On the Query Complexity of Real Functionals Hugo Fre, Walid Gomaa, - - PowerPoint PPT Presentation

on the query complexity of real functionals
SMART_READER_LITE
LIVE PREVIEW

On the Query Complexity of Real Functionals Hugo Fre, Walid Gomaa, - - PowerPoint PPT Presentation

Outline Introduction Complexity of Norms Query Complexity One Oracle Access On the Query Complexity of Real Functionals Hugo Fre, Walid Gomaa, Mathieu Hoyrup Hugo Fre, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real


slide-1
SLIDE 1

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

On the Query Complexity of Real Functionals

Hugo Férée, Walid Gomaa, Mathieu Hoyrup

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-2
SLIDE 2

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

1

Introduction

2

Complexity of Norms

3

Query Complexity

4

One Oracle Access

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-3
SLIDE 3

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

1

Introduction

2

Complexity of Norms

3

Query Complexity

4

One Oracle Access

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-4
SLIDE 4

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Computing with Real Numbers

(qn)n∈N x if ∀n, |x − qn| ≤ 2−n R ∼ (N → N) (1) Model: Oracle Turing Machines Finite-time computation − → finite number of queries − → finite knowledge of the input − → continuity Bounding computation time ⇐ ⇒ bounding the computational power the number of queries

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-5
SLIDE 5

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Computing with Real Numbers

(qn)n∈N x if ∀n, |x − qn| ≤ 2−n R ∼ (N → N) (1) Model: Oracle Turing Machines Finite-time computation − → finite number of queries − → finite knowledge of the input − → continuity Bounding computation time ⇐ ⇒ bounding the computational power the number of queries

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-6
SLIDE 6

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Computing with Real Numbers Cont’d

f ∈ C[0, 1] ⇐ ⇒ ∃µ, fQ : modulus of continuity : µ : N → N |x − y| ≤ 2−µ(n) = ⇒ |f(x) − f(y)| ≤ 2−n approximation function fQ : Q × N → Q |fQ(q, n) − f(q)| ≤ 2−n

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-7
SLIDE 7

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Computing with Real Numbers Cont’d

Theorem f : R → R is computable w.r.t. an oracle ⇐ ⇒ f is continuous. Theorem f : R → R is polynomial time computable w.r.t. to an oracle ⇐ ⇒ its modulus of continuity is bounded by a polynomial.

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-8
SLIDE 8

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

1

Introduction

2

Complexity of Norms

3

Query Complexity

4

One Oracle Access

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-9
SLIDE 9

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Dependence of a Norm on a Point

F is a norm over C[0, 1] f ∈ C[0, 1], α ∈ [0, 1] ∆α implies ∆F(f)

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-10
SLIDE 10

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Dependence of a Norm on a Point

Two problems

1

f is continuous, so must change a neighborhood of α,

2

if f = 0, any change causes F(f) = 0 to change to some positive value.

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-11
SLIDE 11

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Dependence of a Norm on a Point

Quantitative measure

F is always assumed weaker than the uniform norm dF,α(n) = sup{l : ∃f ∈ Lip1, Supp(f) ⊆ N(α, 1/l) and F(f) > 2−n} 1/l f

  • α

dF,α ≤ c2n, non-decreasing, unbounded

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-12
SLIDE 12

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Examples

The uniform norm

The uniform norm is monotonic: F = ||.||∞ = ⇒ dF,α(n) ∼ 2n (2)

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-13
SLIDE 13

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Examples

The L1-norm

The L1-norm is monotonic: F = ||.||1 = ⇒ dF,α(n) ∼ 2

n 2

(3)

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-14
SLIDE 14

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Some Properties of the Dependency Function

Proposition

1

α → dF,α(n) is continuous

2

F is weaker than G = ⇒ dF,α(n) ≤ dG,α(n + k) Maximal dependence: DF(n) = maxα∈[0,1] dF,α(n) Proposition For F weaker than the uniform norm: c12

n 2 ≤ DF(n) ≤ c22n

(4)

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-15
SLIDE 15

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Relevant Points

Rn,l = {α : dF,α(n) ≥ l} Definition α is relevant if ∃c > 0, ∀n, dF,α(n) ≥ c · 2

n 2

R =

  • k
  • n

Rn,2

n 2 −k

  • Rk

Example For ||.||∞ and ||.||1, R = [0, 1].

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-16
SLIDE 16

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Relevant Points

Example

Let Q = {q0, q1, . . . , } be some particular canonical enumeration of the dyadic rationals Define F(f) =

  • i

2−i|f(qi)| Then dF,qi(n) ≥ 2n−i, for n ≥ i dF,α(n) ≤ n2 ε R = D

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-17
SLIDE 17

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Relevant Points

Properties

Theorem R is dense. Theorem f = 0 on R2µf (k) = ⇒ F(f) ≤ c.2−k. Corollary f = g on Rµ(k) = ⇒ |F(f) − F(g)| ≤ 2−k.

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-18
SLIDE 18

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

R is Dense (Proof)

F(h0) ≥ 2−c

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-19
SLIDE 19

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

R is Dense (Proof)

F(h1) ≥ 2−c−2

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-20
SLIDE 20

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

R is Dense (Proof)

F(h2) ≥ 2−c−4

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-21
SLIDE 21

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

R is Dense (Proof)

F(hn) = F(hαn,2−n−p) ≥ 2−c−2n = ⇒ F(2n + c) ≥ 2n+p (αn) → α dF,α(2n + c) ≥ 2n+p−1 = ⇒ α ∈ R

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-22
SLIDE 22

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

1

Introduction

2

Complexity of Norms

3

Query Complexity

4

One Oracle Access

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-23
SLIDE 23

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Query Complexity

Definition Qn : oracle calls of F on x → 0 with precision 2−n. Proposition Rn,l ⊆ N(Qn+1, 1 l ) Definition F has a polynomial query complexity if it is computable by a relativized OTM with |Qn| ≤ P(n).

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-24
SLIDE 24

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Query Complexity Cont’d

Theorem If F has polynomial query complexity, then almost every point has a polynomial dependency (dF,α ∈ P for almost all α). Theorem If F has polynomial query complexity, then R has Hausdorff dimension 0. Proposition F has polynomial query complexity = ⇒ ∃α,

2n dF,α(n) is bounded by a polynomial.

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-25
SLIDE 25

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

Query Complexity

Characterizing polynomial time computable norms

Theorem F is polynomial time computable w.r.t. an oracle ⇐ ⇒ F has polynomial query complexity ⇐ ⇒ Rk can be polynomially covered (wrt. k). Rk = {α ∈ [0, 1]: ∀n, dF,α(n) ≥ 2

n 2 −k} =

  • n

Rn,2

n 2 −k

Open question Can it be generalized for any F : C[0, 1] → R?

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-26
SLIDE 26

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

1

Introduction

2

Complexity of Norms

3

Query Complexity

4

One Oracle Access

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-27
SLIDE 27

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

One Oracle Access Case

Theorem The following are equivalent: F is computable by a polynomial time machine doing only

  • ne oracle query

∀f, F(f) = φ(f(α)) where:

α ∈ Poly(R) (but cannot be efficiently retrived from F!) φ ∈ Poly(R → R) φ is uniformly continuous

Open question Generalization to any finite number of queries?

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals

slide-28
SLIDE 28

Outline Introduction Complexity of Norms Query Complexity One Oracle Access

THANK YOU

Hugo Férée, Walid Gomaa, Mathieu Hoyrup On the Query Complexity of Real Functionals