Chapter VIII

C*-Algebras

[54/0%]
Show
LaTeX
Problem VIII.1.1

Verify the statements made in the following examples.

  • If H\mathcal{H} is a Hilbert space, A=B(H)\mathcal{A} = \mathcal{B}(\mathcal{H}) is a C∗C^*-algebra where for each AA in B(H)\mathcal{B}(\mathcal{H}), A∗=A^* = the adjoint of AA.

  • If H\mathcal{H} is a Hilbert space, B0(H)\mathcal{B}_0(\mathcal{H}) is a C∗C^*-subalgebra of B(H)\mathcal{B}(\mathcal{H}), though B0(H)\mathcal{B}_0(\mathcal{H}) does not have an identity if H\mathcal{H} is infinite dimensional.

  • If XX is a compact space, C(X)C(X) is a C∗C^*-algebra where f∗(x)=f(x)‾f^*(x) = \overline{f(x)} for ff in C(X)C(X) and xx in XX.

  • If (X,Ω,μ)(X, \Omega , \mu ) is a σ\sigma-finite measure space, L∞(X,Ω,μ)L^\infty (X, \Omega , \mu ) is a C∗C^*-algebra where the involution is defined as in the preceding part.

  • If XX is locally compact but not compact, C0(X)C_0(X) is a C∗C^*-algebra without identity.

?
Problem VIII.1.2

Let A={f∈C(cl⁡D):f is analytic in D}\mathcal{A} = \{ f \in C(\operatorname {cl} \mathbb {D}): f \text{ is analytic in } \mathbb {D}\} and for ff in A\mathcal{A} define f∗f^* by f∗(z)=f(zˉ)‾f^*(z) = \overline{f(\bar{z})}. Show that A\mathcal{A} is a Banach algebra, f∗∈Af^* \in \mathcal{A} when f∈Af \in \mathcal{A}, and ∥f∗∥=∥f∥\left\| f^*\right\| = \left\| f\right\|, but A\mathcal{A} is not a C∗C^*-algebra.

?
Problem VIII.1.3

If {Ai:i∈I}\{ \mathcal{A}_i : i \in I\} is a collection of C∗C^*-algebras, show that ⨁∞Ai\bigoplus_\infty \mathcal{A}_i and ⨁0Ai\bigoplus_0 \mathcal{A}_i are C∗C^*-algebras.

?
Problem VIII.1.4

Let XX be a locally compact space and let A\mathcal{A} be a C∗C^*-algebra. If Cb(X,A)=C_b(X, \mathcal{A}) = the collection of bounded continuous functions from X→AX \to \mathcal{A}, show that Cb(X,A)C_b(X, \mathcal{A}) is a C∗C^*-algebra. Let C0(X,A)=C_0(X, \mathcal{A}) = all of the continuous functions f:X→Af: X \to \mathcal{A} such that for every ε>0\varepsilon > 0, {x∈X:∥f(x)∥≥ε}\{ x \in X : \left\| f(x)\right\| \geq \varepsilon \} is compact. Show that C0(X,A)C_0(X, \mathcal{A}) is a C∗C^*-algebra.

?
Problem VIII.2.1

Prove a converse to the following proposition: if A\mathcal{A} is an abelian C∗C^*-algebra with maximal ideal space Σ\Sigma and a∈Aa \in \mathcal{A} is such that A=C∗(a)\mathcal{A} = C^*(a) (the C∗C^*-algebra generated by aa and the identity), then the map τ:Σ→σ(a)\tau : \Sigma \to \sigma (a) defined by τ(h)=h(a)\tau (h) = h(a) is a homeomorphism. Specifically, show: if KK is a compact subset of C\mathbb {C}, C(K)C(K) is a singly generated C∗C^*-algebra.

?
Problem VIII.2.2

If A\mathcal{A} is an abelian C∗C^*-algebra with a finite number of C∗C^*-generators a1,…,ana_1, \dots , a_n, then there is a compact subset XX of Cn\mathbb {C}^n and an isometric ∗*-isomorphism ρ:A→C(X)\rho : \mathcal{A} \to C(X) such that ρ(ak)=zk\rho (a_k) = z_k, 1≤k≤n1 \leq k \leq n, where zk(λ1,…,λn)=λkz_k(\lambda_1, \dots , \lambda_n) = \lambda_k (see Exercise VII.8.12).

?
Problem VIII.2.3

If XX is a compact Hausdorff space, show that XX is totally disconnected if and only if C(X)C(X) is the closed linear span of its projections (≡\equiv hermitian idempotents).

?
Problem VIII.2.4

Using the terminology of the preceding exercise, show that if (X,Ω,μ)(X, \Omega , \mu ) is a σ\sigma-finite measure space, the maximal ideal space of L∞(X,Ω,μ)L^\infty (X, \Omega , \mu ) is totally disconnected.

?
Problem VIII.2.5

If A\mathcal{A} is a C∗C^*-algebra with identity and a=a∗a = a^*, show that exp⁡(ia)=u\exp (ia) = u is unitary. Is the converse true?

?
Problem VIII.2.6

Let XX be compact and fix a point x0x_0 in XX. Let

A={{fn}:fn∈C(X),sup⁡n∥fn∥<∞, and {fn(x0)} is a convergent sequence}. \mathcal{A} = \{ \{ f_n\} : f_n \in C(X), \sup _n \left\| f_n\right\| < \infty , \text{ and } \{ f_n(x_0)\} \text{ is a convergent sequence}\} .

Show that A\mathcal{A} is an abelian C∗C^*-algebra with identity and find its maximal ideal space.

?
Problem VIII.2.7

If XX is completely regular, then Cb(X)C_b(X) is a C∗C^*-algebra and its maximal ideal space is the Stone--Cech compactification of XX.

?
Problem VIII.3.1

Prove the uniqueness statement in the following proposition for the case that A\mathcal{A} is abelian: if a∈Re⁡Aa \in \operatorname {Re} \mathcal{A} (the hermitian elements of A\mathcal{A}), then there are unique positive elements u,vu, v in A\mathcal{A} such that a=u−va = u - v and uv=vu=0uv = vu = 0.

?
Problem VIII.3.2

Prove: if a∈A+a \in \mathcal{A}_+ and n≥1n \geq 1, there is a unique element bb in A+\mathcal{A}_+ such that a=bna = b^n.

?
Problem VIII.3.3

Let A∈B(L2(0,1))A \in \mathcal{B}(L^2(0,1)) be defined by (Af)(t)=tf(t)(Af)(t) = tf(t). Show that A≥0A \geq 0 and find A1/nA^{1/n}.

?
Problem VIII.3.4

Let (X,Ω,μ)(X, \Omega , \mu ) be a σ\sigma-finite measure space, let ϕ∈L∞(X,Ω,μ)\phi \in L^\infty (X, \Omega , \mu ), and define MϕM_\phi as the multiplication operator on L2(X,Ω,μ)L^2(X, \Omega , \mu ) given by Mϕf=ϕfM_\phi f = \phi f. Show that Mϕ≥0M_\phi \geq 0 if and only if ϕ(x)≥0\phi (x) \geq 0 a.e.[μ][\mu ]. What is Mϕ1/nM_\phi^{1/n}? If Mϕ∈Re⁡B(H)M_\phi \in \operatorname {Re} \mathcal{B}(\mathcal{H}), find the positive and negative parts of MϕM_\phi.

?
Problem VIII.3.5

Find an example of a positive operator on a Hilbert space that has a nonhermitian square root.

?
Problem VIII.3.6

If a∈Re⁡Aa \in \operatorname {Re} \mathcal{A}, show that ∣a∣≡(a2)1/2=a++a−\left|a\right| \equiv (a^2)^{1/2} = a_+ + a_-, where a+a_+ and a−a_- are the positive and negative parts of aa (that is, the unique positive elements with a=a+−a−a = a_+ - a_- and a+a−=a−a+=0a_+ a_- = a_- a_+ = 0).

?
Problem VIII.3.7

If a∈A+a \in \mathcal{A}_+, show that x∗ax∈A+x^* a x \in \mathcal{A}_+ for every xx in A\mathcal{A}.

?
Problem VIII.3.8

If a,b∈Aa, b \in \mathcal{A}, 0≤a≤b0 \leq a \leq b, and aa is invertible, then bb is invertible and b−1≤a−1b^{-1} \leq a^{-1}.

?
Problem VIII.3.9

If a,b∈Re⁡Aa, b \in \operatorname {Re} \mathcal{A}, a≤ba \leq b, and ab=baab = ba, then f(a)≤f(b)f(a) \leq f(b) for every increasing continuous function ff on R\mathbb {R}.

?
Problem VIII.3.10

If a∈Re⁡Aa \in \operatorname {Re} \mathcal{A} and ∥a∥≤1\left\| a\right\| \leq 1, show that aa is the sum of two unitaries.

?
Problem VIII.3.11

If α>0\alpha > 0, define fα:(−α−1,∞)→Rf_\alpha : (-\alpha^{-1}, \infty ) \to \mathbb {R} by fα(t)=t/(1+αt)=α−1[1−(1+αt)−1]f_\alpha (t) = t/(1 + \alpha t) = \alpha^{-1}[1 - (1 + \alpha t)^{-1}]. Show:

?
(a)

If 0≤a≤b0 \leq a \leq b in A\mathcal{A}, fα(a)≤fα(b)f_\alpha (a) \leq f_\alpha (b) for all α>0\alpha > 0;

(b)

fα(t)<min⁡{t,α−1}f_\alpha (t) < \min \{ t, \alpha^{-1}\} for t>0t > 0;

(c)

lim⁡α→0fα(t)=t\lim_{\alpha \to 0} f_\alpha (t) = t uniformly on bounded intervals in [0,∞)[0, \infty );

(d)

if 0≤α≤β0 \leq \alpha \leq \beta, fα≤fβf_\alpha \leq f_\beta on [0,∞)[0, \infty );

(e)

fα∘fβ=fα+βf_\alpha \circ f_\beta = f_{\alpha + \beta };

(f)

lim⁡α→∞αfα(t)=1\lim_{\alpha \to \infty } \alpha f_\alpha (t) = 1 uniformly on bounded intervals in [0,∞)[0, \infty ).

Problem VIII.3.12

If a,b∈A+a, b \in \mathcal{A}_+ and a≤ba \leq b, show that aβ≤bβa^\beta \leq b^\beta for 0≤β≤10 \leq \beta \leq 1. (Here aβ=f(a)a^\beta = f(a) where f(t)=tβf(t) = t^\beta.)

?
Problem VIII.3.13

Give an example of a C∗C^*-algebra A\mathcal{A} and positive elements a,ba, b in A\mathcal{A} such that a≤ba \leq b but b2−a2∉A+b^2 - a^2 \notin \mathcal{A}_+.

?
Problem VIII.3.14

Let A=B(l2)\mathcal{A} = \mathcal{B}(l^2), let a=a = the unilateral shift on l2l^2, and let b=a∗b = a^*. Show that σ(ab)≠σ(ba)\sigma (ab) \neq \sigma (ba).

?
Problem VIII.3.15

Let W∈B(H)W \in \mathcal{B}(\mathcal{H}) and show that the following statements are equivalent: (a) WW is a partial isometry (that is, an operator such that for hh in (ker⁡W)⊥(\ker W)^\perp, ∥Wh∥=∥h∥\left\| Wh\right\| = \left\| h\right\|); (b) W∗W^* is a partial isometry; (c) W∗WW^*W is a projection; (d) WW∗WW^* is a projection; (e) WW∗W=WWW^*W = W; (f) W∗WW∗=W∗W^*WW^* = W^*.

?
Problem VIII.3.16

If WW is a partial isometry, show that W∗WW^*W is the projection onto the initial space of WW (that is, the space (ker⁡W)⊥(\ker W)^\perp) and WW∗WW^* is the projection onto the final space of WW (that is, the space ran⁡W\operatorname {ran} W).

?
Problem VIII.3.17

If W1,W2W_1, W_2 are partial isometries, define W1≾W2W_1 \precsim W_2 to mean that W1∗W1≤W2∗W2W_1^* W_1 \leq W_2^* W_2, W1W1∗≤W2W2∗W_1 W_1^* \leq W_2 W_2^*, and W2h=W1hW_2 h = W_1 h whenever hh is in the initial space of W1W_1. Show that ≾\precsim is a partial ordering on the set of partial isometries and that a partial isometry WW is a maximal element in this ordering if and only if either WW or W∗W^* is an isometry.

?
Problem VIII.3.18

Using the terminology of the preceding exercise, show that the extreme points of ball B(H)\mathcal{B}(\mathcal{H}) are the maximal partial isometries. (See Exercise V.7.10.)

?
Problem VIII.3.19

Find the polar decomposition of each of the following operators:

?
(a)

MϕM_\phi as defined in (II.1.5);

(b)

the unilateral shift;

(c)

the weighted unilateral shift [A(x1,x2,… )=(0,α1x1,α2x2,… )[A(x_1, x_2, \dots ) = (0, \alpha_1 x_1, \alpha_2 x_2, \dots ) for xx in l2l^2 and sup⁡n∣αn∣<∞]\sup_n \left|\alpha_n\right| < \infty ] with nonzero weights;

(d)

A⊕αA \oplus \alpha (in terms of the polar decomposition of AA).

Problem VIII.3.20

Let A∈B(H)A \in \mathcal{B}(\mathcal{H}) such that ker⁡A=(0)\ker A = (0) and A≥0A \geq 0 and define SS on K=H⊕H⊕⋯\mathcal{K} = \mathcal{H} \oplus \mathcal{H} \oplus \cdots by S(h1,h2,… )=(0,Ah1,Ah2,… )S(h_1, h_2, \dots ) = (0, Ah_1, Ah_2, \dots ). Find the polar decomposition of SS, S=W∣S∣S = W \left|S\right|, and show that S=∣S∣WS = \left|S\right| W.

?
Problem VIII.3.21

Show that the parts of the polar decomposition of a normal operator commute.

?
Problem VIII.3.22

If A∈B(H)A \in \mathcal{B}(\mathcal{H}), show that there is a positive operator PP and a partial isometry WW such that A=PWA = PW. Discuss the uniqueness of PP and WW.

?
Problem VIII.3.23

If AA is normal and invertible, show that the parts of the polar decomposition of AA belong to C∗(A)C^*(A).

?
Problem VIII.3.24

Give an example of a normal operator AA such that the partial isometry in the polar decomposition of AA does not belong to C∗(A)C^*(A).

?
Problem VIII.3.25

Show that for an arbitrary C∗C^*-algebra A\mathcal{A} it is not necessarily true that ab∈A+ab \in \mathcal{A}_+ whenever aa and b∈A+b \in \mathcal{A}_+. If, however, a,b∈A+a, b \in \mathcal{A}_+ and ab=baab = ba, then ab∈A+ab \in \mathcal{A}_+.

?
Problem VIII.4.1

Complete the proof of the following proposition: if XX is compact and II is a closed ideal of C(X)C(X), then there is a closed subset FF of XX such that I={f∈C(X):f(x)=0 for all x in F}I = \{ f \in C(X) : f(x) = 0 \text{ for all } x \text{ in } F\}. Moreover, C(X)/IC(X)/I is isometrically isomorphic to C(F)C(F).

?
Problem VIII.4.2

Show that Mn(C)M_n(\mathbb {C}) has no nontrivial ideals. Find all of the left ideals.

?
Problem VIII.4.3

If α\alpha is an infinite cardinal number, let Iα={A∈B(H):dim⁡cl⁡(ran⁡A)≤α}I_\alpha = \{ A \in \mathcal{B}(\mathcal{H}) : \dim \operatorname {cl}(\operatorname {ran} A) \leq \alpha \}. Show that IαI_\alpha is a closed ideal in B(H)\mathcal{B}(\mathcal{H}).

?
Problem VIII.4.4

Let SS be the unilateral shift on l2l^2. Show that C∗(S)⊇B0(l2)C^*(S) \supseteq \mathcal{B}_0(l^2) and C∗(S)/B0(l2)C^*(S)/\mathcal{B}_0(l^2) is abelian. Show that the maximal ideal space of C∗(S)/B0(l2)C^*(S)/\mathcal{B}_0(l^2) is homeomorphic to ∂D\partial \mathbb {D}.

?
Problem VIII.4.5

If VV is the Volterra operator on L2(0,1)L^2(0,1), show that C∗(V)=C+B0(L2(0,1))C^*(V) = \mathbb {C} + \mathcal{B}_0(L^2(0,1)).

?
Problem VIII.4.6

If A\mathcal{A} is a C∗C^*-algebra, II is a closed ideal of A\mathcal{A}, and B\mathcal{B} is a C∗C^*-subalgebra of A\mathcal{A}, show that the C∗C^*-algebra generated by I∪BI \cup \mathcal{B} is I+BI + \mathcal{B}.

?
Problem VIII.4.7

If A\mathcal{A} is a C∗C^*-algebra and II and JJ are closed ideals in A\mathcal{A}, show that I+JI + J is a closed ideal of A\mathcal{A}.

?
Problem VIII.5.1

Let A\mathcal{A} be a C∗C^*-algebra with identity and let π:A→B(H)\pi : \mathcal{A} \to \mathcal{B}(\mathcal{H}) be a ∗*-homomorphism [but don't assume that π(1)=1\pi (1) = 1]. Let P1=π(1)P_1 = \pi (1). Show that P1P_1 is a projection and H1≡P1H\mathcal{H}_1 \equiv P_1 \mathcal{H} reduces π(A)\pi (\mathcal{A}). If π1(a)=π(a)∣H1\pi_1(a) = \pi (a) | \mathcal{H}_1, show that π1:A→B(H1)\pi_1: \mathcal{A} \to \mathcal{B}(\mathcal{H}_1) is a representation.

?
Problem VIII.5.2

Show that the following representation is a cyclic representation and find all of the cyclic vectors: if (X,Ω,μ)(X, \Omega , \mu ) is a σ\sigma-finite measure space and H=L2(μ)\mathcal{H} = L^2(\mu ), then π:L∞(μ)→B(H)\pi : L^\infty (\mu ) \to \mathcal{B}(\mathcal{H}) defined by π(ϕ)=Mϕ\pi (\phi ) = M_\phi (the multiplication operator) is a representation.

?
Problem VIII.5.3

Show that the following representation is a cyclic representation and find all the cyclic vectors: if XX is a compact space and μ\mu is a positive Borel measure on XX, then π:C(X)→B(L2(μ))\pi : C(X) \to \mathcal{B}(L^2(\mu )) defined by π(f)=Mf\pi (f) = M_f is a representation.

?
Problem VIII.5.4

If XX is compact and μ\mu is a positive measure on XX, let πμ:C(X)→B(L2(μ))\pi_\mu : C(X) \to \mathcal{B}(L^2(\mu )) be the representation defined by πμ(f)=Mf\pi_\mu (f) = M_f. If μ,ν\mu , \nu are positive measures on XX show that πμ⊕πν\pi_\mu \oplus \pi_\nu is cyclic if and only if μ⊥ν\mu \perp \nu. If μ⊥ν\mu \perp \nu, then πμ⊕πν\pi_\mu \oplus \pi_\nu is equivalent to πμ+ν\pi_{\mu + \nu }. Also, πμ(n)\pi_\mu^{(n)} is not cyclic if n≥2n \geq 2.

?
Problem VIII.5.5

Verify the statements in the following example. Let XX be a compact space and let {μn}\{ \mu_n\} be a sequence of measures on XX. For each nn let πn:C(X)→B(L2(μn))\pi_n : C(X) \to \mathcal{B}(L^2(\mu_n)) be defined by πn(f)=Mf\pi_n(f) = M_f on L2(μn)L^2(\mu_n). Then π=⨁nπn\pi = \bigoplus_n \pi_n is a representation. If the measures {μn}\{ \mu_n\} are pairwise mutually singular, then π\pi is equivalent to the representation f↦Mff \mapsto M_f of C(X)→B(L2(μ))C(X) \to \mathcal{B}(L^2(\mu )), where μ=∑n=1∞μn/2n∥μn∥\mu = \sum_{n=1}^\infty \mu_n / 2^n \left\| \mu_n\right\|.

?
Problem VIII.5.6

If A=C+B0(H)\mathcal{A} = \mathbb {C} + \mathcal{B}_0(\mathcal{H}) and π:A→B(H)\pi : \mathcal{A} \to \mathcal{B}(\mathcal{H}) is the identity representation, show that π(α)\pi^{(\alpha )} is a cyclic representation.

?
Problem VIII.5.7

Fix a Banach limit LIM on l∞(N)l^\infty (\mathbb {N}) and let H\mathcal{H} be a separable Hilbert space with an orthonormal basis {en}\{ e_n\}. Define f:B(H)→Cf: \mathcal{B}(\mathcal{H}) \to \mathbb {C} by f(T)=LIM⁡{⟨Ten,en⟩}f(T) = \operatorname {LIM}\{ \langle Te_n, e_n \rangle \}. Show that ff is a state on B(H)\mathcal{B}(\mathcal{H}). If πf\pi_f is the corresponding cyclic representation, show that ker⁡πf=B0(H)\ker \pi_f = \mathcal{B}_0(\mathcal{H}). Hence πf\pi_f induces a cyclic representation of B(H)/B0(H)\mathcal{B}(\mathcal{H})/\mathcal{B}_0(\mathcal{H}) that is isometric. Is Hf\mathcal{H}_f separable?

?
Problem VIII.5.8

If ff is a positive linear functional on A\mathcal{A} and α∈(0,∞)\alpha \in (0, \infty ), show that πf\pi_f and παf\pi_{\alpha f} are equivalent representations.

?
Problem VIII.5.9

If a∈Aa \in \mathcal{A}, then a≥0a \geq 0 if and only if f(a)≥0f(a) \geq 0 for every state ff.

?
Problem VIII.5.10

If a∈Aa \in \mathcal{A} and a≠0a \neq 0, then there is a state ff on A\mathcal{A} such that f(a)≠0f(a) \neq 0.

?
Problem VIII.5.11

If A\mathcal{A} is a separable C∗C^*-algebra and {fn}\{ f_n\} is a countable weak* dense subset of SAS_{\mathcal{A}} (the state space of A\mathcal{A}), let f=∑n2−nfnf = \sum_n 2^{-n} f_n. Show that πf\pi_f is an isometry.

?