Lecture 2

de Investigación
AboutGrupo
the Mathematical
Foundation of Quantum Mechanics
M. Victoria Velasco Collado
Departamento de Análisis Matemático
Universidad de Granada (Spain)
Operator Theory and The Principles of Quantum Mechanics
CIMPA-MOROCCO research school,
Meknès, September 8-17, 2014
Lecture nº 2
12-09-2014
de Investigación
AboutGrupo
the Mathematical
Foundation of Quantum Mechanics
Lecture 1: About the origins of the Quantum Mechanics
Lecture 2: The mathematical foundations of Quantum Mechanics.
Lecture 3: About the future of Quantum Mechanics. Some problems and challenges
Lecture 2: The mathematical foundations of Quantum Mechanics
Hilbert spaces: orthogonality, summable families, Fourier expansion,
the prototype of Hilbert space, Hilbert basis, the Riesz-Fréchet
theorem, the weak topology.
Operators on Hilbert spaces: The adjoint operator, the spectral
equation, the spectrum, compact operators on Hilbert spaces.
CIMPA-MOROCCO
MEKNÈS, September 2014
Operator Theory and The Principles of Quantum Mechanics
From Quantum Mechanics to Functional Analysis
Definition: A normed space is a (real or complex) linear space 𝑋 equipped
with a norm, i. e. a function βˆ™ : 𝑋 β†’ ℝ satisfying
i)
π‘₯ = 0 β‡’ π‘₯ = 0 (separates points)
ii)
𝛼π‘₯ = 𝛼 π‘₯
(absolute homogeneity)
iii)
π‘₯ + 𝑦 ≀ π‘₯ + 𝑦 triangle inequality (or subadditivity).
Definition: A Banach space is a complete normed space.
Definition: A Hilbert space is a inner product Banach space 𝐻. That is a
Banach space 𝐻 whose norm is given by π‘₯ = π‘₯, π‘₯ , for every π‘₯ ∈ 𝐻,
where βˆ™,βˆ™ is an inner product, i. e. a mapping βˆ™,βˆ™ : 𝐻 × π» β†’ K such that:
i) π‘₯, π‘₯ = 0 β‡’ π‘₯ = 0
ii) π‘₯, π‘₯ β‰₯ 0
iii) 𝛼π‘₯ + 𝛽𝑦, 𝑧 = 𝛼 π‘₯, 𝑧 + 𝛽 𝑦, 𝑧
iv) π‘₯, 𝑦 = 𝑦, π‘₯
Note: If 𝐻 is a real space, then the conjugation is the identity map.
Hilbert spaces
Examples of Banach spaces and Hilbert spaces:
a)
b)
c)
d)
e)
f)
g)
ℝ, β„‚ (Hilbert)
Espacios de Banach
ℝ𝑛 , ℂ𝑛 (Hilbert)
Espacios de Hilbert
Sequences spaces 𝑙𝑝 (Hilbert si 𝑝 = 2)
Spaces of continuous functions 𝐢,π‘Ž, 𝑏Spaces of integrable functions 𝐿𝑝 ,π‘Ž, 𝑏- (Hilbert if 𝑝 = 2)
Matrices 𝑀𝑛×𝑛 (Hilbert)
Spaces of bounded linear operators: 𝐿 𝑋, π‘Œ , 𝐿(𝐻)
Theorem (Cauchy-Schawarz inequality): If 𝐻 is a linear space equipped with a inner
product βˆ™,βˆ™ then
𝑒, 𝑣 ≀
𝑒, 𝑒
v
𝑣, 𝑣
(𝑒, 𝑣 ∈ 𝐻)
From the Cauchy-Schawarz inequality we obtain straightforwardly the Minkowsky
inequality:
𝑒 + 𝑣, 𝑒 + 𝑣 ≀
𝑒, 𝑒 +
𝑣, 𝑣
𝑒, 𝑣 ∈ 𝐻
Theorem: If 𝐻 is a linear space equipped with a inner product
normed space with the norm 𝑒 =
𝑒, 𝑒
(𝑒 ∈ 𝐻).
βˆ™,βˆ™ , then 𝐻 is a
Hilbert spaces
An inner product space is linear space 𝐻 equipped with an inner product βˆ™,βˆ™ .
Let us rewrite Cauchy-Schwarz inequality in terms of the associated norm:
Fact: If 𝐻 is a inner product space, then
𝑒, 𝑣 ≀ 𝑒 𝑣 (𝑒, 𝑣 ∈ 𝐻)
Cauchy-Schwarz
Corollary: If 𝐻 is a inner product space, then βˆ™,βˆ™ is continuous.
This means that if 𝑒 = π‘™π‘–π‘š 𝑒𝑛 and 𝑣 = π‘™π‘–π‘š 𝑣𝑛 , then 𝑒, 𝑣 = lim 𝑒𝑛 , v𝑛 .
I fact, in terms of the associated norm, the inner product is given by:
Polarization identities: Let 𝐻 be a inner product space over K
1
If K = ℝ, then 𝑒, 𝑣 = ( 𝑒 + 𝑣 2 βˆ’ 𝑒 βˆ’ 𝑣 2 ).
If K = β„‚, then 𝑒, 𝑣 =
1
4
4
( 𝑒+𝑣
2
βˆ’ π‘’βˆ’π‘£
2 )+ 𝑖
4
( 𝑒 + 𝑖𝑣
2
βˆ’ 𝑒 βˆ’ 𝑖𝑣
Consequently, if 𝐻 is a inner product space over K (= ℝ or β„‚) then
Parallelogram identity:
𝑒+𝑣
2
+ π‘’βˆ’π‘£
2
= 2( 𝑒
2+
𝑣
2).
2 ).
Hilbert spaces
Question: Given a normed space 𝐻, howv to know if 𝐻 is an inner product space?
If this is the case, then such a norm needs to satisfy the paralelogram indetity
Theorem (Parallelogram theorem): If (𝐻, βˆ™ ) is a normed space then βˆ™ is given
by an inner product if and only of if, for every
𝑒, 𝑣 ∈ 𝐻 we have that
v
𝑒+𝑣
2
+ π‘’βˆ’π‘£
2
= 2( 𝑒
2+
𝑣
2)
(Parallelogram identity)
Therefore we know, for instance that:
𝑣
𝑙𝑝 is Hilbert space ⇔ 𝑝 = 2.
𝑒 βˆ’v
𝑒
Theorem: Any inner product space may be completed to a Hilbert space.
Proof (sketch): If 𝑒 = π‘™π‘–π‘š 𝑒𝑛 and 𝑣 = π‘™π‘–π‘š 𝑣𝑛 (for 𝑒, 𝑣 ∈ 𝐻) then define
𝑒, 𝑣 : = π‘™π‘–π‘š 𝑒𝑛, 𝑣𝑛 .
It happens that this not depend of the choice of the sequences. Moreover,
𝑒 = π‘™π‘–π‘š 𝑒𝑛 = π‘™π‘–π‘š 𝑒𝑛 , 𝑒𝑛 = 𝑒, 𝑒
If follows that 𝐻 is Hilbert space from the parallelogram identity.
Hilbert spaces: orthogonality
Law of cosines (Euclides): π‘Ž2 = 𝑏 2 + 𝑐 2 βˆ’ 2𝑏𝑐 π‘π‘œπ‘ πœƒ
𝑏
π‘Ž
πœƒ
𝑐
Consequently:
π‘π‘œπ‘ πœƒ =
𝑒 2 + 𝑣 2 βˆ’ π‘’βˆ’π‘£ 2
2 𝑒 𝑣
=
𝑒,𝑒 + 𝑣,𝑣 βˆ’ π‘’βˆ’π‘£,π‘’βˆ’π‘£
2 𝑒 𝑣
=
2 𝑒,𝑣
2 𝑒 𝑣
=
𝑒
𝑒
,
𝑣
𝑣
π‘’βˆ’π‘£
𝑣
πœƒ
Fact : 𝑒 βŠ₯ 𝑣 ⟺ π‘π‘œπ‘ πœƒ=0 ⟺ 𝑒, 𝑣 =0
𝑣
Definition: Let 𝐻 be an inner space. It is said that 𝑒, 𝑣 ∈ 𝐻 are orthogonal vectors
if 𝑒, 𝑣 =0. If this is the case, then we write 𝑒 βŠ₯ 𝑣.
Definition: Let 𝐻 be an inner space. The orthogonal set of S βŠ† 𝐻 is defined by
𝑆 βŠ₯ ≔ 𝑒 ∈ 𝐻: 𝑒 βŠ₯ s, βˆ€π‘  ∈ 𝑆 .
Proposition: If 𝐻 is an inner space, and if S, W βŠ† 𝐻 then:
i) 0 ∈ 𝑆 βŠ₯
iv) 𝑆 βŠ† π‘Š ⟹ π‘Š βŠ₯ βŠ† 𝑆 βŠ₯
ii) 0 ∈ 𝑆 ⟹ 𝑆 ∩ 𝑆 βŠ₯ = 0
v) 𝑆 βŠ₯ is a closed linear subspace of H
iii) *0+βŠ₯ = 𝐻;
𝐻βŠ₯ = 0
vi) 𝑆 βŠ† 𝑆 βŠ₯βŠ₯
Hilbert spaces: orthogonality
Definition: Let 𝑋 be a linear space, and let π‘Œ and 𝑍 be linear subspaces. Then:
𝑋 = π‘Œβ¨π‘ ⟺ 𝑋 = π‘Œ + 𝑍 ; Y ∩ 𝑍 = 0 . (Direct sum)
Let 𝑋 be a normed space such that 𝑋 = π‘Œβ¨π‘.
Let π‘₯ = 𝑦 + 𝑧 and π‘₯𝑛 = 𝑦𝑛 + 𝑧𝑛 in π‘Œβ¨π‘.
We hope that : π‘₯𝑛 β†’ π‘₯ ⟺ 𝑦𝑛 β†’ π‘₯ y 𝑧𝑛 β†’ 𝑧.
If this is the case, then we say that 𝑋 = π‘Œβ¨π‘ is a topological direct sum.
Fact: If 𝑋 is a Banach space then,
𝑋 = π‘Œβ¨π‘ is a topological direct sum ⟺ the subspaces π‘Œ and 𝑍 are closed
Theorem (best approximation): Let 𝐻 be a Hilbert space, 𝑒 ∈ 𝐻, and 𝑀 a closed
subspace of 𝐻. Then, there exists a unique π‘š ∈ 𝑀 such that 𝑒 βˆ’ π‘š = 𝑑 𝑒, 𝑀 .
Notation: 𝑃𝑀 𝑒 = π‘š ∈ 𝑀: 𝑒 βˆ’ π‘š = 𝑑 𝑒, 𝑀
= best approximation from 𝑒 to 𝑀.
Orthogonal projection theorem: Let 𝐻 be Hilbert space and M a closed subspace.
(Topological direct sum)
Then, 𝐻 = 𝑀 ⨁ 𝑀 βŠ₯ .
Morever if πœ‹π‘€ : 𝐻 β†’ 𝑀 is the canonical projection then πœ‹π‘€ = 𝑃𝑀 . Also πœ‹π‘€ = 1
and 𝑒 = πœ‹π‘€ (𝑒) + 𝑒 βˆ’ πœ‹π‘€ (𝑒) = 𝑃𝑀 (𝑒) + 𝑒 βˆ’ 𝑃𝑀 (𝑒) .
Remmark: Results also know as Hilbert projection theorem
Hilbert spaces: orthogonality
Fact: Every closed subspace of a Hilbert space admits a topological complement.
This is a very important property. Indeed it characterizes the Hilbert spaces
Theorem (Lindestrauss-Tzafriri, 1971): Assume that every closed subspace of a
Banach space 𝑋 is complemented. Then 𝑋 is isomophic to a Hilbert space.
To be complemented means to admit a topological complement
.
Hilbert spaces: orthogonality
Definition: Let *𝑒𝑖 +π‘–βˆˆπΌ be a family of nonzero vectors of 𝐻. We say that *𝑒𝑖 +π‘–βˆˆπΌ is a
orthogonal family if 𝑒𝑖 , 𝑒𝑗 = 0 βˆ€π‘– β‰  𝑗.
If in addition 𝑒𝑖 = 1, βˆ€π‘– ∈ 𝐼, then we say that *𝑒𝑖 +π‘–βˆˆπΌ is a ortonormal family.
Proposición: Every orthogonal family of vectors in 𝐻 is linearly independent.
The Gram–Schmidt process is a method for orthonormalising a set of vectors in an
inner product space. It applies to a linearly independent countably infinite (or finite)
family of vectors in a inner space.
Proposition (Gram–Schmidt process): Let *𝑣𝑛 +π‘›βˆˆβ„• be a linearly independent family
of vectors in a inner space 𝐻. Then the family *𝑒𝑛 +π‘›βˆˆβ„• given by
𝑒1 = 𝑣1
π‘›βˆ’1
𝑒𝑛 = 𝑣𝑛 βˆ’
𝑗=1
is a orthogonal family in 𝐻.
Morever the family *𝑒𝑛 +π‘›βˆˆβ„• given by 𝑒𝑛 =
𝑒𝑛
𝑒𝑛
𝑒𝑗 , 𝑣𝑛
𝑒𝑗 , 𝑒𝑗
𝑒𝑗
is orthonormal.
Hilbert spaces : summable families
Definition: A family *π‘₯𝑖 +π‘–βˆˆπΌ in a normed space 𝑋 is summable if there exists π‘₯ ∈ 𝑋
such that βˆ€πœ€ > 0 there exists a finite set π½πœ€ βŠ† 𝐼, such that if J βŠ† 𝐼 is finite and
π½πœ€ βŠ† 𝐽 then
π‘–βˆˆπ½ π‘₯𝑖 βˆ’ π‘₯ < πœ€. If this is the case we write π‘₯ = π‘–βˆˆπΌ π‘₯𝑖 .
Proposition: A family of positive real numbers *π‘₯𝑖 +π‘–βˆˆπΌ is summable if and only if the
set * π‘–βˆˆπ½ π‘₯𝑖 : J βŠ† 𝐼 ; 𝐽 π‘“π‘–π‘›π‘–π‘‘π‘œ+ is bounded. If this is the case then,
π‘₯=
π‘₯𝑖 : J βŠ† 𝐼 ; 𝐽 π‘“π‘–π‘›π‘–π‘‘π‘œ+ .
π‘₯𝑖 = sup*
π‘–βˆˆπΌ
π‘–βˆˆπ½
Definition: A family *π‘₯𝑖 +π‘–βˆˆπΌ in a normed space 𝑋 satisfies the Cauchy condition if
βˆ€ πœ€ > 0 there exists a finite π½πœ€ βŠ† 𝐼, such that if J βŠ† 𝐼 is finite, and if 𝐽 ∩ π½πœ€ = βˆ… then,
π‘–βˆˆπ½ π‘₯𝑖 < πœ€.
Theorem: Let 𝑋 be a Banach space and *π‘₯𝑖 +π‘–βˆˆπΌ a family in 𝑋. Then,
i) *π‘₯𝑖 +π‘–βˆˆπΌ is summable ⟺ *π‘₯𝑖 +π‘–βˆˆπΌ satisfies the Cauchy condition.
ii) *π‘₯𝑖 +π‘–βˆˆπΌ summable ⟹ 𝑖 ∈ 𝐼: π‘₯𝑖 β‰  0 is countable.
1
𝑛
Remark: Apply the Cauchy condition with πœ€ = . Then we obtain 𝐽 1 such that
π‘–βˆˆπ½ π‘₯𝑖
then,
<
1
𝑛
π‘₯𝑖 <
if 𝐽 ∩ π½πœ€ = βˆ…. Hence, 𝐼0 ≔
1
,
𝑛
βˆ€π‘›, so that π‘₯𝑖 = 0.
𝑛
𝐽 1 is a countable set. Moreover if 𝑖 ∈ 𝐼 βˆ– 𝐼0
𝑛
Hilbert spaces : summable families
Theorem: Let *𝑒𝑖 +π‘–βˆˆπΌ be a orthogonal family *𝑒𝑖 +π‘–βˆˆπΌ in a Hilbert space 𝐻. Then,
*𝑒𝑖 +π‘–βˆˆπΌ summable (in 𝐻) ⟺ * 𝑒𝑖 2 +π‘–βˆˆπΌ is summable en ℝ,
in whose case:
π‘–βˆˆπΌ 𝑒𝑖
2
=
π‘–βˆˆπΌ
𝑒𝑖
2
.
This result addresses the problem of the summability in 𝐻 to the problem of the
summability in ℝ. On the other hand, it generalizes the Pythagoras's theorem.
Remark: The canonical base of ℝ3 is orthogonal. Moreover if π‘₯ ∈ ℝ3 is such that
π‘₯ = (π‘₯1 , π‘₯2 , π‘₯3 ), that is π‘₯ = 𝑖=3
𝑖=1 π‘₯𝑖 𝑒𝑖 , then
𝑖=3
π‘₯=
π‘₯, 𝑒𝑖 𝑒𝑖
𝑖=1
The idea is to generalize this fact to an arbitrary 𝐻 to have β€œcoordinates”.
Remark: Let *𝑒𝑖 +π‘–βˆˆπΌ be a othonormal family in a Hilbert space 𝐻. Then
*𝑓(𝑒𝑖 )𝑒𝑖 +π‘–βˆˆπΌ summable in 𝐻 ⟺ * 𝑓(𝑒𝑖 ) 2 +π‘–βˆˆπΌ es summable in ℝ .
If this is the case, then:
π‘–βˆˆπΌ 𝑓(𝑒𝑖 )𝑒𝑖
2
=
π‘–βˆˆπΌ
𝑓(𝑒𝑖 ) 2 .
Particularly: * π‘₯, 𝑒𝑖 𝑒𝑖 +π‘–βˆˆπΌ summable in 𝐻 ⟺ * π‘₯, 𝑒𝑖
2+
π‘–βˆˆπΌ
summable ℝ .
Hilbert spaces : summable families
When * π‘₯, 𝑒𝑖
2+
π‘–βˆˆπΌ
is summable in ℝ ? When is π‘₯ =
π‘–βˆˆπΌ
π‘₯, 𝑒𝑖 𝑒𝑖 ?
Let 𝐽 βŠ† 𝐼, such that 𝐽 is finite. Then
2
0≀ π‘₯βˆ’
π‘₯, 𝑒𝑖 𝑒𝑖
= π‘₯βˆ’
π‘–βˆˆπ½
Therefore, π‘–βˆˆπ½ π‘₯, 𝑒𝑖
Consequently
π‘–βˆˆπΌ
π‘₯, 𝑒𝑖. 𝑒𝑖 , π‘₯ βˆ’
π‘–βˆˆπ½
2
𝑒, 𝑒𝑖
This means that * π‘₯, 𝑒𝑖
≀ π‘₯
2
2
2
βˆ’
π‘–βˆˆπ½
π‘₯, 𝑒𝑖
2
π‘–βˆˆπ½
for every 𝐽 βŠ† 𝐼, with 𝐽 finite.
= sup*
2+
π‘–βˆˆπΌ
π‘₯, 𝑒𝑖 𝑒𝑖 =β‹…β‹…β‹…= π‘₯
π‘–βˆˆπ½
𝑒, 𝑒𝑖
2:
J βŠ† 𝐼 ; 𝐽 𝑓𝑖𝑛𝑖𝑑e+ ≀
π‘₯
2
is summable in ℝ and hence we have the following:
Theorem (Bessel's inequality): If *𝑒𝑖 +π‘–βˆˆπΌ is a orthormal family in a Hilbert space 𝐻,
then for every 𝑒 ∈ 𝐻, the family π‘–βˆˆπΌ 𝑒, 𝑒𝑖 𝑒𝑖 is summable. Moreover
𝑒, 𝑒𝑖 𝑒𝑖 ≀ 𝑒
π‘–βˆˆπΌ
Hilbert spaces : summable families
Corollary (characterization of the maximal ortonormal families): Let *𝑒𝑖 +π‘–βˆˆπΌ be an
orthonormal system in a Hilbert space 𝐻. The following conditions are equivalent:
i) 𝑒 = π‘–βˆˆπΌ 𝑒, 𝑒𝑖 𝑒𝑖
ii) 𝑒, 𝑣 = π‘–βˆˆπΌ 𝑒, 𝑒𝑖 𝑒𝑖 , 𝑣 (Parseval’s identy)
2
iii) 𝑒 =
.
π‘–βˆˆπΌ 𝑒, 𝑒𝑖
iv) *𝑒𝑖 +π‘–βˆˆπΌ is a maximal orthonormal maximal
v) 𝑒 βŠ₯ 𝑒𝑖 βˆ€π‘– β‡’ 𝑒 = 0
vi) 𝐻 = 𝑙𝑖𝑛 𝑒𝑖 : 𝑖 ∈ 𝐼 .
Proof: TPO+Bessel: 𝑒 ∈ 𝐻 = 𝑙𝑖𝑛 𝑒𝑖 : 𝑖 ∈ 𝐼 ⟺ 𝑒 =
The other assertions are easy to prove.
π‘–βˆˆπΌ
𝑒, 𝑒𝑖 𝑒𝑖 .
Fact: By Zorn’s lemma there exist maximal ortonormal basis.
Definition: A Hilbert basis (or an orthonormal basis) in 𝐻 is a maximal orthonormal
family of vectors in 𝐻.
Example: *𝑒𝑛 +π‘›βˆˆβ„• is an orthonormal family in 𝑙2 = *𝑒 = 𝛼𝑛 π‘›βˆˆβ„• ∢
𝛼𝑛 2 < ∞+
such that 𝑒 βŠ₯ 𝑒𝑛 βˆ€π‘› β‡’ 𝑒 = 0. Therefore 𝐡 ≔ *𝑒𝑛 : 𝑛 ∈ β„•+ is a Hilbert basis of 𝑙2.
Note that 𝐡 is not a Hamel basis. Inndeed the linear span of 𝐡 is 𝑐00 (sequences
with finite support). Similarly with 𝑙2(I) where 𝑙2 = 𝑙2(β„•).
Hilbert spaces : Fourier expansion
Theorem: Every Hilbert space 𝐻 has an orthonormal basis. Moreover all the
orthonormal bases of 𝐻 have have the same cardinal.
Definition: The Hilbert space dimension of a Hilbert space 𝐻 is the cardinality of an
orthonormal basis of 𝐻.
Remark: Denote by dim𝐻 the algebraic dimension of 𝐻.
If dim𝐻 < ∞ , then 𝐡 = 𝑒1 ,…, 𝑒𝑛 is a Hilbert basis ⟺ 𝐡 is a Hamel basis.
Indeed,
𝐻 = 𝑙𝑖𝑛*𝑒1 ,…, 𝑒𝑛 + = 𝑙𝑖𝑛*𝑒1 ,…, 𝑒𝑛 +.
If dim𝐻 = ∞, and if 𝐡 = *𝑒𝑖 +π‘–βˆˆπΌ is a Hilbert basis then we have 𝐻 = 𝑙𝑖𝑛*𝑒𝑖 : 𝑖 ∈ 𝐼+ .
But *𝑒𝑖 +π‘–βˆˆπΌ do not need to be a spanning set of 𝐻. Therefore
Hilbert basis ⇏ Hamel basis
Definition: Let 𝐻 be a Hilbert space and *𝑒𝑖 +π‘–βˆˆπΌ a Hilbert basis. For every 𝑒 ∈ 𝐻,
𝑒 = π‘–βˆˆπΌ 𝑒, 𝑒𝑖 𝑒𝑖 (Fourier expansion)
Fact: The Fourier coefficients * 𝑒, 𝑒𝑖 +π‘–βˆˆπΌ are the unique family of scalars *𝛼𝑖 +π‘–βˆˆπΌ
such that 𝑒 = π‘–βˆˆπΌ 𝛼𝑖 𝑒𝑖 vector coordinates in ∞-dim!!
Hilbert spaces : Fourier expansion
Example: Let
be the Lebesgue 𝜎-algebra of βˆ’πœ‹, πœ‹ and let π‘š be the
π‘š
π‘š
Lebesgue measure. Let πœ‡ =
if K = ℝ, and πœ‡ =
if K = β„‚ . Then
πœ‹
2πœ‹
The family 𝑒 𝑖𝑛𝑑 : 𝑛 ∈ β„€ is a Hilbert basis of 𝐿2 ( βˆ’πœ‹, πœ‹ , , πœ‡)β„‚ . It is known as the
trigonometric system. Therefore, for 𝑓 ∈ 𝐿2 βˆ’πœ‹, πœ‹ we have
𝑓 𝑑 = π‘›βˆˆβ„€ 𝑓 (𝑛) 𝑒 𝑖𝑛𝑑 (Fourier expansion)
convergence in βˆ™ 2, where
1 πœ‹
𝑓 (𝑛):= βˆ«βˆ’πœ‹ 𝑓(𝑑)𝑒 βˆ’π‘–π‘›π‘‘ 𝑑𝑑
2πœ‹
1
Similarly, the family , cos 𝑛𝑑 , sin(𝑛𝑑) : 𝑛 ∈ β„• is a Hilbert basis of
2
𝐿2 ( βˆ’πœ‹, πœ‹ , , πœ‡)ℝ. Therefore, for 𝑓 ∈ 𝐿2 βˆ’πœ‹, πœ‹ we have
π‘Ž
𝑓(𝑑) = 0+ ∞
𝑛=1(π‘Žπ‘› cos 𝑛𝑑 + 𝑏𝑛 sen 𝑛𝑑 ) (Fourier expansion)
2
Convergence in βˆ™ 2, where:
1
πœ‹
π‘Ž0 = βˆ«βˆ’πœ‹ 𝑓(𝑑) 𝑑𝑑 ;
πœ‹
1
πœ‹
π‘Žπ‘› = βˆ«βˆ’πœ‹ 𝑓(𝑑) cos 𝑛𝑑 𝑑𝑑 ;
πœ‹
1
πœ‹
𝑏𝑛 = βˆ«βˆ’πœ‹ 𝑓(𝑑) sen 𝑛𝑑 𝑑𝑑
πœ‹
Hilbert spaces : The prototype
Fact: If 𝐻 is a Hilbert space and 𝐡 ≔ *𝑒𝑖 : 𝑖 ∈ 𝑆+ is a Hilbert basis then, every
𝑒 ∈ 𝐻 has a unique expansion given by
𝑒 = π‘–βˆˆπΌ 𝑒, 𝑒𝑖 𝑒𝑖
and
2.
𝑒 =
π‘–βˆˆπΌ 𝑒, 𝑒𝑖
Consequently, the mapping 𝒖 β†’ 𝒖, π’†π’Š π’Šβˆˆπ‘° defines an isometric isomorphism
from 𝑯 onto π’πŸ(𝑰). Therefore, 𝑙2(𝐼) becomes the Hilbert space prototype.
Example: The mapping 𝐿2 βˆ’πœ‹, πœ‹
isometric isomorphism.
β„‚
β†’ 𝑙2(β„€), given by 𝑓 β†’ 𝑓 (𝑛)
nβˆˆβ„€
defines an
For this reason the Hilbert spaces have β€œcoordinates”, as well as the euclidean
space (which is also a Hilbert space) has its own coordinates.
Theorem: Two Hilbert spaces are isometrically isomorphic, if and only if, they have
the same Hilbert dimension.
(that is that an orthonormal basis of one of this spaces has the same cardinality
of an orthonormal basis of the other one).
Hilbert spaces: the Riesz-Fréchet theorem
Let 𝐻 be a Hilbert space over K. Who is 𝐻 βˆ— ?
This is to determine the set of continuous
linear functionals 𝑓: 𝐻 β†’ K.
For general Banach spaces, this was open problem along
more of 20 years. The answer was the Hahn-Banach
theorem (a cornerstone theorem of the Functional Analysis).
Fréchet, Maurice
1878-1973
If dim 𝐻 = 𝑛 then 𝐻 = K𝑛 . Fix an orthonormal basis 𝐡 = *𝑒1 ,…, 𝑒𝑛 +. If 𝑓: 𝐻 β†’ K
is linear then, for 𝑒 = 𝑛𝑖=1 𝛼𝑖 𝑒𝑖 , we have
𝑓 𝑒 =
𝑛
𝑖=1 𝛼𝑖 𝑓(𝑒𝑖 )
= 𝑒, 𝑣
where 𝑣 = 𝑛𝑖=1 𝑓(𝑒𝑖 )𝑒𝑖 . Therefore 𝑓 = βˆ™, 𝑣 .
Conversely, 𝑓 = βˆ™, 𝑣 is a continuous linear functional, obviously. Thus,
𝐻 βˆ— = βˆ™, 𝑣 : 𝑣 ∈ 𝐻 .
If π‘‘π‘–π‘šπ» = ∞ then, βˆ™, 𝑣 : 𝑣 ∈ 𝐻 βŠ† 𝐻 βˆ— trivially. What F. Riesz and M. Fréchet
showed (in independent papers published in Comptes Redus in 1907) is that, in
fact, βˆ™, 𝑣 : 𝑣 ∈ 𝐻 = 𝐻 βˆ— .
Note also that if 𝑓𝑣 : = βˆ™, 𝑣 , then
𝑓𝑣 = 𝑣 .
Espacios de Hilbert: El teorema de Riesz-Fréchet
Riesz-Fréchet theorem (1907): Let 𝐻 be a Hilbert space and let 𝑓 ∈ 𝐻 βˆ— . Then, there
exists a unique 𝑣 ∈ 𝐻 such that 𝑓 = 𝑓𝑣 = βˆ™, 𝑣 . Moreover, 𝑓 = 𝑣 .
Proof: Let 𝑓 ∈ 𝐻 βˆ— . If 𝑓 = 0, then obviously 𝑓 = 𝑓0 = βˆ™, 0 . If 𝑓 β‰  0, then ker𝑓 is a
proper closed subspace of 𝐻, so that
ker𝑓 βŠ₯ is nonzero. Therefore, by the OPT
there exists πœ” ∈ ker𝑓
βŠ₯
such that 𝑓(πœ”) β‰  0. Now it is easy to check that 𝑣 =
𝑓(πœ”)
πœ” 2
πœ”
is the vector that we look for, and the result follows.
Corollary: If 𝐻 is a Hilbert space, then there exists an isometric conjugate-linear
bijection between 𝐻 and 𝐻 βˆ— .
Proof: Consider the map 𝐻 βˆ— β†’ 𝐻 given by 𝑓𝑣 = βˆ™, 𝑣 β†’ 𝑣.
Corollary: If 𝐻 is a Hilbert space, then 𝐻 βˆ— is also a Hilbert space.
Proof: Define 𝑓𝑒 , 𝑓𝑣 ≔ 𝑒, 𝑣 and check it.
Corollary: Every Hilbert space is reflexive (i. e. 𝐻 = 𝐻 βˆ—βˆ— ).
Corollary: Any inner product space may be completed to a Hilbert space.
Hilbert spaces: The weak topology
Corollary: If 𝐻 is a Hilbert space and 𝑀 is a subspace, then every bounded linear
functional 𝑓: 𝑀 β†’K admits a bounded linear extension 𝑓 : 𝐻 β†’K with 𝑓 = 𝑓 .
Bounded linear functionals in a Hilbert space do exist in abundance, as we have
shown. Consider the smallest topology that makes that all of them are continuous.
This is the so-called weak topology (the strong topology is the norm topology).
Unless that dim 𝐻 < ∞ , the weak topology is not normable.
Definition: A sequence 𝑒𝑛 , in a Hilbert 𝐻, converges weakly to 𝑒 ∈ 𝐻 (and we
πœ”
write 𝑒𝑛 β†’ 𝑒 ) whenever 𝑒𝑛 , 𝑣 β†’ 𝑒, 𝑣 , for every 𝑣 ∈ 𝐻.
From the continuity of the inner product we deduce the following result.
Corollary: If in a Hilbert space, 𝑒𝑛
βˆ™
πœ”
𝑒 β‡’ 𝑒𝑛 β†’ 𝑒.
The converse does not hold. For instance, in 𝑙2 we have that 𝑒𝑛 , 𝑣 β†’0 = 0, 𝑣 ,
πœ”
so that 𝑒𝑛 β†’ 0, meanwhile 𝑒𝑛 βˆ’ π‘’π‘š = 2.
Corollary: The (norm)-closed unit ball in a Hilbert space is weakly compact.
Operators in Hilbert spaces: The adjoint operator
Let 𝐻1 and 𝐻2 be Hilbert spaces and let 𝑇 ∈ 𝐿 𝐻1 , 𝐻2 . As a consequence of the
Riesz-Fréchet theorem, there exists a unique bounded linear operator 𝑇 βˆ— : 𝐻2 β†’ 𝐻1
such that
𝑇𝑒, πœ” = 𝑒, 𝑇 βˆ— πœ” ,
(𝑒 ∈ 𝐻1 , πœ” ∈ 𝐻2).
Note that
π‘‡βˆ— πœ”
2
= π‘‡βˆ— πœ” , π‘‡βˆ— πœ”
= 𝑇𝑇 βˆ— πœ” , πœ” ≀ 𝑇
π‘‡βˆ— πœ”
πœ” ,
Therefore
π‘‡βˆ— πœ”
Consequently, 𝑇 βˆ— is continuous and
≀ 𝑇
πœ” .
π‘‡βˆ— ≀ 𝑇 .
Definition: Let 𝐻1 and 𝐻2 be Hilbert spaces and let 𝑇: 𝐻1 β†’ 𝐻2 be a bounded linear
operator. The adjoint operador of 𝑇 is defined as the unique bounded linear
operator 𝑇 βˆ— : 𝐻2 β†’ 𝐻1 such that 𝑇𝑒, πœ” = 𝑒, 𝑇 βˆ— πœ” , (𝑒 ∈ 𝐻1 , πœ” ∈ 𝐻2).
Fact: 𝑇 βˆ—βˆ— = 𝑇. Indeed, if 𝑒 ∈ 𝐻2 , and πœ” ∈ 𝐻1, then:
𝑇 βˆ— 𝑒, πœ” = πœ”, 𝑇 βˆ— 𝑒 = π‘‡πœ”, 𝑒 = 𝑒, π‘‡πœ” = 𝑒, 𝑇 βˆ—βˆ— πœ” .
Since 𝑇 βˆ—βˆ— πœ” is unique, we obtain that 𝑇 βˆ—βˆ— πœ” = π‘‡πœ”.
Operators in Hilbert spaces: The adjoint operator
Proposition: Let 𝐻1 and 𝐻2 be Hilbert spaces and let 𝑇, 𝑆 ∈ 𝐿(𝐻1 , 𝐻2 ). Then:
i) 𝑇 βˆ—βˆ— = 𝑇
ii) 𝛼𝑇 βˆ— = 𝛼𝑇 βˆ— (𝛼 ∈ K)
iii) 𝑇 + 𝑆 βˆ— = 𝑇 βˆ— + 𝑆 βˆ—
Moreover, if 𝐻3 is a Hilbert space and 𝑅 ∈ 𝐿(𝐻2 , 𝐻3 ), then (𝑅𝑇)βˆ— = 𝑇 βˆ— 𝑅 βˆ— .
Since 𝑇 βˆ—βˆ— = 𝑇 and
Indeed:
𝑇 βˆ—βˆ— ≀ 𝑇 βˆ— , we obtain that
π‘‡βˆ— = 𝑇 .
Proposition: Let 𝐻1 and 𝐻2 be Hilbert spaces and let 𝑇 ∈ 𝐿(𝐻1 , 𝐻2 ). Then,
π‘‡βˆ— = 𝑇 =
π‘‡βˆ—π‘‡ =
𝑇𝑇 βˆ— (Gelfand-Naimark).
Fact: The completeness is essential for the existence of the adjoint.
Example: Let 𝑇: 𝑐00 β†’ 𝑐00 be the operador 𝑇π‘₯ =
π‘₯ = π‘₯1 , π‘₯2 … . . π‘₯𝑛 , … β†’ 𝑇π‘₯ = *
∞ π‘₯𝑛
𝑛=1 𝑛 𝑒1 . That is
∞ π‘₯𝑛
𝑛=1 𝑛 , 0 … , 0, … +.
Let 𝑦 = *𝑦𝑛 + with 𝑦1 β‰  0. Let 𝑇 βˆ— 𝑦 = 𝑧𝑛 . Then 𝑇 βˆ— 𝑦 βˆ‰ 𝑐00.
Operators in Hilbert spaces: The adjoint operator
From the above result we obtain straightforwardly the following proposition.
Proposition: Let 𝐻1 and 𝐻2 be Hilbert spaces and let 𝑇 ∈ 𝐿(𝐻1 , 𝐻2 ).
Then,
i) 𝑇 βˆ— is injective ⟺ 𝑇 has dense range.
(Ker 𝑇 βˆ— = (Im 𝑇)βŠ₯ )
ii) 𝑇 is injective ⟺ 𝑇 βˆ— has dense range. (Ker 𝑇 = (Im𝑇 βˆ— )βŠ₯ )
iii) 𝑇 is biyective ⟺ T βˆ— is bijective, in whose case (T βˆ— )βˆ’1 = (T βˆ’1 )βˆ— .
Notation: 𝐿(𝐻) = 𝐿(𝐻, 𝐻).
Examples: Let 𝐻 be a Hilbert space and let B = *𝑒𝑖 +π‘–βˆˆπΌ be a Hilbert basis.
i) If 𝑇 ∈ 𝐿 𝐻 is diagonal (i.e. 𝑇𝑒𝑖 = πœ†π‘– 𝑒𝑖 ), then 𝑇 βˆ— 𝑒𝑖 = πœ†π‘– 𝑒𝑖 .
ii) If 𝑇 ∈ 𝐿 𝐻 is the orthogonal projection on a closed subspace, then 𝑇 = 𝑇 βˆ— .
Examples:
i) If, repect to the basis B, the matrix associated to 𝑇 ∈ 𝐿 𝑙2 is π‘Žπ‘–π‘— , then the
matrix associated to T βˆ— is 𝑏𝑖𝑗 = π‘Žπ‘–π‘— .
ii) If 𝑇 ∈ 𝐿 𝐿2 π‘Ž, 𝑏 is an integral operator with kernel π‘˜ ∈ 𝐿2 ( π‘Ž, 𝑏 × π‘Ž, 𝑏 ), then
𝑇 βˆ— is an integral operator with kernel π‘˜ βˆ— 𝑑, 𝑠 = π‘˜ 𝑠, 𝑑 .
Operators in Hilbert spaces: The adjoint operator
Fact: From now on, we will consider operators from a Hilbert space into itself, that is
the space 𝐿 𝐻 : = 𝐿 𝐻, 𝐻 of bounded linear operators 𝑇: 𝐻 β†’ 𝐻.
Recall that if 𝑇 ∈ 𝐿 𝐻 , then 𝑇 βˆ— is the unique operator 𝐿(𝐻) such that
𝑇𝑒, πœ” = 𝑒, 𝑇 βˆ— πœ” , (𝑒, 𝑣 ∈ 𝐻).
Definition: 𝑇 ∈ 𝐿(𝐻) is a self-adjoint operator if 𝑇 = 𝑇 βˆ— .
Proposición: Let S, 𝑇 ∈ 𝐿 𝐻 be selft-adjoint operators. Then,
i) 𝑆 + 𝑇 is selft-adjoint.
ii) 𝑆𝑇 selft-adjoint ⟺ 𝑆𝑇 = 𝑇𝑆.
Proposition: Let 𝑇 ∈ 𝐿 𝐻 . If 𝑇 = 𝑇 βˆ— , then 𝐻 = ker 𝑇 ⨁ πΌπ‘šπ‘‡.
Proposition: Let 𝐻 be a complex Hilbert space and let 𝑇 ∈ 𝐿 𝐻 . Then:
𝑇 = 𝑇 βˆ— ⟺ 𝑇𝑒, 𝑒 ∈ ℝ, for every 𝑒 ∈ 𝐻.
Moreover, if 𝑇 is selft-adjoint, then
𝑇 = sup 𝑇𝑒, 𝑒 : 𝑒 ≀ 1 = sup*|⟨Tu, u⟩|: β€–uβ€– = 1+.
Operators in Hilbert spaces: The adjoint operator
Proposition: Let 𝐻 be a complex Hilbert space and let 𝑇 ∈ 𝐿(𝐻). Then, there exist
self-adjoint operators R, S ∈ 𝐿 H (which are unique) such that 𝑇 = 𝑅 + 𝑖𝑆.
Proof: 𝑇 =
𝑇+𝑇 βˆ—
2
+
π‘‡βˆ’π‘‡ βˆ—
𝑖
.
2𝑖
The unique self-adjoint operators R, S ∈ 𝐿 𝐻 , such that 𝑇 = 𝑅 + 𝑖𝑆, are called
the real part and the imaginary part of 𝑇, respectivelty.
Definition: 𝑇 ∈ 𝐿 H is called normal if 𝑇𝑇 βˆ— = 𝑇 βˆ— 𝑇.
𝑇 ∈ 𝐿 𝐻 self-adjoint ⟹ 𝑇 normal
𝑇 ∈ 𝐿(𝐻) diagonal ⟹ 𝑇 normal
Proposition: Let 𝑇 ∈ 𝐿 H . Then 𝑇 normal ⟺
𝑇𝑒 = 𝑇 βˆ— 𝑒
π‘’βˆˆπ» .
Proposition: Let 𝐻 be a complex Hilbert space. Then 𝑇 ∈ 𝐿 𝐻 is normal if, and
only if, its real and imaginary parts commute.
Proposition: If 𝑇 ∈ 𝐿(𝐻) is such that 𝑇 2 = 𝑇 (i.e. 𝑇 is a projection), then,
T normal ⟺ T is a orthogonal projection ⟺ 𝑇 = 1.
Operators in Hilbert spaces: The spectral equation
Goal: To solve the spectral equation 𝑇 βˆ’ πœ†πΌ π‘₯ = 𝑦.
Let 𝑇: ℂ𝑛 β†’ ℂ𝑛 be a linear operator. If 𝐴 is the matrix associated to 𝑇 respect to the
canonical basis, then we like to solve the equation
𝐴 βˆ’ πœ†πΌ π‘₯ = 𝑦,
where 𝐴 is an 𝑛 × π‘› matrix, 𝐼 is the identity matrix, and π‘₯ and 𝑦 are the
coordinates of the corresponding vectors (where 𝑦 is known and π‘₯ is not).
If det 𝐴 βˆ’ πœ†πΌ β‰  0, then 𝐴 βˆ’ πœ†πΌ is an invertible matrix and the given equation has
a unique solution given by π‘₯ = 𝐴 βˆ’ πœ†πΌ βˆ’1 𝑦.
If det 𝐴 βˆ’ πœ†πΌ = 0, then 𝑇 βˆ’ πœ†πΌ is not injective, nor surjective. Therefore the
system has a solution (not unique) if, and only if, 𝑦 ∈ πΌπ‘š 𝑇 βˆ’ πœ†πΌ .
Thus, determining the set of all πœ† ∈ β„‚ such that 𝑇 βˆ’ πœ†πΌ is not invertible is relevant:
*πœ† ∈ β„‚ ∢ det 𝐴 βˆ’ πœ†πΌ = 0+.
Definition: 𝑇 ∈ 𝐿 𝐻 is invertible if there exists 𝑅 ∈ 𝐿 𝐻 such that 𝑇𝑅 = 𝑅𝑇 = 𝐼.
If turns out that 𝑅 is unique (whenever it exists). We denote 𝑅 = 𝑇 βˆ’1.
Operators in Hilbert spaces: The spectrum
Definition: Let 𝐻 be a complex Hilbert space. The spectrum of 𝑇 ∈ 𝐿 H is the set
𝜎 𝑇 = *πœ† ∈ β„‚ ∢ 𝑇 βˆ’ πœ†πΌ is not invertible+
If 𝑇: ℂ𝑛 β†’ ℂ𝑛 is a linear operator, and if 𝐴 is its matrix respecto to the canonical
basis, then
𝜎 𝑇 = *πœ† ∈ β„‚ ∢ 𝑇 βˆ’ πœ†πΌ is not invertible+ = πœŽπ‘ 𝑇 = *πœ† ∈ β„‚ ∢ det 𝐴 βˆ’ πœ†πΌ = 0+.
In fact:
𝑇 invertible ⇔ 𝑇 injective ⇔ 𝑇 surjective
Thus,
πœŽπ‘ 𝑇 = *πœ† ∈ β„‚ ∢ det 𝐴 βˆ’ πœ†πΌ = 0+ = *πœ† ∈ β„‚ ∢ 𝑇 βˆ’ πœ†πΌ is not injective+.
Definition: Let 𝐻 be a complex Hilbert space. The pointwise spectrum of 𝑇 ∈ 𝐿(𝐻)
is the set given by
πœŽπ‘ 𝑇 = *πœ† ∈ β„‚ ∢ 𝑇 βˆ’ πœ†πΌ is not injective+.
The elements of πœŽπ‘ 𝑇 are called eigenvalues of 𝑇.
If π‘₯ ∈ ker 𝑇 βˆ’ πœ†πΌ then π‘₯ is an eigenvector of 𝑇 associated to the eigenvalue πœ†.
Fact: Let 𝐻 be a finite-dimensional complex Hilbert space. Then, 𝜎 𝑇 = πœŽπ‘ 𝑇 , for
every 𝑇 ∈ 𝐿 𝐻 (i. e. the spectrum and the pointwise spectrum coincide).
Operators in Hilbert spaces: The spectrum
Fact: If 𝐻 is an infinite-dimensional complex Hilbert space, then πœŽπ‘ 𝑇 βŠ† 𝜎 𝑇
and these sets do not need to coincide.
Example: If πœ‹: 𝐻 β†’ 𝑀 is the projection of 𝐻 over a closed subspace 𝑀, then
πœŽπ‘ 𝑇 = 0,1 . Moreover 𝑀 is the invariant subspace associated to the eigenvalue
0 and 𝑀βŠ₯ the one associated to the eigenvalue 1.
Example: The Volterra operator 𝑇: 𝐿2 ,0,1- β†’ 𝐿2 0,1 is such that πœŽπ‘ 𝑇 = βˆ….
Fact: Let 𝑇 ∈ 𝐿 𝐻 . By the Banach isomorphism theorem we have that there exists
𝑇 βˆ’1 ∈ 𝐿(𝐻) such that 𝑇𝑇 βˆ’1 = 𝑇 βˆ’1 𝑇 = 𝐼 if, and only if, 𝑇 is bijective.
Definition: Let 𝐻 be a complex Hilbert space. The surjective spectrum of 𝑇 ∈ 𝐿 H
is defined by
πœŽπ‘ π‘’ 𝑇 = *πœ† ∈ β„‚ ∢ 𝑇 βˆ’ πœ†πΌ is not surjective+.
Proposition: Let 𝐻 be a complex Hilbert space. For every 𝑇 ∈ 𝐿 H we have
𝜎 𝑇 = πœŽπ‘ 𝑇 βˆͺ πœŽπ‘ π‘’ 𝑇 .
Operators in Hilbert spaces: The spectrum
Theorem: Let 𝐻 be a complex Hilbert space, and let 𝑇 ∈ 𝐿 H . Then, 𝜎 𝑇 is a
non-empty compact subset of β„‚.
Remark: If 𝐻 is a real Hilbert space, then the set *πœ† ∈ ℝ ∢ 𝑇 βˆ’ πœ†πΌ is not bijective+ may
be empty (in whose case, no information is provided).
Example: 𝑇 ∈ 𝐿 ℂℝ
given by 𝑇 π‘₯ = 𝑖π‘₯.
In fact, (𝑇 βˆ’ πœ†πΌ)π‘₯ = 𝑖 βˆ’ πœ† π‘₯ always is bijective because if πœ† ∈ ℝ, then 𝑖 βˆ’ πœ† β‰  0.
Definition: Let 𝐻 be a real Hilbert space. The complexification of 𝐻 is defined as the
complex Hilbert space 𝐻ℂ ≔ 𝐻⨁𝑖𝐻. Moreover, the spectrum of 𝑇 ∈ 𝐿(𝐻) is defined
as the spectrum of the operator 𝑇ℂ ∈ 𝐿(𝐻ℂ) given by 𝑇ℂ 𝑒 + 𝑖𝑣 = 𝑇 𝑒 + 𝑖𝑇 𝑣 .
Theorem: Let 𝐻 be a Hilbert space (either real or complex). If 𝑇 ∈ 𝐿(𝐻) then 𝜎 𝑇 is
a non-empty compact subset of β„‚.
Agreement: From now on, only complex Hilbert spaces will be considered.
Operators in Hilbert spaces: The spectrum
Proposition : If 𝑇 ∈ 𝐿 H then 𝜎 𝑇 βˆ— = *Ξ» ∢ πœ† ∈ 𝜎 𝑇 + (the conjugate set of 𝜎 𝑇 ).
Proof: 𝑇 βˆ’ πœ†πΌ invertible ⇔ 𝑇 βˆ’ πœ†πΌ 𝑅 = 𝑅 𝑇 βˆ’ πœ†πΌ = 𝐼 ⇔ π‘…βˆ— 𝑇 βˆ’ πœ†πΌ
⇔ 𝑅 βˆ— 𝑇 βˆ— βˆ’ πœ†πΌ = π‘…βˆ— 𝑇 βˆ— βˆ’ πœ†πΌ = 𝐼 ⇔ 𝑇 βˆ— βˆ’ πœ†πΌ invertible.
βˆ—
= 𝑇 βˆ’ πœ†πΌ βˆ— 𝑅 βˆ— = 𝐼 βˆ—
Corollary: If 𝑇 ∈ 𝐿 H is self-adjoint then 𝜎 𝑇 βŠ† ℝ.
Proposition: If 𝑇 ∈ 𝐿 H is normal then,
πœ† ∈ 𝜎(𝑇) ⟺ there exists π‘₯𝑛 with π‘₯𝑛 = 1 such that
Note: This is equivalent to the following fact:
πœ† ∈ β„‚\𝜎(𝑇) ⟺ there exists 𝑐 > 0 such that
𝑇 βˆ’ πœ†πΌ π‘₯
(𝑇 βˆ’ πœ†πΌ)π‘₯𝑛 β†’ 0.
β‰₯𝑐 π‘₯ ,
Corollary: If 𝑇 ∈ 𝐿 H is normal, then 𝑇 = max πœ† : πœ† ∈ 𝜎 𝑇 .
Proof: This is a consequence of the fact that if 𝑇 ∈ 𝐿 H is normal, then
𝑇 = 𝑠𝑒𝑝 π‘₯ =1 𝑇π‘₯, π‘₯ (the numerical radius).
π‘₯∈𝐻 .
Compact operators on Hilbert spaces
Let (Ξ©, Ξ£, πœ‡) be a measure space and let π‘˜(𝑠, 𝑑) ∈ 𝐿2(Ξ© × Ξ©, Ξ£ × Ξ£, πœ‡ × πœ‡). Then, the
integral operator with kernel π‘˜(𝑠, 𝑑) is the operator 𝑇: 𝐿2 (πœ‡) β†’ 𝐿2 (πœ‡) given by
𝑇π‘₯ 𝑠 = ∫Ω π‘˜ 𝑠, 𝑑 π‘₯ 𝑑 π‘‘πœ‡(𝑑)
(𝑠 ∈ Ξ©, π‘₯ ∈ 𝐿2 (πœ‡))
In Quantum Mechanics the integral equation 𝑇 βˆ’ πœ†πΌ π‘₯ 𝑠 = 𝑦 𝑠 , where 𝑇 is the
integral operator with kernel π‘˜, is essential.
Definition: Let 𝑋 and π‘Œ be normed spaces. An operator 𝑇 ∈ 𝐿 X, Y is compact if,
for every bounded sequence π‘₯𝑛 in 𝑋, the sequence 𝑇π‘₯𝑛 has a convergent
subsequence in π‘Œ.
Example: The integral operator 𝑇: 𝐿2(πœ‡) β†’ 𝐿2 (πœ‡) with kernel π‘˜ is compact.
Proposition: Let 𝑋 and π‘Œ be normed spaces and let 𝑇 ∈ 𝐿 X, Y . Then, the
following assertions are equivalent:
i)
𝑇 is compact,
ii) For every sequence π‘₯𝑛 in 𝑋 with π‘₯𝑛 = 1, the sequence 𝑇π‘₯𝑛 has a convergent
subsequence in π‘Œ,
iii) 𝑇(𝐡𝑋 ) is compact in π‘Œ (where 𝐡𝑋 denotes the closed unit ball of 𝑋).
Compact operators on Hilbert spaces
Example: If 𝑇 ∈ 𝐿 X, Y is a finite rank opertor, then 𝑇 is compact.
If π‘₯𝑛 = 1, then 𝑇π‘₯𝑛 is a bounded sequence in a finite dimensional normed space
π‘Œ, and hence it has a convergent subsequence.
Example: If 𝐻 is an infinite dimensional Hilbert space, then the identity operator
𝐼: 𝐻 β†’ 𝐻 is not compact.
In fact, if 𝑒𝑛 is an orthonormal sequence, then 𝐼𝑒𝑛 does not have a convergent
subsequence.
Notation: Let 𝑋 and π‘Œ be normed spaces. Let denote
𝐾(𝑋, π‘Œ)= *𝑇 ∈ 𝐿 𝑋, π‘Œ : 𝑇 is compact+.
Proposition: 𝐾 X, Y is closed whenever π‘Œ is a Banach space.
Proposition: Let 𝑋 and π‘Œ be normed spaces and let 𝑇 ∈ 𝐾 X, Y . Then,
i)
𝑇(𝑋) is a separable subspace of π‘Œ.
ii) If π‘Œ is a Hilbert space, and if 𝐡 = *𝑒𝑛 : 𝑛 ∈ β„•+ is a Hilbert basis, then 𝑇 = π‘™π‘–π‘šπœ‹π‘› 𝑇
where πœ‹π‘› is the orthogonal projection on 𝐿𝑖𝑛 𝑒1 , … , 𝑒𝑛 .
Compact operators on Hilbert spaces
Notation: Let 𝐻 and 𝐾 be Hilbert spaces. Let denote
𝐹 𝐻, 𝐾 = *𝑇 ∈ 𝐿 𝐻, 𝐾 : 𝑇 has finite rank+.
Corollary (Jordan): Let 𝐻 and 𝐾 be Hilbert spaces. Then, 𝐾 𝐻, 𝐾 = 𝐹(𝐻, 𝐾).
Note that the integral operator 𝑇 with kernel π‘˜ is the limit of the sequence of finiterank operators 𝑇𝑛 , where 𝑇𝑛 is the integral operators with kernel π‘˜π‘› for a sequence
π‘˜π‘› of simple functions with π‘˜π‘› β†’ π‘˜.
Corollary (Theorem of the adjoint): Let 𝐻 and 𝐾 Hilbert spaces. Then,
𝑇 ∈ 𝐾 𝐻, 𝐾 ⟺ 𝑇 βˆ— ∈ 𝐾 𝐻, 𝐾 .
Theorem: Let 𝑇 ∈ 𝐿(𝐻) be a compact operator and let πœ† β‰  0. Then,
i)
dim ker 𝑇 βˆ’ πœ† 𝐼 < ∞ (Theorem of the kernel) (Riesz)
ii) (𝑇 βˆ’ πœ†πΌ)(𝐻) is closed (Theorem of the rank).
Corollary: Let 𝑇 ∈ 𝐿(𝐻) a compact operators and let πœ† β‰  0. Then,
(𝑇 βˆ’ πœ†πΌ)(𝐻) = 𝐻 ⟺ ker 𝑇 βˆ’ πœ† 𝐼 = *0+.
Consequently,
πœŽπ‘ π‘’ (𝑇)\*0+ = πœŽπ‘ (𝑇)\*0+ = 𝜎(𝑇)\*0+.
In finite dimension πœŽπ‘ π‘’ (𝑇) = πœŽπ‘ (𝑇) = 𝜎(𝑇).
Compact operators on Hilbert spaces
Theorem (Fredholm alternative): Let 𝑇 ∈ 𝐿(𝐻) be a compact operator and let
πœ† β‰  0. Consider the following equations:
(a) 𝑇 βˆ’ πœ† 𝐼 π‘₯ = 𝑦
(b) 𝑇 βˆ— βˆ’ πœ†πΌ π‘₯ = 𝑦
(c) 𝑇 βˆ’ πœ† 𝐼 π‘₯ = 0
(d) 𝑇 βˆ— βˆ’ πœ†πΌ π‘₯ = 0
Then either
i) the equations (a) and (b) has a solution π‘₯ and π‘₯, for every 𝑦, 𝑦 ∈ 𝐻, resp.
ii) or the homogeneous system of equations (c) and (d) have a non-trivial
solution.
In the case (i), the solutions π‘₯ and π‘₯ are unique and depend continously on 𝑦
and 𝑦 respectively.
In the case (ii), the equation (a) has a unique solution π‘₯ if, and only if, 𝑦 is
orthogonal to all the solutions of (d). Similarly (b) has a unique solution π‘₯ if,
and only if, 𝑦 is orthogonal to all the solutions of (a).
Proof: If πœ† βˆ‰ 𝜎(𝑇) then we have (i) obviously, because 𝜎(𝑇) = 𝜎(𝑇 βˆ— ). In fact,
π‘₯ = (𝑇 βˆ’ πœ†πΌ)βˆ’1 𝑦 meanwhile π‘₯ = (𝑇 βˆ’ πœ†πΌ)βˆ’1 𝑦.
Otherwise, πœ† ∈ 𝜎(𝑇) βˆ– *0+ = πœŽπ‘ (𝑇) βˆ– *0+, and hence (c) and (d) have a non-trivial
solution. Moreover, (a) has a solution ⟺ 𝑦 ∈ 𝑇 βˆ’ πœ† 𝐼 𝐻 = ker 𝑇 βˆ— βˆ’ πœ†πΌ
Similarly (b) has a solution ⟺ 𝑦 ∈ 𝑇 βˆ— βˆ’ πœ†πΌ 𝐻 = ker 𝑇 βˆ’ πœ† 𝐼 βŠ₯ .
βŠ₯
.
Compact operators on Hilbert spaces
Corollary: If 𝑇 ∈ 𝐿(𝐻) is a self-adjoint compact operator, then 𝑇 or βˆ’ 𝑇
eigenvalue of 𝑇.
Indeed it is known that if 𝑇 is normal, then
is an
𝑇 = max πœ† : πœ† ∈ 𝜎 𝑇 .
Remarks:
i) Recall that if 𝑇 is compact, then πœŽπ‘ (𝑇)\*0+ = 𝜎(𝑇)\*0+.
ii) The restriction of a self-adjoint compact operator to an invariant subspace is a
self-adjoint compact operator.
Diagonalization of a selft-adjoint compact operator
Let 𝑇 ∈ 𝐿 𝐻 a self-adjoint compact operator (𝑇 β‰  0).
Let πœ†1 ∈ πœŽπ‘ (𝑇) such that πœ†1 = 𝑇 . Let 𝑒1 with 𝑒1 = 1 such that 𝑇 βˆ’ πœ†1 𝐼 𝑒1 = 0.
Let 𝐻1 = 𝐻 and 𝐻2 = 𝑒 ∈ 𝐻1 : 𝑒 βŠ₯ 𝑒1 = 𝑒1 βŠ₯ . If 𝑒 ∈ 𝐻2 , then 𝑒, 𝑒1 = 0 and hence,
0 = πœ†1 𝑒, 𝑒1 = 𝑒, πœ†1 𝑒1 = 𝑒, 𝑇𝑒1 = 𝑇 βˆ— 𝑒, 𝑒1 = πœ†1 𝑒, 𝑒1 = 𝑇𝑒, 𝑒1 ,
Thus, 𝑇(𝐻2)= 𝐻2, so that π‘‡βˆ•π»2 is self-adjoint and compact.
Compact operators on Hilbert spaces
Repeat the process with π‘‡βˆ•π»2 .
Let πœ†2 ∈ πœŽπ‘ (π‘‡βˆ•π»2 ) βŠ† πœŽπ‘ (𝑇) such that πœ†2 = π‘‡βˆ•π»2 .
Note that πœ†2 ≀ πœ†1 because π‘‡βˆ•π»2 ≀ 𝑇 .
Let 𝐻3 = 𝑒 ∈ 𝐻2 : 𝑒 βŠ₯ 𝑒2 . Then π‘‡βˆ•π»2 is self-adjoint and compact and π‘‡βˆ•π»2 (𝐻3 ) = 𝐻3 .
Note that
𝐻2 = 𝑒 ∈ 𝐻1 : 𝑒 βŠ₯ 𝑒1 = 𝑒1 βŠ₯
𝐻3 = 𝑒 ∈ 𝐻2 : 𝑒 βŠ₯ 𝑒2 = 𝑒1 βŠ₯ ∩ 𝑒2 βŠ₯ = 𝑒1 , 𝑒2 βŠ₯
Reiterating the process we obtain nonzero eigenvalues πœ†1 , πœ†2 ,…, πœ†π‘› such that
πœ†π‘› ≀ β‹― ≀ πœ†2 ≀ πœ†1 ,
with unital eigenvectors 𝑒1 , 𝑒2 ,…, 𝑒𝑛 , and closed invariant subspaces 𝐻1 , 𝐻2,…, 𝐻𝑛 ,
where 𝐻𝑗+1 = 𝑒 ∈ 𝐻𝑗 : 𝑒 βŠ₯ 𝑒𝑗 is such that 𝐻𝑛 βŠ† β‹― βŠ† 𝐻2 βŠ† 𝐻1 and
πœ†π‘— = π‘‡βˆ•π»π‘— .
The process stops only when π‘‡βˆ•π»π‘›+1 = 0.
Since 𝐻𝑛+1 = 𝑒1 , 𝑒2 ,…, 𝑒𝑛 βŠ₯ and 𝐻 = 𝑒1 , 𝑒2 ,…, 𝑒𝑛 ⨁ 𝑒1 , 𝑒2 ,…, 𝑒𝑛 βŠ₯ (OPT) we
have: if 𝑒 ∈ 𝐻, then 𝑒 = 𝑛𝑖=1 𝑒, 𝑒𝑖 𝑒𝑖 + 𝑣 with 𝑣 ∈ 𝐻𝑛+1 . Thus, if π‘‡βˆ•π»π‘›+1 = 0 then
𝑇𝑒 =
(this is the case, particularly, if π‘‘π‘–π‘šπ»<∞).
𝑛
𝑖=1 πœ†π‘–
𝑒, 𝑒𝑖 𝑒𝑖
Compact operators on Hilbert spaces
If the process does not stop, then we obtain a sequence of eigenvalues πœ†π‘› β†’ 0.
1
1
Indeed, if πœ†π‘› β†’ πœ† > 0 then
𝑒𝑛 is bounded so that 𝑇( 𝑒𝑛 )= 𝑒𝑛 has a
πœ†π‘›
πœ†π‘›
covergent subsequence which contradicts that
𝑒𝑛 βˆ’ π‘’π‘š = 2.
Hence, in this case, for every 𝑛 ∈ β„• we have that
𝐻 = 𝑒1 , 𝑒2 ,…, 𝑒𝑛 ⨁ 𝑒1 , 𝑒2 ,…, 𝑒𝑛 βŠ₯.
𝑛
𝑖=1
𝑒, 𝑒𝑖 𝑒𝑖 + 𝑣𝑛 with 𝑣𝑛 ∈ 𝑒1 , 𝑒2 ,…, 𝑒𝑛 βŠ₯ = 𝐻𝑛+1 , then we obtain that
𝑇𝑣𝑛 = π‘‡βˆ•π»π‘›+1 𝑣𝑛 ≀ π‘‡βˆ•π»π‘›+1 𝑣𝑛 = πœ†π‘›+1 𝑣𝑛 ≀ πœ†π‘›+1 𝑒 β†’ 0,
And hence,
If 𝑒 =
∞
𝑇𝑒 =
πœ†π‘– 𝑒, 𝑒𝑖 𝑒𝑖
𝑖=1
Note that if πœ† is a nonzero eigenvalue of 𝑇, then πœ† ∈ πœ†π‘› : 𝑛 ∈ β„• . Otherwise, if 𝑒
is an associated eigenvalue then, πœ†π‘’ = 𝑇𝑒 = 0, which is impossible.
Compact operators on Hilbert spaces: spectral theorem
Theorem (spectral theorem for compact self-adjoint operators):
Let 𝑇 ∈ 𝐿(𝐻) be a compact self-adjoint operator. Then 𝑇 is diagonalizable.
Indeed, one of the following assertions holds:
i) There exist eignevalues πœ†1 , πœ†2 ,…, πœ†π‘› and a system of associated
orthonormal eigenvectors 𝑒1 , 𝑒2 ,…, 𝑒𝑛 such that, for every 𝑒 ∈ 𝐻,
𝑛
𝑇𝑒 =
πœ†π‘– 𝑒, 𝑒𝑖 𝑒𝑖
𝑖=1
(uniform convergence over the compact subsets of 𝐻).
ii) There exists a sequence πœ†π‘› of eigenvalues such that πœ†π‘› β†’ 0, and a
sequence of associated orthonormal eigenvectors 𝑒𝑛 such that, βˆ€π‘’ ∈ 𝐻,
∞
𝑇𝑒 =
πœ†π‘– 𝑒, 𝑒𝑖 𝑒𝑖
𝑖=1
(uniform convergence over the compact subsets of 𝐻).
In (i) as well as in (ii), if πœ† is a non-zero eigenvalue, then πœ† ∈ πœ†1 , πœ†2 … .
Moreover, the dimension of the invariant subspace associated to πœ†
coincides with the number of times that πœ† appears in πœ†1 , πœ†2 … .
Compact operators on Hilbert spaces: spectral theorem
Corollary: 𝑇 ∈ 𝐿 𝐻 is a compact self-adjoint operator ⟺ 𝑇 is diagonalizable, i.e.
𝑇=
πœ†π‘– βˆ™, 𝑒𝑖 𝑒𝑖
𝑖
for a countable family of
𝑒1 , 𝑒2 … .
real numbers
πœ†1 , πœ†2 …
and a orthonormal system
Rearranging the above sum, fix π‘˜ and denote π‘ƒπœ†π‘˜ = πœ†π‘–=πœ†π‘˜ βˆ™, 𝑒𝑖 𝑒𝑖 . Since the linear
subspace generated by 𝑒𝑖 , with πœ†π‘– = πœ†π‘˜ , is precisely ker 𝑇 βˆ’ πœ†π‘˜ 𝐼 , we obtain that
π‘ƒπ‘˜ is nothing but the projection of 𝐻 over ker 𝑇 βˆ’ πœ†π‘˜ 𝐼 .
Theorem (spectral resolution of a compact self-adjoint operator):
Let 𝑇 ∈ 𝐿(𝐻) be a compact self-adjoint operator. For every eignevalue πœ† let π‘ƒπœ† be
the orthogonal projection on ker 𝑇 βˆ’ πœ†πΌ . Then the family πœ†π‘ƒπœ† πœ†βˆˆπœŽπ‘ (𝑇) is summable
in the Banach space 𝐿 𝐻 , and
Remark: Now each Ξ»
𝑇=
πœ†π‘ƒπœ† .
appears only once.
πœ†βˆˆπœŽπ‘(𝑇)
Moreover, for every πœ† ∈ πœŽπ‘ (𝑇)\*0+, the corresponding projection π‘ƒπœ† has finite rank
and these projections are mutually orthogonal, i.e. if Ξ» and πœ‡ are nonβˆ’equal
eigenvalues, then π‘ƒπœ† π‘ƒπœ‡ = π‘ƒπœ‡ π‘ƒπœ† = 0.
Compact operators on Hilbert spaces: spectral theorem
Recall that 𝑇 ∈ 𝐿(𝐻) is normal ⇔ 𝑇 = 𝑅 + 𝑖𝑆 where 𝑅 and 𝑆 are self-adjoint
operators such that 𝑅𝑆 = 𝑆𝑅.
Theorem (spectral resolution of a compact normal operator):
Let 𝑇 ∈ 𝐿 𝐻 be a compact normal operator. For every eigenvalue πœ†, let π‘ƒπœ† be
the orthogonal projection on the invariant subspace ker 𝑇 βˆ’ πœ†πΌ . Then, the family
πœ†π‘ƒπœ† πœ†βˆˆπœŽπ‘ (𝑇) is summable in the Banach space 𝐿 𝐻 , and
𝑇 = πœ†βˆˆπœŽπ‘ (𝑇) πœ†π‘ƒπœ† .
Moreover, for every πœ† ∈ πœŽπ‘ (𝑇)\*0+, the corresponding projection π‘ƒπœ† has finite
rank, and for every non-equal eigenvalues Ξ», and πœ‡, we have that
π‘ƒπœ† π‘ƒπœ‡ =π‘ƒπœ‡ π‘ƒπœ† =0.
Corollary: If 𝑇 ∈ 𝐿(𝐻) is a compact normal operator, then 𝑇 is diagonalizable. In
fact,
𝑇=
πœ†π‘– βˆ™, 𝑒𝑖 𝑒𝑖
𝑖
where πœ†1 , πœ†2 … is the set of non-zero eigenvalues of 𝑇 and
orthonormal system of associated eigenvectors.
𝑒1 , 𝑒2 … is an
Compact operators on Hilbert spaces: spectral theorem
Corollary: Let 𝑇 ∈ 𝐿(𝐻) be a compact normal operator. If 𝑇 = 𝑖 πœ†π‘– βˆ™, 𝑒𝑖 𝑒𝑖
then, for every 𝑦 ∈ 𝐻 and πœ† β‰  0 we have that:
i) If πœ† βˆ‰ πœ†1 , πœ†2 … , then the equation 𝑇 βˆ’ πœ†πΌ π‘₯ = 𝑦 has a unique solution for
every 𝑦 ∈ 𝐻. This solution is given by
1
πœ†
π‘₯= (
𝑦,π‘’π‘˜
πœ†β‰ πœ†π‘˜ πœ†π‘˜ πœ† βˆ’πœ†
π‘˜
π‘’π‘˜ βˆ’ 𝑦).
ii) Otherwise, the equation 𝑇 βˆ’ πœ†πΌ π‘₯ = 𝑦 has a solution ⟺ 𝑦 βŠ₯ ker 𝑇 βˆ’ πœ†πΌ .
In this is the case, the general solution is given by
1
πœ†
π‘₯= (
𝑦,π‘’π‘˜
πœ†β‰ πœ†π‘˜ πœ†π‘˜ πœ† βˆ’πœ†
π‘˜
π‘’π‘˜ βˆ’ 𝑦) + 𝑧
(𝑧 ∈ ker 𝑇 βˆ’ πœ†πΌ ).
Proof: If 𝑇 βˆ’ πœ†πΌ π‘₯ = 𝑦, then 𝑇π‘₯ = πœ†π‘₯ + 𝑦, so that
𝑇π‘₯, π‘’π‘˜ = πœ†π‘₯ + 𝑦, π‘’π‘˜ = πœ† π‘₯, π‘’π‘˜ + 𝑦, π‘’π‘˜
Since 𝑇π‘₯, π‘’π‘˜ = 𝑖 πœ†π‘– π‘₯, 𝑒𝑖 𝑒𝑖 , π‘’π‘˜ = πœ†π‘˜ π‘₯, π‘’π‘˜ we obtain that
πœ† π‘₯, π‘’π‘˜ + 𝑦, π‘’π‘˜ = πœ†π‘˜ π‘₯, π‘’π‘˜ ,
and hence (πœ†π‘˜ βˆ’ πœ†) π‘₯, π‘’π‘˜ = 𝑦, π‘’π‘˜ .
1
Therefore if (πœ†π‘˜ βˆ’ πœ†)β‰  0, then π‘₯, π‘’π‘˜ =
𝑦, π‘’π‘˜ . Consequenly
1
πœ†
1
πœ†
πœ†π‘˜ βˆ’πœ†
π‘₯ = (𝑇π‘₯ βˆ’ 𝑦) = ( πœ†π‘– π‘₯, 𝑒𝑖 𝑒𝑖 βˆ’ 𝑦) =
1
πœ†
(
πœ†β‰ πœ†π‘– πœ†π‘–
𝑦,𝑒𝑖
πœ†π‘– βˆ’πœ†
𝑒𝑖 βˆ’ 𝑦).