiaik,j,k∈N;a_{j k} = \sum_{i = 1}^{\infty } a_{j i} a_{i k}, j, k \in \mathbb {N} ;

(4)

AA is isometric ⇔{(a1k,a2k,…,ajk,…):k≥1}\Leftrightarrow \left\{ \left(a_{1 k}, a_{2 k}, \ldots , a_{j k}, \ldots \right): k \geq 1\right\} is an orthonormal system in l2l_{2}.

Find the necessary and sufficient conditions for the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty } to represent a unitary operator in HH.

Problem 9.52
?
(1)

Let S∈L(H)S \in \mathcal{L}(H) be a bounded operator in a Hilbert space HH such that ∥S∥≤1\left\| S\right\| \leq 1. Put T1=0,Tn=12(S+Tn−12),n≥2T_{1} = 0, T_{n} = \frac{1}{2}\left(S+T_{n-1}^{2}\right), n \geq 2. Prove that

(a) for every n≥2n \geq 2 the operators TnT_{n} and Tn−Tn−1T_{n}-T_{n-1} are polynomials of SS with nonnegative coefficients;

(b) 0≤Tn−1≤Tn0 \leq T_{n-1} \leq T_{n} and ∥Tn∥≤1\left\| T_{n}\right\| \leq 1 for every n≥2n \geq 2;

(c) there exists T∈L(H)T \in \mathcal{L}(H) such that Tn→sTT_{n} \xrightarrow {s} T, moreover T≥0,∥T∥≤1,Tn2→sT2T \geq 0,\left\| T\right\| \leq 1, T_{n}^{2} \xrightarrow {s} T^{2} and (I−T)2=I−S(I-T)^{2} = I-S;

(d) TT commutes with all operators that commute with SS.

(2)

Let A∈L(H)A \in \mathcal{L}(H) be a nonnegative operator in a Hilbert space HH. Prove that there exists B∈L(H)B \in \mathcal{L}(H) such that B≥0B \geq 0 and B2=AB^{2} = A (the operator BB is called the square root of the operator AA and is denoted B=AB = \sqrt{A} or B=A12B = A^{\frac{1}{2}}), moreover BB commutes with all operators that commute with AA.

(3)

Prove the uniqueness of the square root BB of the nonnegative operator AA.

Problem 9.53

For an operator A∈L(H)A \in \mathcal{L}(H) prove that A≥0A \geq 0 and find A\sqrt{A}.

?
(1)

H=l2,Ax=(α1x1,α2x2,…)H = l_{2}, A x = \left(\alpha_{1} x_{1}, \alpha_{2} x_{2}, \ldots \right), where αk≥0,k≥1\alpha_{k} \geq 0, k \geq 1;

(2)

H=l2,Ax=(x1,…,xj,0,0,…)H = l_{2}, A x = \left(x_{1}, \ldots , x_{j}, 0,0, \ldots \right), where j∈Nj \in \mathbb {N} is fixed,

(3)

H=l2,Ax=(2x1+3x2,3x1+5x2,x3,0,0,…)H = l_{2}, A x = \left(2 x_{1}+3 x_{2}, 3 x_{1}+5 x_{2}, x_{3}, 0,0, \ldots \right);

(4)

H=L2(R),(Ax)(t)=χ[−α,α](t)x(t)H = L_{2}(\mathbb {R}),(A x)(t) = \chi_{\left[-\alpha , \alpha \right]}(t) x(t), where α>0\alpha > 0;

(5)

H=L2([0,1]),(Ax)(t)=tx(t)H = L_{2}(\left[0,1\right]),(A x)(t) = t x(t);

(6)

H=L2(R),(Ax)(t)=12(x(t)+x(−t))H = L_{2}(\mathbb {R}),(A x)(t) = \frac{1}{2}(x(t)+x(-t));

(7)

H=L2([0,1]),(Ax)(t)=∫01et+τx(τ)dτH = L_{2}(\left[0,1\right]),(A x)(t) = \int_{0}^{1} e^{t+\tau } x(\tau ) d \tau.

Problem 9.54

Prove that there does not exist an operator B∈L(l2)B \in \mathcal{L}\left(l_{2}\right) such that B2=AB^{2} = A if

?
(1)

Ax=(x2,x3,…),x∈l2A x = \left(x_{2}, x_{3}, \ldots \right), x \in l_{2};

(2)

Ax=(0,x1,x2,…),x∈l2A x = \left(0, x_{1}, x_{2}, \ldots \right), x \in l_{2}.

Problem 9.55

Let HH be a Hilbert space, A,B,C,D∈L(H)A, B, C, D \in \mathcal{L}(H) are nonnegative operators that pairwise commute. Prove that

?
(1)

AB≥0A B \geq 0;

(2)

if A≤BA \leq B, then A2≤B2A^{2} \leq B^{2};

(3)

if A≤BA \leq B, then AC≤BCA C \leq B C;

(4)

if A≤BA \leq B and C≤DC \leq D, then AC≤BDA C \leq B D.

Problem 9.56

Let AA be a nonnegative operator in a Hilbert space HH. Prove that ∥A∥=∥A∥\left\| \sqrt{A}\right\| = \sqrt{\left\| A\right\| }.

?
Problem 9.57

Let AA be a nonnegative operator in a Hilbert space HH. Prove that the following statements are equivalent:

?
(1)

R(A)R(A) is dense in HH;

(2)

Ker⁡A={0}\operatorname {Ker} A = \left\{ 0\right\};

(3)

(Ax,x)>0(A x, x) > 0 for every x∈H,x≠0x \in H, x \neq 0.

Problem 9.58

Let HH be a Hilbert space, A,B∈L(H)A, B \in \mathcal{L}(H). Is it correct that

?
(1)

if AB=BA,A≥0,AB≥0A B = B A, A \geq 0, A B \geq 0, then B≥0B \geq 0;

(2)

if A2=AA^{2} = A, then A=0A = 0 or A=IA = I;

(3)

if AB=BA\sqrt{A} B = B \sqrt{A}, then AB=BAA B = B A?

Problem 9.59

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that there exists a unique nonnegative operator B∈L(H)B \in \mathcal{L}(H) such that ∥Ax∥=∥Bx∥,x∈H\left\| A x\right\| = \left\| B x\right\| , x \in H.

?
Problem 9.60

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that

?
(1)

if 0≤A≤I0 \leq A \leq I, then A≤A≤IA \leq \sqrt{A} \leq I;

(2)

if I≤AI \leq A, then I≤A≤AI \leq \sqrt{A} \leq A.

Problem 9.61

Let HH be a Hilbert space, A,B∈L(H),I≤A≤BA, B \in \mathcal{L}(H), I \leq A \leq B. Prove that the operators A,BA, B are continuously invertible and B−1≤A−1B^{-1} \leq A^{-1}.

?
Problem 9.62

Let HH be a Hilbert space, A∈L(H),∣A∣=A∗AA \in \mathcal{L}(H),\left|A\right| = \sqrt{A^{*} A}. Prove that

?
(1)

∣A∣=0⇔A=0\left|A\right| = 0 \Leftrightarrow A = 0;

(2)

A≥0⇔∣A∣=AA \geq 0 \Leftrightarrow \left|A\right| = A;

(3)

the norms of the operators ∣A∣\left|A\right| and AA are equal;

(4)

Ker⁡∣A∣=Ker⁡A,R(∣A∣)‾=R(A∗)‾\operatorname {Ker}\left|A\right| = \operatorname {Ker} A, \overline{R(\left|A\right|)} = \overline{R\left(A^{*}\right)}.

Problem 9.63

Let A∈L(H)A \in \mathcal{L}(H) be a self-adjoint operator, ∣A∣=A2\left|A\right| = \sqrt{A^{2}}. We put A+=12(∣A∣+A^{+} = \frac{1}{2}(\left|A\right|+ A), A−=12(∣A∣−A)A^{-} = \frac{1}{2}(\left|A\right|-A). Prove that

?
(1)

A+A−=A−A+=0A^{+} A^{-} = A^{-} A^{+} = 0;

(2)

A−=0⇔A=A+=∣A∣A^{-} = 0 \Leftrightarrow A = A^{+} = \left|A\right|;

(3)

A+=0⇔A=−A−=−∣A∣A^{+} = 0 \Leftrightarrow A = -A^{-} = -\left|A\right|.

Problem 9.64

Let HH be a Hilbert space and operator A∈L(H)A \in \mathcal{L}(H) be continuously invertible. Prove that

?
(1)

operator ∣A∣\left|A\right| is continuously invertible;

(2)

operator U=A∣A∣−1U = A\left|A\right|^{-1} is unitary.

Thus there is a representation A=U∣A∣A = U\left|A\right|, which is called the polar decomposition of AA (in the case of H=CH = \mathbb {C} this decomposition is reduced to the representation of a complex number aa in the form a=reiφa = r e^{i \varphi }).

Problem 9.65

Let HH be a Hilbert space. An operator U∈L(H)U \in \mathcal{L}(H) is called partially isometric if ∥Ux∥=∥x∥,x∈(Ker⁡U)⊥\left\| U x\right\| = \left\| x\right\| , x \in (\operatorname {Ker} U)^{\perp }.

?
(1)

Prove that the operator UU is partially isometric if and only if U∗UU^{*} U is an orthogonal projector on H1=(Ker⁡U)⊥H_{1} = (\operatorname {Ker} U)^{\perp }.

(2)

Let the operator UU be partially isometric. Prove that the set H2=R(U)H_{2} = R(U) is closed, UU∗U U^{*} is an orthoprojector on H2H_{2} and the operator U∗U^{*} is also partially isometric.

Problem 9.66

Let HH be a Hilbert space, A∈L(H)A \in \mathcal{L}(H). Representation A=UBA = U B, where operator B∈L(H)B \in \mathcal{L}(H) is nonnegative, operator U∈L(H)U \in \mathcal{L}(H) is partially isometric and Ker⁡U=Ker⁡B\operatorname {Ker} U = \operatorname {Ker} B, is called the polar decomposition of AA (compare with problem 9.64). Prove that the polar decomposition exists and is unique, moreover B=∣A∣B = \left|A\right|.

?
Problem 9.67

Let A=U∣A∣A = U\left|A\right| be the polar decomposition of an operator A∈L(H)A \in \mathcal{L}(H). Prove that ∣A∗∣=U∣A∣U∗,A∗=U∗∣A∗∣,A=∣A∗∣U,A∗=∣A∣U∗\left|A^{*}\right| = U\left|A\right| U^{*}, A^{*} = U^{*}\left|A^{*}\right|, A = \left|A^{*}\right| U, A^{*} = \left|A\right| U^{*}.

?
Problem 9.68

Find the polar decomposition of an operator A∈L(l2)A \in \mathcal{L}\left(l_{2}\right).

?
(1)

Ax=(α1x1,α2x2,…)A x = \left(\alpha_{1} x_{1}, \alpha_{2} x_{2}, \ldots \right), where {αn:n≥1}⊂C\left\{ \alpha_{n}: n \geq 1\right\} \subset \mathbb {C} are given numbers;

(2)

Ax=(x2,x3,…)A x = \left(x_{2}, x_{3}, \ldots \right);

(3)

Ax=(x1,…,xj,0,0,…)A x = \left(x_{1}, \ldots , x_{j}, 0,0, \ldots \right), where j∈Nj \in \mathbb {N} is fixed.

Problem 9.69

Let A=U∣A∣A = U\left|A\right| be the polar decomposition of an operator A∈L(H)A \in \mathcal{L}(H).

?
(1)

Under what conditions on AA the operator UU will be

(a) isometric;

(b) unitary?

(2)

Let operator AA be normal. Prove that the operators ∣A∣\left|A\right| and UU commute.

(3)

Let C∈L(H)C \in \mathcal{L}(H) be a unitary operator such that AC=CAA C = C A. Prove that ∣A∣\left|A\right| and UU commute with CC.

(4)

Is the statement of item (3) correct if CC is not necessarily a unitary operator?

Problem 9.70

Let HH be a complex Hilbert space, A∈L(H)A \in \mathcal{L}(H). Prove that

?
(1)

if AA is self-adjoint, then the operator U=(A+λI)(A+λˉI)−1,λ∈C\RU = (A+\lambda I)(A+\bar{\lambda } I)^{-1}, \lambda \in \mathbb {C} \backslash \mathbb {R}, is unitary;

(2)

if (A−iI)−1∈L(H)(A-i I)^{-1} \in \mathcal{L}(H) exists and the operator U=(A+iI)(A−iI)−1U = (A+i I)(A-i I)^{-1} is unitary, then AA is self-adjoint;

(3)

if the operator U∈L(H)U \in \mathcal{L}(H) is unitary and (U−I)−1∈L(H)(U-I)^{-1} \in \mathcal{L}(H) exists, then the operator A=i(U+I)(U−I)−1A = i(U+I)(U-I)^{-1} is self-adjoint.

Problem 9.71

Prove that

?
(1)

if (T,μ)(T, \mu ) is a space with a finite measure μ\mu and AA is an integral operator in Lp(T,μ),1<p<+∞L_{p}(T, \mu ), 1 < p < +\infty, with a kernel K∈L∞(T×T,μ×μ)K \in L_{\infty }(T \times T, \mu \times \mu ), i.e.

(Ax)(t)=∫TK(t,s)x(s)dμ(s),t∈T,x∈Lp(T,μ) (A x)(t) = \int _{T} K(t, s) x(s) d \mu (s), t \in T, x \in L_{p}(T, \mu )

then A′A^{\prime } is an integral operator in Lq(T,μ)L_{q}(T, \mu ) with the kernel K~(t,s)=K(s,t),(t,s)∈T×T\widetilde{K}(t, s) = K(s, t),(t, s) \in T \times T;

(2)

if A∈L(lp),1<p<+∞A \in \mathcal{L}\left(l_{p}\right), 1 < p < +\infty, is represented by the matrix (ajk)j,k=1∞\left(a_{j k}\right)_{j, k = 1}^{\infty }, then A′∈L(lq)A^{\prime } \in \mathcal{L}\left(l_{q}\right) is represented by the matrix (a~jk)j,k=1∞\left(\widetilde{a}_{j k}\right)_{j, k = 1}^{\infty }, where a~jk=akj,j,k∈N\widetilde{a}_{j k} = a_{k j}, j, k \in \mathbb {N}.

Problem 9.72

Find the adjoint operators to operators AA from problem 9.15∘9.15^{\circ }, items (1)-(11), if A:lp→lp,1<p<+∞A: l_{p} \rightarrow l_{p}, 1 < p < +\infty.

?
Problem 9.73

Find the adjoint operator to the operator AA defined by the formula Ax=(x2,x3,…)A x = \left(x_{2}, x_{3}, \ldots \right), if

?
(1)

A:c0→c0A: c_{0} \rightarrow c_{0};

(2)

A:l1→l2A: l_{1} \rightarrow l_{2};

(3)

A:l1→c0A: l_{1} \rightarrow c_{0}.

Problem 9.74

Find the adjoint operator to an operator AA if

?
(1)

A:l1→l2A: l_{1} \rightarrow l_{2} is the embedding operator: Ax=x,x∈l1A x = x, x \in l_{1};

(2)

A:C([0,2])→C([0,1])A: C(\left[0,2\right]) \rightarrow C(\left[0,1\right]) is the restriction operator: (Ax)(t)=x(t),t∈[0,1],x∈C([0,2]);(A x)(t) = x(t), t \in \left[0,1\right], x \in C(\left[0,2\right]) ;

(3)

A:Lp([0,2])→Lp([0,1]),1≤p<+∞A: L_{p}(\left[0,2\right]) \rightarrow L_{p}(\left[0,1\right]), 1 \leq p < +\infty, is the restriction operator;

(4)

A:C([0,1])→C([0,1]),(Ax)(t)=a(t)x(t)A: C(\left[0,1\right]) \rightarrow C(\left[0,1\right]),(A x)(t) = a(t) x(t), where a∈C([0,1])a \in C(\left[0,1\right]) is a fixed function;

(5)

A:C([0,1])→C([0,1]),(Ax)(t)=∫0tx(s)dsA: C(\left[0,1\right]) \rightarrow C(\left[0,1\right]),(A x)(t) = \int_{0}^{t} x(s) d s;

(6)

X,YX, Y are normed linear spaces, f∈X∗f \in X^{*} and y∈Yy \in Y are fixed, Ax=f(x)y,x∈XA x = f(x) y, x \in X.

Problem 9.75

Let X1,X2,X3X_{1}, X_{2}, X_{3} be normed linear spaces over the field K,A∈L(X1,X2)\mathbb {K}, A \in \mathcal{L}\left(X_{1}, X_{2}\right). Prove that

?
(1)

if B∈L(X1,X2),α,β∈KB \in \mathcal{L}\left(X_{1}, X_{2}\right), \alpha , \beta \in \mathbb {K}, then (αA+βB)′=αA′+βB′(\alpha A+\beta B)^{\prime } = \alpha A^{\prime }+\beta B^{\prime };

(2)

if B∈L(X2,X3)B \in \mathcal{L}\left(X_{2}, X_{3}\right), then (AB)′=B′A′(A B)^{\prime } = B^{\prime } A^{\prime };

(3)

the restriction of the operator A′′∈L(X1∗∗,X2∗∗)A^{\prime \prime } \in \mathcal{L}\left(X_{1}^{* *}, X_{2}^{* *}\right) to X1X_{1} coincides with AA;

(4)

if the spaces X1X_{1} and X2X_{2} are reflexive, then A′′=AA^{\prime \prime } = A;

(5)

if AA is continuously invertible, then A′A^{\prime } is also continuously invertible and (A′)−1=(A−1)′\left(A^{\prime }\right)^{-1} = \left(A^{-1}\right)^{\prime }.

Problem 9.76

Let X1,X2X_{1}, X_{2} be Banach spaces over the field K,A∈L(X1,X2)\mathbb {K}, A \in \mathcal{L}\left(X_{1}, X_{2}\right). Prove that

?
(1)

Ker⁡A={x∈X1∣∀f∈R(A′):f(x)=0}\operatorname {Ker} A = \left\{ x \in X_{1} \mid \forall f \in R\left(A^{\prime }\right): f(x) = 0\right\};

(2)

the set R(A)‾={y∈X2∣∀f∈Ker⁡A′:f(y)=0}\overline{R(A)} = \left\{ y \in X_{2} \mid \forall f \in \operatorname {Ker} A^{\prime }: f(y) = 0\right\};

(3)

the operator AA is continuously invertible if and only if

∃c>0∀x∈X1∀f∈X2∗:∥Ax∥2≥c∥x∥1,∥A′f∥≥c∥f∥; \exists c > 0 \forall x \in X_{1} \forall f \in X_{2}^{*}:\left\| A x\right\| _{2} \geq c\left\| x\right\| _{1},\left\| A^{\prime } f\right\| \geq c\left\| f\right\| ;
(4)

the operator AA is continuously invertible if and only if the operator A′A^{\prime } is continuously invertible.