Scribd
Upload
37 views

Oa 2

The document introduces some basic concepts from spectral theory for C*-algebras: (1) It defines the spectrum of an element a in a C*-algebra as the set of complex numbers where a - λ is not invertible. (2) Several theorems are proved about the properties of spectra, such as being non-empty and contained within a disk centered at 0. (3) It introduces the concept of the spectral radius of an element as the radius of the smallest disk containing its spectrum.

Uploaded by

Severus Nietzsche
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
37 views

Oa 2

The document introduces some basic concepts from spectral theory for C*-algebras: (1) It defines the spectrum of an element a in a C*-algebra as the set of complex numbers where a - λ is not invertible. (2) Several theorems are proved about the properties of spectra, such as being non-empty and contained within a disk centered at 0. (3) It introduces the concept of the spectral radius of an element as the radius of the smallest disk containing its spectrum.

Uploaded by

Severus Nietzsche
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Basic spectral theory

Teo Banica

"Introduction to operator algebras", 2/6

07/20
Framework

Definition. An abstract C ∗ -algebra is a complex algebra A, with:


(1) A norm a → ||a||, making it a Banach algebra.
(2) An involution a → a∗ , such that ||aa∗ || = ||a||2 , ∀a ∈ A.

– We know that B(H) is a C ∗ -algebra in the above sense.


– And so are all the norm closed ∗-subalgebras A ⊂ B(H).
– We’ll see later that any abstract C ∗ -algebra is of this form.

=⇒ However, useful formalism, because we can construct many


examples of C ∗ -algebras with generators and relations.
Spectra

Definition. The spectrum of an element a ∈ A is the set

σ(a) = λ ∈ C a − λ 6∈ A−1


where A−1 ⊂ A is the set of invertible elements.

Remark. For the usual matrices we obtain the eigenvalues,

Mv = λv ⇐⇒ (M − λ)v = 0
⇐⇒ / MN (C)−1
M −λ∈

or rather the eigenvalue set, with no multiplicities.


Basics

Theorem. The spectrum of any element a ∈ A is:


(1) A compact subset of C.
(2) Contained in the disk D(0, ||a||).

Proof. The spectrum of a norm 1 element is in the unit disk. This


comes from the following formula, valid for any ||a|| < 1:

1
= 1 + a + a2 + . . .
1−a

But this gives (2) by dilation, and shows as well that A−1 is open,
and so that σ(a) ⊂ C is closed, and so we get (1) as well.
Nonzero

Theorem. The spectrum of any element a ∈ A is non-empty.

Proof. Assume σ(a) = ∅. Pick a linear form ϕ ∈ A∗ and set:


 
1
f (λ) = ϕ
λ−a

Then f is differentiable, so holomorphic. For λ >> 0 we have

1 1 a a2
= × 1 + + 2 + ...
λ−a |λ| λ λ
1
≤
|λ| − ||a||

so f (λ) → 0 with λ → ∞. By Liouville f = 0, contradiction.


Products
Theorem. For any two elements a, b ∈ A we have

σ(ab) = σ(ba)

outside {0}. Non-equality at 0 can happen.

Proof. For the equality, by dilation it is enough to prove that


σ(ab) = σ(ba) at λ = 1. But this follows from:

c = (1 − ab)−1 =⇒ 1 + cba = (1 − ba)−1

Consider the shift S : l 2 (N) → l 2 (N), δi → δi+1 . Then

S ∗S = 1 , SS ∗ = Proje ⊥
0

/ σ(S ∗ S), but 0 ∈ σ(SS ∗ ).


and so 0 ∈
Rational functions

Definition. Given a ∈ A, and a rational function f = P/Q having


poles outside σ(a), we set f (a) = P(a)Q(a)−1 . We write:

P(a)
f (a) =
Q(a)

Theorem. We have the “rational functional calculus” formula

σ(f (a)) = f (σ(a))

valid for any f ∈ C(X ) having poles outside σ(a).


Proof

Case f ∈ C[X ]. With f (X ) − λ = c(X − r1 ) . . . (X − rn ):

λ∈
/ σ(f (a)) ⇐⇒ c(a − r1 ) . . . (a − rn ) ∈ A−1
⇐⇒ a − r1 , . . . , a − rn ∈ A−1
⇐⇒ r1 , . . . , rn ∈
/ σ(a)
⇐⇒ λ∈
/ f (σ(a))

Case f ∈ C(X ). With f = P/Q and F = P − λQ:

λ ∈ σ(f (a)) ⇐⇒ 0 ∈ σ(F (a))


⇐⇒ 0 ∈ F (σ(a))
⇐⇒ ∃µ ∈ σ(a), F (µ) = 0
⇐⇒ λ ∈ f (σ(a))
Unitaries

Theorem. The spectrum of a unitary element,

a∗ = a−1

is on the unit circle T ⊂ C.

Proof. This follows by using f (z) = z −1 . Indeed, we have:

σ(a)−1 = σ(a−1 ) = σ(a∗ ) = σ(a)

Thus σ(a) consists of numbers satisfying λ−1 = λ̄.


Self-adjoints

Theorem. The spectrum of a self-adjoint element,

a = a∗

consists of real numbers.

Proof. This follows by using f (z) = (z + it)/(z − it), with t ∈ R.


Indeed, for t >> 0 the element f (a) is well-defined, and:
 ∗  −1
a + it a − it a + it
= =
a − it a + it a − it

Thus f (a) is unitary, with spectrum contained in T. We conclude


that f (σ(a)) = σ(f (a)) ⊂ T, and so σ(a) ⊂ f −1 (T) = R.
Spectral radius 1/2

Definition. Given an element a ∈ A, its spectral radius ρ(a) is the


radius of the smallest disk centered at 0 containing σ(a).

Theorem. The spectral radius of a normal element,

aa∗ = a∗ a

equals its norm.

Proof. We already know that ρ(a) ≤ ||a||, for any a ∈ A.


Spectral radius 2/2

For the converse, if we fix ρ > ρ(a), we have:


∞
!
zn
Z X Z
dz = z n−k−1 dz ak = an−1
|z|=ρ z −a |z|=ρ
k=0

By applying the norm and taking n-th roots we obtain:

ρ ≥ lim ||an ||1/n


n→∞

(1) In the case a = a∗ we have ||an || = ||a||n for any exponent of


the form n = 2k , and by taking n-th roots we get ρ ≥ ||a||.

(2) In general we have an (an )∗ = (aa∗ )n , so ρ(a)2 = ρ(aa∗ ). Now


since aa∗ is self-adjoint, ρ(aa∗ ) = ||a||2 , and we are done.

You might also like