Section VI.3

Compact Operators

[21/0%]
Show
LaTeX
Problem VI.3.1

If X\mathscr {X} is reflexive and A∈B(X,Y)A \in \mathcal{B}(\mathscr {X}, \mathscr {Y}), show that A(ball X)A(\mathrm{ball}\ \mathscr {X}) is closed in Y\mathscr {Y}.

?
Problem VI.3.2

Prove Proposition 3.5. Let X\mathscr {X}, Y\mathscr {Y}, and Z\mathscr {Z} be Banach spaces.

?
(a)

B0(X,Y)\mathcal{B}_0(\mathscr {X}, \mathscr {Y}) is a closed linear subspace of B(X,Y)\mathcal{B}(\mathscr {X}, \mathscr {Y}).

(b)

If K∈B0(X,Y)K \in \mathcal{B}_0(\mathscr {X}, \mathscr {Y}) and A∈B(Y,Z)A \in \mathcal{B}(\mathscr {Y}, \mathscr {Z}), then AK∈B0(X,Z)AK \in \mathcal{B}_0(\mathscr {X}, \mathscr {Z}).

(c)

If K∈B0(X,Y)K \in \mathcal{B}_0(\mathscr {X}, \mathscr {Y}) and A∈B(Z,X)A \in \mathcal{B}(\mathscr {Z}, \mathscr {X}), then KA∈B0(Z,Y)KA \in \mathcal{B}_0(\mathscr {Z}, \mathscr {Y}).

Problem VI.3.3

If A∈B0(X,Y)A \in \mathcal{B}_0(\mathscr {X}, \mathscr {Y}), show that cl[ran A]\mathrm{cl}[\mathrm{ran}\ A] is separable.

?
Problem VI.3.4

If A∈B0(X,Y)A \in \mathcal{B}_0(\mathscr {X}, \mathscr {Y}) and ran A\mathrm{ran}\ A is closed, show that ran A\mathrm{ran}\ A is finite dimensional.

?
Problem VI.3.5

If A∈B0(X)A \in \mathcal{B}_0(\mathscr {X}) and AA is invertible, show that dim⁡X<∞\dim \mathscr {X} < \infty.

?
Problem VI.3.6

Let (X,Ω,μ)(X, \Omega , \mu ) be a finite measure space, 1<p<∞1 < p < \infty, and 1/p+1/q=11/p + 1/q = 1. If k:X×X→Fk: X \times X \to \mathbb {F} is an Ω×Ω\Omega \times \Omega-measurable function such that sup⁡{∫∣k(x,y)∣q dμ(y):x∈X}<∞\sup \left\{ \int \left|k(x, y)\right|^q\, d\mu (y): x \in X\right\} < \infty, then (Kf)(x)=∫k(x,y)f(y) dμ(y)(Kf)(x) = \int k(x, y) f(y)\, d\mu (y) defines a compact operator on Lp(μ)L^p(\mu ).

?
Problem VI.3.7

Let (X,Ω,μ)(X, \Omega , \mu ) be an arbitrary measure space, 1<p<∞1 < p < \infty, and 1/p+1/q=11/p + 1/q = 1. If k:X×X→Fk: X \times X \to \mathbb {F} is an Ω×Ω\Omega \times \Omega-measurable function such that

M=[∫(∫∣k(x,y)∣p dμ(x))q/pdμ(y)]1/q<∞ M = \left[ \int \left( \int \left|k(x, y)\right|^p\, d\mu (x) \right)^{q/p} d\mu (y) \right]^{1/q} < \infty

and if (Kf)(x)=∫k(x,y)f(y) dμ(y)(Kf)(x) = \int k(x, y) f(y)\, d\mu (y), then K∈B0(Lp(μ))K \in \mathcal{B}_0(L^p(\mu )) and ∥K∥≤M\left\| K\right\| \le M.

?
Problem VI.3.8

Let XX be a compact space and let μ\mu be a positive Borel measure on XX. Let T∈B(Lp(μ),C(X))T \in \mathcal{B}(L^p(\mu ), C(X)) where 1<p<∞1 < p < \infty. Show that if A:Lp(μ)→Lp(μ)A: L^p(\mu ) \to L^p(\mu ) is defined by Af=TfAf = Tf, then AA is compact.

?
Problem VI.3.9

(B.J. Pettis) If X\mathscr {X} is reflexive and T∈B(X,l1)T \in \mathcal{B}(\mathscr {X}, l^1), then TT is a compact operator. Also, if Y\mathscr {Y} is reflexive and T∈B(c0,Y)T \in \mathcal{B}(c_0, \mathscr {Y}), TT is compact.

?
Problem VI.3.10

If XX is compact and {f1,…,fn,g1,…,gn}⊆C(X)\left\{ f_1, \ldots , f_n, g_1, \ldots , g_n\right\} \subseteq C(X), define k(x,y)=∑j=1nfj(x)gj(y)k(x, y) = \sum_{j=1}^{n} f_j(x) g_j(y) for x,y∈Xx, y \in X. Let μ\mu be a regular Borel measure on XX and put Kf(x)=∫k(x,y)f(y) dμ(y)Kf(x) = \int k(x, y) f(y)\, d\mu (y). Show that K∈B(C(X))K \in \mathcal{B}(C(X)) and KK has finite rank.

?
Problem VI.3.11

If XX is compact, k∈C(X×X)k \in C(X \times X), and μ\mu is a regular Borel measure on XX, show that Kf(x)=∫k(x,y)f(y) dμ(y)Kf(x) = \int k(x, y) f(y)\, d\mu (y) defines a compact operator on C(X)C(X).

?
Problem VI.3.12

Let (X,Ω,μ)(X, \Omega , \mu ) be a σ\sigma-finite measure space and for ϕ\phi in L∞(μ)L^\infty (\mu ) let Mϕ:Lp(μ)→Lp(μ)M_\phi : L^p(\mu ) \to L^p(\mu ) be the multiplication operator defined in Example III.2.2. Give necessary and sufficient conditions on (X,Ω,μ)(X, \Omega , \mu ) and ϕ\phi for MϕM_\phi to be compact.

?
Problem VI.3.13

Let τ:[0,1]→[0,1]\tau : [0, 1] \to [0, 1] be continuous and define A:C[0,1]→C[0,1]A: C[0, 1] \to C[0, 1] by Af=f∘τAf = f \circ \tau. Give necessary and sufficient conditions on τ\tau for AA to be compact.

?
Problem VI.3.14

Let A∈B(c0)A \in \mathcal{B}(c_0) and let (αmn)(\alpha_{mn}) be the corresponding matrix as in Exercise 1.7. Give necessary and sufficient conditions on (αmn)(\alpha_{mn}) for AA to be compact.

?
Problem VI.3.15

Let A∈B(l1)A \in \mathcal{B}(l^1) and let (αmn)(\alpha_{mn}) be the corresponding matrix as in Exercise 1.8. Give a necessary and sufficient condition on (αmn)(\alpha_{mn}) for AA to be compact.

?
Problem VI.3.16

If (X,d)(X, d) is a compact metric space and F⊆C(X)\mathscr {F} \subseteq C(X), show that F\mathscr {F} is equicontinuous if and only if for every ε>0\varepsilon > 0 there is a δ>0\delta > 0 such that ∣f(x)−f(y)∣<ε\left|f(x) - f(y)\right| < \varepsilon whenever d(x,y)<δd(x, y) < \delta and f∈Ff \in \mathscr {F}.

?
Problem VI.3.17

If XX is locally compact and F⊆C0(X)\mathscr {F} \subseteq C_0(X), show that F\mathscr {F} is totally bounded if and only if (a) F\mathscr {F} is bounded; (b) F\mathscr {F} is equicontinuous; (c) for every ε>0\varepsilon > 0 there is a compact subset KK of XX such that ∣f(x)∣<ε\left|f(x)\right| < \varepsilon for all ff in F\mathscr {F} and xx in X∖KX \setminus K.

?
Problem VI.3.18

If XX is locally compact and A∈B0(C0(X))A \in \mathcal{B}_0(C_0(X)), then there is a sequence {An}\left\{ A_n\right\} of finite-rank operators such that ∥An−A∥→0\left\| A_n - A\right\| \to 0.

?
Problem VI.3.19

Let X\mathscr {X} be a Banach space and suppose there is a net {Fi}\left\{ F_i\right\} of finite-rank operators on X\mathscr {X} such that (a) sup⁡i∥Fi∥<∞\sup_i \left\| F_i\right\| < \infty; (b) ∥Fix−x∥→0\left\| F_i x - x\right\| \to 0 for all xx in X\mathscr {X}. Show that if A∈B0(X)A \in \mathcal{B}_0(\mathscr {X}), then ∥FiA−A∥→0\left\| F_i A - A\right\| \to 0 and hence there is a sequence {An}\left\{ A_n\right\} of finite-rank operators on X\mathscr {X} such that ∥An−A∥→0\left\| A_n - A\right\| \to 0.

?
Problem VI.3.20

Let 1≤p≤∞1 \le p \le \infty and let (X,Ω,μ)(X, \Omega , \mu ) be a σ\sigma-finite measure space. If A∈B0(Lp(μ))A \in \mathcal{B}_0(L^p(\mu )), show that there is a sequence {An}\left\{ A_n\right\} of finite-rank operators such that ∥An−A∥→0\left\| A_n - A\right\| \to 0.

?
Problem VI.3.21

Let XX be compact and let U\mathscr {U} be the collection of all pairs (C,F)(C, F) where C={U1,…,Un}C = \left\{ U_1, \ldots , U_n\right\} is a finite open cover of XX and F={x1,…,xn}⊆XF = \left\{ x_1, \ldots , x_n\right\} \subseteq X such that xj∈Ujx_j \in U_j for 1≤j≤n1 \le j \le n. If (C1,F1)(C_1, F_1) and (C2,F2)∈U(C_2, F_2) \in \mathscr {U}, define (C1,F1)≤(C2,F2)(C_1, F_1) \le (C_2, F_2) to mean:

  1. C2C_2 is a refinement of C1C_1; that is, each member of C2C_2 is contained in some member of C1C_1.

  2. F1⊆F2F_1 \subseteq F_2. If α=(C,F)∈U\alpha = (C, F) \in \mathscr {U} let {ϕ1,…,ϕn}\left\{ \phi_1, \ldots , \phi_n\right\} be a partition of unity subordinate to CC. If F={x1,…,xn}F = \left\{ x_1, \ldots , x_n\right\}, define Tα:C(X)→C(X)T_\alpha : C(X) \to C(X) by

(Tαf)(x)=∑j=1nf(xj)ϕj(x). (T_\alpha f)(x) = \sum _{j=1}^{n} f(x_j) \phi _j(x).

Then:

?
(a)

Tα∈B00(C(X))T_\alpha \in \mathcal{B}_{00}(C(X));

(b)

∥Tα∥=1\left\| T_\alpha \right\| = 1;

(c)

(U,≤)(\mathscr {U}, \le ) is a directed set and {Tα:α∈U}\left\{ T_\alpha : \alpha \in \mathscr {U}\right\} is a net;

(d)

∥Tαf−f∥→0\left\| T_\alpha f - f\right\| \to 0 for each ff.

Now apply Exercise 19 to obtain a new proof of Theorem 3.11.

Theorem 3.11: If XX is compact, then B00(C(X))\mathcal{B}_{00}(C(X)) is dense in B0(C(X))\mathcal{B}_0(C(X)); that is, the finite-rank operators on C(X)C(X) are dense in the compact operators on C(X)C(X).