Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Evan Camrud
Iowa State University
June 2, 2018 1 R: 1 ⇥ ⇤1 0 0 2 R3: 0 , 1 , 0 2 3 2 3 2 3 0 0 1 4 15 4 05 4 5 0 0 0 1 0 0 2 3 2 3 2 3 2 3 n 0 0 . . 3 R : , ,..., . , . 6.7 6.7 6 7 6 7 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 Question: How could we construct an ONB for an infinite-dimensional case?
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Let’s look at some orthonormal bases (ONBs): 1 0 0 2 R3: 0 , 1 , 0 2 3 2 3 2 3 0 0 1 4 15 4 05 4 5 0 0 0 1 0 0 2 3 2 3 2 3 2 3 n 0 0 . . 3 R : , ,..., . , . 6.7 6.7 6 7 6 7 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 Question: How could we construct an ONB for an infinite-dimensional case?
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Let’s look at some orthonormal bases (ONBs): 1 R: 1 ⇥ ⇤ 1 0 0 0 0 1 0 0 2 3 2 3 2 3 2 3 n 0 0 . . 3 R : , ,..., . , . 6.7 6.7 6 7 6 7 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 Question: How could we construct an ONB for an infinite-dimensional case?
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Let’s look at some orthonormal bases (ONBs): 1 R: 1 ⇥ ⇤1 0 0 2 R3: 0 , 1 , 0 2 3 2 3 2 3 0 0 1 4 5 4 5 4 5 Question: How could we construct an ONB for an infinite-dimensional case?
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Let’s look at some orthonormal bases (ONBs): 1 R: 1 ⇥ ⇤1 0 0 2 R3: 0 , 1 , 0 2 3 2 3 2 3 0 0 1 4 15 4 05 4 5 0 0 0 1 0 0 2 3 2 3 2 3 2 3 n 0 0 . . 3 R : , ,..., . , . 6.7 6.7 6 7 6 7 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Let’s look at some orthonormal bases (ONBs): 1 R: 1 ⇥ ⇤1 0 0 2 R3: 0 , 1 , 0 2 3 2 3 2 3 0 0 1 4 15 4 05 4 5 0 0 0 1 0 0 2 3 2 3 2 3 2 3 n 0 0 . . 3 R : , ,..., . , . 6.7 6.7 6 7 6 7 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 Question: How could we construct an ONB for an infinite-dimensional case? For ease of notation, we refer to these as e where the { n}n1=1 subscript corresponds to the only nonzero entry to the vector.
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Answer: Exactly like you would expect:
1 0 0 0 0 1 0 0 2 3 2 3 2 3 2 3 0 0 . . . . 6.7, 6.7,...,6 7, 6 7,... 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Answer: Exactly like you would expect:
1 0 0 0 0 1 0 0 2 3 2 3 2 3 2 3 0 0 . . . . 6.7, 6.7,...,6 7, 6 7,... 6.7 6.7 617 607 6 7 6 7 6 7 6 7 607 607 607 617 6 7 6 7 6 7 6 7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6.7 6 7 6 7 6 7 6 7 4 5 4 5 4 5 4 5 For ease of notation, we refer to these as e where the { n}n1=1 subscript corresponds to the only nonzero entry to the vector. and we can define an inner product as a direct extension of the dot product of vectors, such that
1 u, v = µ (2) h i n n n=1 X
for u = n1=1 µnen and v = n1=1 nen. P P
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Since this is an orthonormal basis, we may construct elements of this infinite-dimensional Hilbert space by means of (infinite) linear combinations:
1 v = nen for j C (1) 2 n=1 X Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Since this is an orthonormal basis, we may construct elements of this infinite-dimensional Hilbert space by means of (infinite) linear combinations:
1 v = nen for j C (1) 2 n=1 X and we can define an inner product as a direct extension of the dot product of vectors, such that
1 u, v = µ (2) h i n n n=1 X
for u = n1=1 µnen and v = n1=1 nen. P P Answer: It must be that
1 1 µ 2 < and 2 < (3) | n| 1 | n| 1 n=1 n=1 X X Note that these conditions are both necessary and su cient for finite inner products as well as norms, a fact arising from the Cauchy-Schwartz inequality. This very special infinite-dimensional Hilbert space is known as `2(N), the space of square-summable sequences. (Often this may just be written as `2.)
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: What do we require about µ , in { n}n1=1 { n}n1=1 order to have well-defined inner products? Note that these conditions are both necessary and su cient for finite inner products as well as norms, a fact arising from the Cauchy-Schwartz inequality. This very special infinite-dimensional Hilbert space is known as `2(N), the space of square-summable sequences. (Often this may just be written as `2.)
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: What do we require about µ , in { n}n1=1 { n}n1=1 order to have well-defined inner products?
Answer: It must be that
1 1 µ 2 < and 2 < (3) | n| 1 | n| 1 n=1 n=1 X X This very special infinite-dimensional Hilbert space is known as `2(N), the space of square-summable sequences. (Often this may just be written as `2.)
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: What do we require about µ , in { n}n1=1 { n}n1=1 order to have well-defined inner products?
Answer: It must be that
1 1 µ 2 < and 2 < (3) | n| 1 | n| 1 n=1 n=1 X X Note that these conditions are both necessary and su cient for finite inner products as well as norms, a fact arising from the Cauchy-Schwartz inequality. Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: What do we require about µ , in { n}n1=1 { n}n1=1 order to have well-defined inner products?
Answer: It must be that
1 1 µ 2 < and 2 < (3) | n| 1 | n| 1 n=1 n=1 X X Note that these conditions are both necessary and su cient for finite inner products as well as norms, a fact arising from the Cauchy-Schwartz inequality. This very special infinite-dimensional Hilbert space is known as `2(N), the space of square-summable sequences. (Often this may just be written as `2.) Question: Is R a separable vector space?
Answer: Yes. Consider Q.
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Definition A normed linear space is called separable if it contains a countable subset of vectors whose span is dense in the vector space. (i.e. There exists v such that for v V , ✏>0, { n}n1=1 2 v N v <✏.) n=1 n n P Answer: Yes. Consider Q.
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Definition A normed linear space is called separable if it contains a countable subset of vectors whose span is dense in the vector space. (i.e. There exists v such that for v V , ✏>0, { n}n1=1 2 v N v <✏.) n=1 n n Question:P Is R a separable vector space? Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Definition A normed linear space is called separable if it contains a countable subset of vectors whose span is dense in the vector space. (i.e. There exists v such that for v V , ✏>0, { n}n1=1 2 v N v <✏.) n=1 n n Question:P Is R a separable vector space?
Answer: Yes. Consider Q. Proof. (= ) Existence of such a basis is guaranteed by Zorn’s Lemma. ) (A maximal orthonormal set is known to be a basis.) ( =) The ONB is the countable dense subset. (
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Theorem A Hilbert space is separable if and only if it has a countable orthonormal basis. Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Theorem A Hilbert space is separable if and only if it has a countable orthonormal basis.
Proof. (= ) Existence of such a basis is guaranteed by Zorn’s Lemma. ) (A maximal orthonormal set is known to be a basis.) ( =) The ONB is the countable dense subset. ( Proof. Consider a separable Hilbert space , and let " be an H { n}n1=1 ONB of . H Let ' : `2 be defined to be linear, and also H!
'("n)=en (4)
where e is the ONB of `2. It is trivial to check that ' is { n}n1=1 an isomorphism.
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Introduction
Theorem All separable infinite-dimensional Hilbert spaces are isomorphic to `2. Reproducing Kernel Hilbert Spaces and Hardy Spaces
Introduction
Theorem All separable infinite-dimensional Hilbert spaces are isomorphic to `2.
Proof. Consider a separable Hilbert space , and let " be an H { n}n1=1 ONB of . H Let ' : `2 be defined to be linear, and also H!
'("n)=en (4)
where e is the ONB of `2. It is trivial to check that ' is { n}n1=1 an isomorphism. Answer: Yes. Consider L2(R,n), where n is the counting measure.
Remark Non-separable infinite-dimensional Hilbert spaces are hardly ever considered, so in the future, when a Hilbert space is introduced, it is seen as either finite-dimensional (isomorphic to Cn) or separable infinite-dimensional (isomorphic to `2).
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: Do non-separable infinite-dimensional Hilbert spaces exist? Remark Non-separable infinite-dimensional Hilbert spaces are hardly ever considered, so in the future, when a Hilbert space is introduced, it is seen as either finite-dimensional (isomorphic to Cn) or separable infinite-dimensional (isomorphic to `2).
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: Do non-separable infinite-dimensional Hilbert spaces exist?
Answer: Yes. Consider L2(R,n), where n is the counting measure. Reproducing Kernel Hilbert Spaces and Hardy Spaces
Infinite Dimensional Hilbert Spaces
Question: Do non-separable infinite-dimensional Hilbert spaces exist?
Answer: Yes. Consider L2(R,n), where n is the counting measure.
Remark Non-separable infinite-dimensional Hilbert spaces are hardly ever considered, so in the future, when a Hilbert space is introduced, it is seen as either finite-dimensional (isomorphic to Cn) or separable infinite-dimensional (isomorphic to `2). Question: What does f “look like”?
Answer: f(x)= x for some R. 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
Consider f : R R such that for all x1,x2 R, a, b C ! 2 2
f(ax1 + bx2)=af(x1)+bf(x2). (5) That is, suppose f is a linear functional. Answer: f(x)= x for some R. 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
Consider f : R R such that for all x1,x2 R, a, b C ! 2 2
f(ax1 + bx2)=af(x1)+bf(x2). (5) That is, suppose f is a linear functional.
Question: What does f “look like”? Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
Consider f : R R such that for all x1,x2 R, a, b C ! 2 2
f(ax1 + bx2)=af(x1)+bf(x2). (5) That is, suppose f is a linear functional.
Question: What does f “look like”?
Answer: f(x)= x for some R. 2 Question: What does f “look like”?
x1 Answer: For ~x = x , 2 23 x3 ~ ~ ~ 3 f(~x )= 1x1 + 2x24+ 53x3 = ~x = ,~x for some R . · h i 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
3 3 Consider f : R R such that for all ~x 1,~x2 R , a, b C ! 2 2
f(a~x1 + b~x2)=af(~x 1)+bf(~x 2). (6)
That is, once again, suppose f is a linear functional. x1 Answer: For ~x = x , 2 23 x3 ~ ~ ~ 3 f(~x )= 1x1 + 2x24+ 53x3 = ~x = ,~x for some R . · h i 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
3 3 Consider f : R R such that for all ~x 1,~x2 R , a, b C ! 2 2
f(a~x1 + b~x2)=af(~x 1)+bf(~x 2). (6)
That is, once again, suppose f is a linear functional.
Question: What does f “look like”? Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
3 3 Consider f : R R such that for all ~x 1,~x2 R , a, b C ! 2 2
f(a~x1 + b~x2)=af(~x 1)+bf(~x 2). (6)
That is, once again, suppose f is a linear functional.
Question: What does f “look like”?
x1 Answer: For ~x = x , 2 23 x3 ~ ~ ~ 3 f(~x )= 1x1 + 2x24+ 53x3 = ~x = ,~x for some R . · h i 2 Question: What does f “look like”?
x1 x2 Answer: For ~x = 2 . 3, . 6 7 6x 7 6 n7 ~ ~ ~ n f(~x )= 1x1 + 2x24+ ...5+ nxn = ~x = ,~x for some R . · h i 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
n n Consider f : R R such that for all ~x 1,~x2 R , a, b C ! 2 2
f(a~x1 + b~x2)=af(~x 1)+bf(~x 2). (7)
That is, once again, suppose f is a linear functional. x1 x2 Answer: For ~x = 2 . 3, . 6 7 6x 7 6 n7 ~ ~ ~ n f(~x )= 1x1 + 2x24+ ...5+ nxn = ~x = ,~x for some R . · h i 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
n n Consider f : R R such that for all ~x 1,~x2 R , a, b C ! 2 2
f(a~x1 + b~x2)=af(~x 1)+bf(~x 2). (7)
That is, once again, suppose f is a linear functional.
Question: What does f “look like”? Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
n n Consider f : R R such that for all ~x 1,~x2 R , a, b C ! 2 2
f(a~x1 + b~x2)=af(~x 1)+bf(~x 2). (7)
That is, once again, suppose f is a linear functional.
Question: What does f “look like”?
x1 x2 Answer: For ~x = 2 . 3, . 6 7 6x 7 6 n7 ~ ~ ~ n f(~x )= 1x1 + 2x24+ ...5+ nxn = ~x = ,~x for some R . · h i 2 Question: What does f “look like”?
Answer: f( )=⌫ + ⌫ + ... + ⌫ + ... = ⌫ , { n}n1=1 1 1 2 2 n n h{ n}n1=1 { n}n1=1i for some ⌫ `2. { }n1=1 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
2 2 Consider f : ` R, such that for all n 1 , µn 1 ` , ! { }n=1 { }n=1 2 a, b C 2 f a 1 + b µ 1 = af 1 + bf( µ 1 . (8) { n}n=1 { n}n=1 { n}n=1 { n}n=1 as well as f( 1 ) M 1 (9) { n}n=1 · { n}n=1 for some M R. That is, once again, suppose f is a linear 2 functional, but also that it is bounded . This boundedness is equivalent to continuity of the functional. Answer: f( )=⌫ + ⌫ + ... + ⌫ + ... = ⌫ , { n}n1=1 1 1 2 2 n n h{ n}n1=1 { n}n1=1i for some ⌫ `2. { }n1=1 2
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
2 2 Consider f : ` R, such that for all n 1 , µn 1 ` , ! { }n=1 { }n=1 2 a, b C 2 f a 1 + b µ 1 = af 1 + bf( µ 1 . (8) { n}n=1 { n}n=1 { n}n=1 { n}n=1 as well as f( 1 ) M 1 (9) { n}n=1 · { n}n=1 for some M R. That is, once again, suppose f is a linear 2 functional, but also that it is bounded . This boundedness is equivalent to continuity of the functional.
Question: What does f “look like”? Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
2 2 Consider f : ` R, such that for all n 1 , µn 1 ` , ! { }n=1 { }n=1 2 a, b C 2 f a 1 + b µ 1 = af 1 + bf( µ 1 . (8) { n}n=1 { n}n=1 { n}n=1 { n}n=1 as well as f( 1 ) M 1 (9) { n}n=1 · { n}n=1 for some M R. That is, once again, suppose f is a linear 2 functional, but also that it is bounded . This boundedness is equivalent to continuity of the functional.
Question: What does f “look like”?
Answer: f( )=⌫ + ⌫ + ... + ⌫ + ... = ⌫ , { n}n1=1 1 1 2 2 n n h{ n}n1=1 { n}n1=1i for some ⌫ `2. { }n1=1 2 Theorem (Riesz-Represenation for Hilbert spaces) Let be a Hilbert space. Let f : R be a bounded linear H H! functional. Then there is a v such that for all u 2H 2H f(u)= u, v . (10) h i
Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
Aside from the non-separable infinite dimensional case, we have just “proved” the Riesz-Representation theorem for Hilbert spaces. Reproducing Kernel Hilbert Spaces and Hardy Spaces
Reproducing Kernel Hilbert Space
Aside from the non-separable infinite dimensional case, we have just “proved” the Riesz-Representation theorem for Hilbert spaces.
Theorem (Riesz-Represenation for Hilbert spaces) Let be a Hilbert space. Let f : R be a bounded linear H H! functional. Then there is a v such that for all u 2H 2H f(u)= u, v . (10) h i That is, if x : R (for some fixed x K)isdefinedby H! 2