Chapter 20

Continuous Linear Operators on Hilbert Spaces

[66/0%]
Show
LaTeX
Problem 20.1

Let [a,b][a, b] be a closed, bounded interval of real numbers. Show that the L2[a,b]L^{2}[a, b] inner-product is also an inner-product on C[a,b]C[a, b]. Is C[a,b]C[a, b], considered as an inner-product space with the L2[a,b]L^{2}[a, b] inner-product, a Hilbert space?

?
Problem 20.2

Show that the maximum norm on C[a,b]C[a, b] is not induced by an inner-product and neither is the usual norm on ℓ1\ell^{1}.

?
Problem 20.3

Let H1H_{1} and H2H_{2} be Hilbert spaces. Show that the Cartesian product H1×H2H_{1} \times H_{2} also is a Hilbert space with an inner-product with respect to which H1×{0}=[{0}×H2]⊥H_{1} \times \left\{ 0\right\} = \left[\left\{ 0\right\} \times H_{2}\right]^{\perp }.

?
Problem 20.4

Show that if SS is a subset of an inner-product space HH, then S⊥S^{\perp } is a closed subspace of HH.

?
Problem 20.5

Let SS be a subset of HH. Show that S=(S⊥)⊥S = \left(S^{\perp }\right)^{\perp } if and only if SS is a closed subspace of HH.

?
Problem 20.6

Show that for any two vectors u,v∈Hu, v \in H,

⟨u,v⟩=14[∥u+v∥2−∥u−v∥2]. \left\langle u, v \right\rangle = \frac{1}{4}\left[\left\| u+v\right\| ^{2}-\left\| u-v\right\| ^{2}\right] .
?
Problem 20.7

Let XX be a linear space normed by ∥⋅∥\left\| \cdot \right\|. Use the polarization identity to show that a norm ∥⋅∥\left\| \cdot \right\| is induced by an inner-product if and only if the parallelogram identity holds.

?
Problem 20.8

Let VV be a closed subspace of HH and PP a projection of HH onto VV. Show that PP is the orthogonal projection of HH onto VV if and only if (4) holds.

?
Problem 20.9

Let TT belong to L(H)\mathcal{L}(H). Show that TT is an isometry if and only if

⟨T(u),T(v)⟩=⟨u,v⟩ for all u,v∈H. \left\langle T(u), T(v) \right\rangle = \left\langle u, v \right\rangle \text{ for all } u, v \in H .
?
Problem 20.10

Let VV be a finite dimensional subspace of HH and φ1,…,φn\varphi_{1}, \ldots , \varphi_{n} a basis for VV consisting of unit vectors, each pair of which is orthogonal. Show that the orthogonal projection PP of HH onto VV is given by

P(h)=∑k=1n⟨h,φk⟩φk for all h∈V. P(h) = \sum _{k = 1}^{n} \left\langle h, \varphi _{k} \right\rangle \varphi _{k} \text{ for all } h \in V .
?
Problem 20.11

For hh a vector in HH, show that the function u↦⟨h,u⟩u \mapsto \left\langle h, u \right\rangle belongs to H∗H^{*}.

?
Problem 20.12

For any vector h∈Hh \in H, show that there is a bounded linear functional ψ∈H∗\psi \in H^{*} for which

∥ψ∥=1 and ψ(h)=∥h∥. \left\| \psi \right\| = 1 \text{ and } \psi (h) = \left\| h\right\| .
?
Problem 20.13

Let VV be a closed subspace of HH and PP the orthogonal projection of HH onto VV. For any normed linear space XX and T∈L(V,X)T \in \mathcal{L}(V, X), show that T∘PT \circ P belongs to L(H,X)\mathcal{L}(H, X), and is an extension of T:V→XT: V \rightarrow X for which ∥T∘P∥=∥T∥\left\| T \circ P\right\| = \left\| T\right\|.

?
Problem 20.14

Prove the Hyperplane Separation Theorem for HH, considered as a locally convex topological vector space with respect to the strong topology, by directly using Proposition 1.

Proposition 1: Let KK be a non-empty, closed, convex subset of a Hilbert space HH and h0h_{0} be a vector in HH that lies outside of KK. Then there is exactly one vector h∗∈Kh^{*} \in K that is closest to h0h_{0}, in the sense that ∥h0−h∗∥=dist⁡K(h0)=inf⁡h∈K∥h0−h∥\left\| h_{0} - h^{*}\right\| = \operatorname {dist}_{K}(h_{0}) = \inf_{h \in K} \left\| h_{0} - h\right\|.

?
Problem 20.15

Use Proposition 1 to prove the Krein-Milman Lemma in a Hilbert space.

Proposition 1: Let KK be a non-empty, closed, convex subset of a Hilbert space HH and h0h_{0} be a vector in HH that lies outside of KK. Then there is exactly one vector h∗∈Kh^{*} \in K that is closest to h0h_{0}, in the sense that ∥h0−h∗∥=dist⁡K(h0)=inf⁡h∈K∥h0−h∥\left\| h_{0} - h^{*}\right\| = \operatorname {dist}_{K}(h_{0}) = \inf_{h \in K} \left\| h_{0} - h\right\|.

?
Problem 20.16

Show that an orthonormal subset of a separable Hilbert space HH must be countable.

?
Problem 20.17

Let {φk}\left\{ \varphi_{k}\right\} be an orthonormal sequence in a Hilbert space HH. Show that {φk}\left\{ \varphi_{k}\right\} converges weakly to 0 in HH.

?
Problem 20.18

Let {φk}\left\{ \varphi_{k}\right\} be an orthonormal basis for the separable Hilbert space HH. Show that {un}→u\left\{ u_{n}\right\} \rightarrow u weakly if and only if {un}\left\{ u_{n}\right\} is bounded and, for each k,lim⁡n→∞⟨un,φk⟩=⟨u,φk⟩k, \lim_{n \rightarrow \infty } \left\langle u_{n}, \varphi_{k} \right\rangle = \left\langle u, \varphi_{k} \right\rangle.

?
Problem 20.19

Show that any two infinite dimensional separable Hilbert spaces are isometrically isomorphic and that any such isomorphism preserves the inner-product.

?
Problem 20.20

Let VV a closed, separable subspace of HH for which {φk}\left\{ \varphi_{k}\right\} is an orthonormal basis. Show that the orthogonal projection of HH onto V,PV, P, is given by

P(h)=∑k=1∞⟨φk,h⟩φk for all h∈H. P(h) = \sum _{k = 1}^{\infty } \left\langle \varphi _{k}, h \right\rangle \varphi _{k} \text{ for all } h \in H .
?
Problem 20.21

Let {φk}\left\{ \varphi_{k}\right\} be an orthonormal basis for a Hilbert space HH. Verify that

∥h∥2=∑k=1∞⟨φk,h⟩2 for all h∈H. \left\| h\right\| ^{2} = \sum _{k = 1}^{\infty } \left\langle \varphi _{k}, h \right\rangle ^{2} \text{ for all } h \in H .

Also verify that

⟨u,v⟩=∑k=1∞ak⋅bk for all u,v∈H, \left\langle u, v \right\rangle = \sum _{k = 1}^{\infty } a_{k} \cdot b_{k} \text{ for all } u, v \in H,

where, for each natural number k,ak=⟨u,φk⟩k, a_{k} = \left\langle u, \varphi_{k} \right\rangle and bk=⟨v,φk⟩b_{k} = \left\langle v, \varphi_{k} \right\rangle.

?
Problem 20.22

Verify the assertions in the example of the orthonormal basis for L2[0,2π]L^{2}[0,2 \pi ].

?
Problem 20.23

Use Proposition 5 and the Stone-Weierstrass Theorem to show that for each f∈L2[−π,π]f \in L^{2}[-\pi , \pi ],

f(x)=a0/2+∑k=1∞[ak⋅cos⁡kx+bk⋅sin⁡kx], f(x) = a_{0} / 2+\sum _{k = 1}^{\infty }\left[a_{k} \cdot \cos k x+b_{k} \cdot \sin k x\right],

where the convergence is in L2[−π,π]L^{2}[-\pi , \pi ] and each

ak=1π∫−ππf(x)cos⁡kxdx and bk=1π∫−ππf(x)sin⁡kxdx. a_{k} = \frac{1}{\pi } \int _{-\pi }^{\pi } f(x) \cos k x d x \text{ and } b_{k} = \frac{1}{\pi } \int _{-\pi }^{\pi } f(x) \sin k x d x .

Proposition 5: An orthonormal sequence {φk}\left\{ \varphi_{k}\right\} in a Hilbert space HH is complete if and only if it is an orthonormal basis.

?
Problem 20.24

Show that neither ℓ1,ℓ∞,L1[a,b]\ell^{1}, \ell^{\infty }, L^{1}[a, b] nor L∞[a,b]L^{\infty }[a, b] is Hilbertable.

?
Problem 20.25

Let HH be an inner-product space. Show that since HH is a dense subset of a Banach space XX whose norm restricts to the norm induced by the inner-product on HH, the inner-product on HH extends to XX and induces the norm on XX. Therefore, inner-product spaces have Hilbert space completions.

?
Problem 20.26

The proof is based on the RieszFréchet Representation Theorem. Let (X,M,μX, \mathcal{M}, \mu) be a finite measure space, and ν:M→[0,∞)\nu : \mathcal{M} \rightarrow [0, \infty ) be a finite measure that is absolutely continuous with respect to μ\mu. Fill in the following outline of the proof that there is a non-negative measurable function h:X→[0,∞)h: X \rightarrow [0, \infty ) for which ν(E)=∫Ehdμ\nu (E) = \int_{E} h d \mu for all E∈ME \in \mathcal{M}.

?
(i)

Define the measure λ=μ+ν\lambda = \mu +\nu on the σ\sigma-algebra M\mathcal{M} and the functional ψ\psi on L2(X,λ)L^{2}(X, \lambda ) by

ψ(f)=∫Xfdμ for all f∈L2(X,λ) \psi (f) = \int _{X} f d \mu \text{ for all } f \in L^{2}(X, \lambda )

Show that, since λ(X)<∞,ψ\lambda (X) < \infty , \psi is a bounded linear functional on L2(X,λ)L^{2}(X, \lambda ).

(ii)

By the Riesz-Fréchet Representation Theorem for the Hilbert space L2(X,λ)L^{2}(X, \lambda ), there is a function g∈L2(X,λ)g \in L^{2}(X, \lambda ) such that

∫Xfdμ=∫Xf⋅gdλ for all f∈L2(X,λ) \int _{X} f d \mu = \int _{X} f \cdot g d \lambda \text{ for all } f \in L^{2}(X, \lambda )

and therefore

∫Xfdμ=∫Xf⋅gdμ+∫Xf⋅gdν for all f∈L2(X,λ)(*) \int _{X} f d \mu = \int _{X} f \cdot g d \mu +\int _{X} f \cdot g d \nu \text{ for all } f \in L^{2}(X, \lambda ) \tag {*}
(iii)

Conclude that, since λ(X)<∞\lambda (X) < \infty,

μ(E)=∫Egdμ+∫Egdν for all E∈M \mu (E) = \int _{E} g d \mu +\int _{E} g d \nu \text{ for all } E \in \mathcal{M}

Therefore, λ{x∈X∣g(x)>1}=0\lambda \left\{ x \in X \mid g(x) > 1\right\} = 0, and, using the absolute continuity of ν\nu with respect to μ\mu, we also have λ{x∈X∣g(x)=0}=0\lambda \left\{ x \in X \mid g(x) = 0\right\} = 0. Consequently, we may assume that 0<g≤10 < g \leq 1 on XX.

(iv)

For a natural number nn and E∈ME \in \mathcal{M}, substitute f=χE⋅1g+1/nf = \chi_{E} \cdot \frac{1}{g+1 / n} in (*) and, using the Monotone Convergence Theorem, take the limit as n→∞n \rightarrow \infty to conclude that

ν(E)=∫Ehdμ for all E∈M, where h=[1/g−1] \nu (E) = \int _{E} h d \mu \text{ for all } E \in \mathcal{M}, \text{ where } h = [1 / g-1]

and hh is non-negative and measurable.

(v)

Show that, by taking a countable measurable partition of XX such that μ\mu and ν\nu are finite on each set in the partition, the above special case of the Radon-Nikodym Theorem extends to the case that the measures μ\mu and ν\nu are σ\sigma-finite.

Problem 20.27

Verify (14).

?
Problem 20.28

Let T,S∈L(H)T, S \in \mathcal{L}(H) be symmetric. Show that T=ST = S if and only if QT=QSQ_{T} = Q_{S}.

?
Problem 20.29

Show the bounded, symmetric operators are a closed subspace of L(H)\mathcal{L}(H). Also show that if TT and SS are symmetric, then so is the composition S∘TS \circ T if and only if TT commutes with SS with respect to composition, that is, S∘T=T∘SS \circ T = T \circ S.

?
Problem 20.30

Let HH be a Hilbert space and the linear operator T:H→HT: H \rightarrow H have the property that ⟨T(u),v⟩=⟨u,T(v)⟩\left\langle T(u), v \right\rangle = \left\langle u, T(v) \right\rangle for all u,v∈Hu, v \in H. Show that TT belongs to L(H)\mathcal{L}(H), that is, TT is bounded.

?
Problem 20.31

Exhibit an operator T∈L(R2)T \in \mathcal{L}\left(\mathbb {R}^{2}\right) for which ∥T∥>sup⁡∥u∥=1∣⟨T(u),u⟩∣\left\| T\right\| > \sup_{\left\| u\right\| = 1}\left| \left\langle T(u), u \right\rangle \right|.

?
Problem 20.32

Let S,T∈L(H)S, T \in \mathcal{L}(H) be symmetric. Assume that S≥TS \geq T and T≥ST \geq S. Prove that T=ST = S.

?
Problem 20.33

Let VV be a closed, nontrivial subspace of a Hilbert space HH and PP the orthogonal projection of HH onto VV. Show that P=P∗,P≥0P = P^{*}, P \geq 0, and ∥P∥=1\left\| P\right\| = 1.

?
Problem 20.34

Let P∈L(H)P \in \mathcal{L}(H) be a projection. Show that PP is the orthogonal projection of HH onto P(H)P(H) if and only if P=P∗P = P^{*}.

?
Problem 20.35

Let {φk}\left\{ \varphi_{k}\right\} be an orthonormal basis for a Hilbert space HH and for each natural number nn, define PnP_{n} to be the orthogonal projection of HH onto the linear span of {φ1,…,φn}\left\{ \varphi_{1}, \ldots , \varphi_{n}\right\}. Show that PnP_{n} is symmetric and

0≤Pn≤Pn+1≤ Id for all n 0 \leq P_{n} \leq P_{n+1} \leq \text{ Id for all } n

Show that {Pn}\left\{ P_{n}\right\} converges pointwise on HH to Id but does not converge uniformly on the unit ball.

?
Problem 20.36

Show that if T∈L(H)T \in \mathcal{L}(H) is invertible, so is T∗∘TT^{*} \circ T and therefore so is T∗T^{*}.

?
Problem 20.37

Let T∈L(H)T \in \mathcal{L}(H) be symmetric and non-negative. Show that for all u,v∈Hu, v \in H,

∣⟨T(u),v⟩∣2≤⟨T(u),u⟩⋅⟨T(v),v⟩. \left| \left\langle T(u), v \right\rangle \right|^{2} \leq \left\langle T(u), u \right\rangle \cdot \left\langle T(v), v \right\rangle .
?
Problem 20.38

Use the preceding problem to show that if S,T∈L(H)S, T \in \mathcal{L}(H) are symmetric and S≥TS \geq T, then for each u∈Hu \in H, ∥S(u)−T(u)∥4=⟨(S−T)(u),(S−T)(u)⟩2≤∣⟨(S−T)(u),u⟩∣∣⟨(S−T)2(u),(S−T)(u)⟩∣\left\| S(u)-T(u)\right\|^{4} = \left\langle (S-T)(u), (S-T)(u) \right\rangle^{2} \leq \left| \left\langle (S-T)(u), u \right\rangle \right|\left| \left\langle (S-T)^{2}(u), (S-T)(u) \right\rangle \right| and thereby conclude that

∥S(u)−T(u)∥4≤∣⟨S(u),u⟩−⟨T(u),u⟩∣⋅∥S−T∥3⋅∥u∥2. \left\| S(u)-T(u)\right\| ^{4} \leq \left| \left\langle S(u), u \right\rangle - \left\langle T(u), u \right\rangle \right| \cdot \left\| S-T\right\| ^{3} \cdot \left\| u\right\| ^{2} .
?
Problem 20.39

A sequence {Tn}\left\{ T_{n}\right\} of symmetric operators in L(H)\mathcal{L}(H) is said to be monotone increasing provided that Tn+1≥TnT_{n+1} \geq T_{n} for each nn, and said to be bounded above provided that there is a symmetric operator SS in L(H)\mathcal{L}(H) such that Tn≤ST_{n} \leq S for all nn.

?
(i)

Use the preceding problem to show that a monotone increasing sequence {Tn}\left\{ T_{n}\right\} of symmetric operators in L(H)\mathcal{L}(H) converges pointwise to a symmetric operator in L(H)\mathcal{L}(H) if and only if it is bounded above.

(ii)

Show that a monotone increasing sequence {Tn}\left\{ T_{n}\right\} of symmetric operators in L(H)\mathcal{L}(H) is bounded above if and only if it is pointwise bounded, that is, for each h∈Hh \in H, the sequence {Tn(h)}\left\{ T_{n}(h)\right\} is bounded.

Problem 20.40

Let S∈L(H)S \in \mathcal{L}(H) be a symmetric operator for which 0≤S≤Id⁡0 \leq S \leq \operatorname {Id}. Define a sequence {Tn}\left\{ T_{n}\right\} in L(H)\mathcal{L}(H) by letting T1=1/2(Id⁡−S)T_{1} = 1 / 2(\operatorname {Id}-S) and if nn is a natural number for which Tn∈L(H)T_{n} \in \mathcal{L}(H) has been defined, defining Tn+1=1/2(Id−S+Tn2)T_{n+1} = 1 / 2\left(\mathrm{Id}-S+T_{n}^{2}\right).

?
(i)

Show that for each natural number n,Tnn, T_{n} and Tn+1−TnT_{n+1}-T_{n} are polynomials in Id −S-S with non-negative coefficients.

(ii)

Show that {Tn}\left\{ T_{n}\right\} is a monotone increasing sequence of symmetric operators that is bounded above by Id.

(iii)

Use the preceding problem to show that {Tn}\left\{ T_{n}\right\} converges pointwise to a symmetric operator TT for which 0≤T≤Id⁡0 \leq T \leq \operatorname {Id} and T=1/2(Id⁡−S+T2)T = 1 / 2\left(\operatorname {Id}-S+T^{2}\right).

(iv)

Define A=(Id⁡−T)A = (\operatorname {Id}-T). Show that A2=SA^{2} = S.

Problem 20.41

Let T∈L(H)T \in \mathcal{L}(H) be a non-negative, symmetric operator. A non-negative, symmetric operator A∈L(H)A \in \mathcal{L}(H) is called a square root of TT provided that A2=TA^{2} = T. Use the inductive construction in the preceding problem to show that TT has a square root AA which commutes with each operator in L(H)\mathcal{L}(H) that commutes with TT. Show that the square root is unique: it is denoted by T\sqrt{T}. Finally, show that TT is invertible if and only if T\sqrt{T} is invertible.

?
Problem 20.42

An invertible operator T∈L(H)T \in \mathcal{L}(H) is said to be orthogonal provided that T−1=T∗T^{-1} = T^{*}. Show that an invertible operator is orthogonal if and only if it is an isometry.

?
Problem 20.43

Let T∈L(H)T \in \mathcal{L}(H) be invertible. Show that there is an orthogonal invertible operator A∈L(H)A \in \mathcal{L}(H) and a non-negative, symmetric invertible operator B∈L(H)B \in \mathcal{L}(H) such that T=B∘AT = B \circ A.

?
Problem 20.44

Show that if HH is infinite dimensional and T∈L(H)T \in \mathcal{L}(H) is invertible, then TT is not compact.

?
Problem 20.45

Prove Proposition 17.

Proposition 17: For T∈L(H)T \in \mathcal{L}(H), the following three properties are equivalent: (i) TT is compact; (ii) T(B)T(B) is totally bounded, where BB is the closed unit ball in HH; (iii) if {hn}\left\{ h_{n}\right\} is a bounded sequence in HH, then {T(hn)}\left\{ T(h_{n})\right\} has a strongly convergent subsequence.

?
Problem 20.46

Let K(H)\mathcal{K}(H) denote the set of compact operators in L(H)\mathcal{L}(H). Show that K(H)\mathcal{K}(H) is a linear subspace of L(H)\mathcal{L}(H). Moreover, show that for K∈K(H)K \in \mathcal{K}(H) and T∈L(H)T \in \mathcal{L}(H), both K∘TK \circ T and T∘KT \circ K belong to K(H)\mathcal{K}(H).

?
Problem 20.47

Show that a linear operator T:H→HT: H \rightarrow H is continuous if and only if it maps weakly convergent sequences to weakly convergent sequences.

?
Problem 20.48

Show that K∈L(H)K \in \mathcal{L}(H) is compact if and only if whenever if {un}→u\left\{ u_{n}\right\} \rightarrow u weakly and {vn}→v\left\{ v_{n}\right\} \rightarrow v weakly, then ⟨K(un),vn⟩→⟨K(u),v⟩\left\langle K\left(u_{n}\right), v_{n} \right\rangle \rightarrow \left\langle K(u), v \right\rangle.

?
Problem 20.49

Let {Pn}\left\{ P_{n}\right\} be a sequence of orthogonal projections in L(H)\mathcal{L}(H) with the property that for natural numbers nn and m,Pn(H)m, P_{n}(H) and Pm(H)P_{m}(H) are orthogonal finite dimensional subspaces of HH. Let {λn}\left\{ \lambda_{n}\right\} be a bounded sequence of real numbers. Show that

K=∑n=1∞λn⋅Pn K = \sum _{n = 1}^{\infty } \lambda _{n} \cdot P_{n}

is a properly defined symmetric operator in L(H)\mathcal{L}(H) that is compact if and only if {λn}\left\{ \lambda_{n}\right\} converges to 0.

?
Problem 20.50

For XX a Banach space, define an operator T∈L(X)T \in \mathcal{L}(X) be compact provided that T(B)T(B) has compact closure. Show that Proposition 17 holds for a general Banach space and Proposition 18 holds for a reflexive Banach space.

Proposition 17: For T∈L(H)T \in \mathcal{L}(H), the following three properties are equivalent: (i) TT is compact; (ii) T(B)T(B) is totally bounded, where BB is the closed unit ball in HH; (iii) if {hn}\left\{ h_{n}\right\} is a bounded sequence in HH, then {T(hn)}\left\{ T(h_{n})\right\} has a strongly convergent subsequence. Proposition 18: An operator TT in L(H)\mathcal{L}(H) is compact if and only if it maps weakly convergent sequences to strongly convergent sequences, that is, if {hn}→h\left\{ h_{n}\right\} \rightarrow h weakly, then {T(hn)}→T(h)\left\{ T(h_{n})\right\} \rightarrow T(h) strongly.

?
Problem 20.51

Let T∈L(H)T \in \mathcal{L}(H) be compact and symmetric. Define

α=inf⁡∥h∥=1⟨T(h),h⟩ and β=sup⁡∥h∥=1⟨T(h),h⟩. \alpha = \inf _{\left\| h\right\| = 1} \left\langle T(h), h \right\rangle \text{ and } \beta = \sup _{\left\| h\right\| = 1} \left\langle T(h), h \right\rangle .

Show that if α<0\alpha < 0, then α\alpha is an eigenvalue of TT and if β>0\beta > 0, then β\beta is an eigenvalue of TT. Exhibit an example where α=0\alpha = 0 and yet α\alpha is not an eigenvalue of TT, that is, TT is one-to-one.

?
Problem 20.52

Let K∈L(H)K \in \mathcal{L}(H) be compact and symmetric. Suppose

sup⁡∥h∥=1⟨K(h),h⟩=β>0 \sup _{\left\| h\right\| = 1} \left\langle K(h), h \right\rangle = \beta > 0

Let {hn}\left\{ h_{n}\right\} be a sequence of unit vectors for which lim⁡n→∞⟨K(hn),hn⟩=β\lim_{n \rightarrow \infty } \left\langle K\left(h_{n}\right), h_{n} \right\rangle = \beta. Show that a subsequence of {hn}\left\{ h_{n}\right\} converges strongly to an eigenvector of TT with corresponding eigenvalue β\beta.

?
Problem 20.53

Let K∈L(H)K \in \mathcal{L}(H) be compact. Show that T=K∗KT = K^{*} K is compact and symmetric. Then use the Hilbert-Schmidt Theorem to show that there is an orthonormal sequence {φk}\left\{ \varphi_{k}\right\} of HH such that T(φk)=λkφkT\left(\varphi_{k}\right) = \lambda_{k} \varphi_{k} for all kk and T(h)=0T(h) = 0 if hh is orthogonal to {φk}k=0∞\left\{ \varphi_{k}\right\}_{k = 0}^{\infty }. Conclude that if hh is orthogonal to {φk}k=0∞\left\{ \varphi_{k}\right\}_{k = 0}^{\infty }, then

∥K(h)∥2=⟨K(h),K(h)⟩=⟨T(h),h⟩=0. \left\| K(h)\right\| ^{2} = \left\langle K(h), K(h) \right\rangle = \left\langle T(h), h \right\rangle = 0.

Define H0H_{0} to be the closed linear span of {Km(φk)∣m≥0,k≥1}\left\{ K^{m}\left(\varphi_{k}\right) \mid m \geq 0, k \geq 1\right\}. Show that H0H_{0} is closed and separable, K(H0)⊆H0K\left(H_{0}\right) \subseteq H_{0} and K=0K = 0 on H0⊥H_{0}^{\perp }.

?
Problem 20.54

Let K(H)\mathcal{K}(H) denote the set of compact operators in L(H)\mathcal{L}(H). Show that K(H)\mathcal{K}(H) is a closed subspace of L(H)\mathcal{L}(H) that has the set of operators of finite rank as a dense subspace. Is K(H)\mathcal{K}(H) an open subset of L(H)\mathcal{L}(H) ?

?
Problem 20.55

Show that the composition in either order of a Fredholm operator of index 0 with an invertible operator is also Fredholm of index 0.

?
Problem 20.56

Show that the composition of two Fredholm operators of index 0 is also Fredholm of index 0.

?
Problem 20.57

Show that an operator T∈L(H)T \in \mathcal{L}(H) is Fredholm of index 0 if and only if it is the perturbation of an invertible operator by an operator of finite rank.

?
Problem 20.58

Argue as follows to show that the collection of invertible operators in L(H)\mathcal{L}(H) is an open subset of L(H)\mathcal{L}(H).

?
(i)

For A∈L(H)A \in \mathcal{L}(H) with ∥A∥<1\left\| A\right\| < 1, use the completeness of L(H)\mathcal{L}(H) to show that the so-called Neumann series ∑n=0∞An\sum_{n = 0}^{\infty } A^{n} converges to an operator in L(H)\mathcal{L}(H) that is the inverse of Id⁡−A\operatorname {Id}-A.

(ii)

For an invertible operator S∈L(H)S \in \mathcal{L}(H) show that for any T∈L(H),T=S[Id⁡+S−1(T−S)]T \in \mathcal{L}(H), T = S\left[\operatorname {Id}+S^{-1}(T-S)\right].

(iii)

Use (i) and (ii) to show that if S∈L(H)S \in \mathcal{L}(H) is invertible then so is any T∈L(H)T \in \mathcal{L}(H) for which ∥S−T∥<1/∥S−1∥\left\| S-T\right\| < 1 /\left\| S^{-1}\right\|.

Problem 20.59

Show that the set of operators in L(H)\mathcal{L}(H) that are Fredholm of index 0 is an open subset of L(H)\mathcal{L}(H).

?
Problem 20.60

By following the orthogonal approximation sequence method used in the proof of Proposition 22, provide another proof of Lemma 14 in case HH is separable.

?
Problem 20.61

For T∈L(H)T \in \mathcal{L}(H), suppose that ⟨T(h),h⟩≥∥h∥2\left\langle T(h), h \right\rangle \geq \left\| h\right\|^{2} for all h∈Hh \in H. Assume that K∈L(H)K \in \mathcal{L}(H) is compact and T+KT+K is one-to-one. Show that T+KT+K is onto.

?
Problem 20.62

Let K∈L(H)K \in \mathcal{L}(H) be compact and μ∈R\mu \in \mathbb {R} have ∣μ∣>∥K∥\left|\mu \right| > \left\| K\right\|. Show that μ−K\mu -K is invertible.

?
Problem 20.63

Let S∈L(H)S \in \mathcal{L}(H) have ∥S∥<1,K∈L(H)\left\| S\right\| < 1, K \in \mathcal{L}(H) be compact and (Id⁡+S+K)(H)=H(\operatorname {Id}+S+K)(H) = H. Show that Id +S+K+S+K is one-to-one.

?
Problem 20.64

Let GL(H)\mathcal{G} L(H) denote the set of invertible operators in L(H)\mathcal{L}(H).

?
(i)

Show that under the operation of composition of operators, GL(H)\mathcal{G} L(H) is a group: it is called the general linear group of HH.

(ii)

An operator TT in GL(H)\mathcal{G} L(H) is said be orthogonal, provided that T∗=T−1T^{*} = T^{-1}. Show that the set of orthogonal operators is a subgroup of GL(H)\mathcal{G} L(H) : it is called the orthogonal group.

Problem 20.65

Let T∈L(H)T \in \mathcal{L}(H) be Fredholm of index zero, and K∈L(H)K \in \mathcal{L}(H) be compact. Show that T+KT+K is Fredholm of index zero.

?
Problem 20.66

Let X0X_{0} be a finite codimensional subspace of a linear space XX. Show that all finite dimensional linear complements of X0X_{0} in XX have the same dimension.

?