Computability of the Radon-Nikodym derivative Mathieu Hoyrup, - - PowerPoint PPT Presentation

computability of the radon nikodym derivative
SMART_READER_LITE
LIVE PREVIEW

Computability of the Radon-Nikodym derivative Mathieu Hoyrup, - - PowerPoint PPT Presentation

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Frchet-Riesz representation theorem Computability of the Radon-Nikodym derivative Mathieu Hoyrup, Cristbal Rojas and Klaus Weihrauch LORIA, INRIA Nancy - France


slide-1
SLIDE 1

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Computability of the Radon-Nikodym derivative

Mathieu Hoyrup, Cristóbal Rojas and Klaus Weihrauch

LORIA, INRIA Nancy - France

slide-2
SLIDE 2

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

  • Let λ be the uniform measure over [0, 1].
  • Let f ∈ L1(λ) be nonnegative.
  • Let µ(A) =
  • Af dλ.
  • One has µ ≪ λ, i.e. for all A, λ(A) = 0 =

⇒ µ(A) = 0.

slide-3
SLIDE 3

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

  • Let λ be the uniform measure over [0, 1].
  • Let f ∈ L1(λ) be nonnegative.
  • Let µ(A) =
  • Af dλ.
  • One has µ ≪ λ, i.e. for all A, λ(A) = 0 =

⇒ µ(A) = 0. Conversely,

Theorem (Radon-Nikodym, 1930)

For every measure µ ≪ λ there exists f ∈ L1(λ) such that µ(A) =

  • A

f dλ for all Borel sets A. f is the Radon-Nikodym derivative of µ w.r.t. λ, denoted f = dµ

dλ.

slide-4
SLIDE 4

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

  • Let λ be the uniform measure over [0, 1].
  • Let f ∈ L1(λ) be nonnegative.
  • Let µ(A) =
  • Af dλ.
  • One has µ ≪ λ, i.e. for all A, λ(A) = 0 =

⇒ µ(A) = 0. Conversely,

Theorem (Radon-Nikodym, 1930)

For every measure µ ≪ λ there exists f ∈ L1(λ) such that µ(A) =

  • A

f dλ for all Borel sets A. f is the Radon-Nikodym derivative of µ w.r.t. λ, denoted f = dµ

dλ.

Our problem

Is dµ

dλ computable from µ?

slide-5
SLIDE 5

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Theorem

On [0, 1], there is a computable measure µ ≪ λ (even µ ≤ 2λ) such that

dµ dλ is not L1(λ)-computable.

slide-6
SLIDE 6

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Theorem

On [0, 1], there is a computable measure µ ≪ λ (even µ ≤ 2λ) such that

dµ dλ is not L1(λ)-computable.

Proof.

The measure will be defined as µ(A) = λ(A|K) = λ(A∩K)

λ(K)

where K ⊆ [0, 1]:

  • is a recursive compact set,
  • λ(K) > 0 is not computable (only upper semi-computable, or

right-c.e.). K Iǫ I0 I1

I00 I01 I10 I11

slide-7
SLIDE 7

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Theorem

On [0, 1], there is a computable measure µ ≪ λ (even µ ≤ 2λ) such that

dµ dλ is not L1(λ)-computable.

Proof cont’d.

There is a computable homeomorphim φ : {0, 1}N → K and µ is the push-forward φ∗λ of the uniform on Cantor space, so it is computable.

dµ dλ = 1 λ(K)1K is not L1(λ)-computable.

K Iǫ I0 I1

I00 I01 I10 I11

slide-8
SLIDE 8

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

(Non-)computability of RN

Theorem

There exists a computable measure µ ≪ λ such that dµ

dλ is not

L1(λ)-computable.

Question

How much is the Radon-Nikodym theorem non-computable?

slide-9
SLIDE 9

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

(Non-)computability of RN

Theorem

There exists a computable measure µ ≪ λ such that dµ

dλ is not

L1(λ)-computable.

Question

How much is the Radon-Nikodym theorem non-computable?

Answer

No more than the Fréchet-Riesz representation theorem.

Theorem

RN ≤W FR.

slide-10
SLIDE 10

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

(Non-)computability of RN

Theorem

There exists a computable measure µ ≪ λ such that dµ

dλ is not

L1(λ)-computable.

Question

How much is the Radon-Nikodym theorem non-computable?

Answer

No more than the Fréchet-Riesz representation theorem.

Theorem

RN ≤W FR. And even,

Theorem

RN ≡W FR ≡W EC.

slide-11
SLIDE 11

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Weihrauch degrees

  • ≤W is due to Weihrauch (1992).
slide-12
SLIDE 12

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Weihrauch degrees

  • ≤W is due to Weihrauch (1992).
  • According to Klaus Weihrauch, W in ≤W stands for Wadge.
slide-13
SLIDE 13

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Weihrauch degrees

  • ≤W is due to Weihrauch (1992).
  • According to Klaus Weihrauch, W in ≤W stands for Wadge.
  • Nevertheless, ≤W is now called Weihrauch-reducibility.
slide-14
SLIDE 14

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Weihrauch degrees

  • ≤W is due to Weihrauch (1992).
  • According to Klaus Weihrauch, W in ≤W stands for Wadge.
  • Nevertheless, ≤W is now called Weihrauch-reducibility.
  • f ≤W g if given x, one can compute f (x) applying g once.
slide-15
SLIDE 15

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Weihrauch degrees

  • ≤W is due to Weihrauch (1992).
  • According to Klaus Weihrauch, W in ≤W stands for Wadge.
  • Nevertheless, ≤W is now called Weihrauch-reducibility.
  • f ≤W g if given x, one can compute f (x) applying g once.
  • f ≡W g if f ≤W g and g ≤W f .
slide-16
SLIDE 16

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Consider two representations En and Cf of 2N: En(p) = {n ∈ N : 100n1 is a subword of p}, Cf(p) = {n ∈ N : pn = 1}. Let E ⊆ N:

  • E is r.e. ⇐

⇒ it is En-computable,

  • E is recursive ⇐

⇒ it is Cf-computable. Let EC : (2N, En) → (2N, Cf) be the identity: it is not computable for these representations.

slide-17
SLIDE 17

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Properties of EC

  • ∆0

2 objects can be computed from one application of EC (subsets of

N, real numbers, real fonctions, points of computable metric spaces, etc.)

  • Actually, for f : NN → NN,

f ∈ ∆0

2 ⇐

⇒ f ≤W EC.

  • Let J(X) be the Turing jump of X ⊆ N: J ≡W EC.
  • EC ≡W limR.

Due to Brattka, Gherardi, Yoshikawa, Marcone (2005-2011).

slide-18
SLIDE 18

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Properties of EC

Used to classify mathematical theorems: to a theorem ∀X∃YP(X, Y ) is associated the operator X → Y (possibly multi-valued).

Due to Brattka, Gherardi, Yoshikawa, Marcone (2005-2011).

slide-19
SLIDE 19

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Properties of EC

Used to classify mathematical theorems: to a theorem ∀X∃YP(X, Y ) is associated the operator X → Y (possibly multi-valued).

  • Let FR be the operator associated to the Fréchet-Riesz

representation theorem (on suitable spaces): EC ≡W FR.

Due to Brattka, Gherardi, Yoshikawa, Marcone (2005-2011).

slide-20
SLIDE 20

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

Properties of EC

Used to classify mathematical theorems: to a theorem ∀X∃YP(X, Y ) is associated the operator X → Y (possibly multi-valued).

  • Let FR be the operator associated to the Fréchet-Riesz

representation theorem (on suitable spaces): EC ≡W FR.

  • Let BWR be the Bolzano-Weierstrass operator:

EC <W BWR.

Due to Brattka, Gherardi, Yoshikawa, Marcone (2005-2011).

slide-21
SLIDE 21

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

EC ≤W RN

Let RN be the Radon-Nikodym operator, that maps µ ≪ λ to dµ

dλ ∈ L1(λ).

Corollary

EC ≤W RN.

Proof.

Given an enumeration of E ⊆ N:

1 construct KE such that λ(KE) = n / ∈E 2−n, 2 apply RN to compute λ(KE), 3 compute E from λ(KE).

enumeration of E

  • EC
  • KE

µE

RN

  • characteristic function of E

λ(KE)

  • dµE

slide-22
SLIDE 22

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

A classical proof of the Radon-Nikodym theorem works as follows:

  • apply the Fréchet-Riesz representation theorem to the continuous

linear operator φµ : L2(λ + µ) → R f →

  • f dµ.

It gives g ∈ L2(λ + µ) such that for all f ∈ L2(λ + µ), φµ(f ) = f , g, i.e.

  • f dµ

=

  • fg d(λ + µ).
  • show that

g 1−g has the required properties for being dµ dλ.

slide-23
SLIDE 23

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

To compute the Radon-Nikodym derivative, µ

  • RN
  • φµ

FR

dλ = g 1−g ∈ L1(λ)

g ∈ L2(λ + µ)

  • ne shows that from g ∈ L2(λ + µ) one can compute g ∈ L1(λ), knowing

that

  • g dλ = 1.

(a simple proof can be obtained using Martin-Löf randomness!)

slide-24
SLIDE 24

Radon-Nikodym theorem A counter-example Weihrauch degrees A computable Fréchet-Riesz representation theorem

It was proved by Brattka and Yoshikawa that on suitable spaces, FR ≡W EC. Hence we get EC ≤W RN ≤W FR ≡W EC.

FR: Fréchet-Riesz RN: Radon-Nikodym EC: Enumeration → Characteristic function