While this submission is a draft, it cannot be used by other submissions.

Quantitative soundness at a fixed evaluation point

Lax253009.CNAPointSoundness · concepts/Lax253009/CNAPointSoundness.lean · lax-253009

proven

Loading review…

Sign in with ORCID

Community review

Flags

Each flag is tied to a public ORCID identity and explains why this concept may be incorrect.

No flags have been submitted.

    Community review

    Flag this concept

    State precisely what appears incorrect. This explanation will be public under your ORCID name.

    No source line selected.

    Natural Language Statement

    Theorem

    The three Fourier estimates apply to the actual event that every balanced query agrees with evaluation at a fixed point while its label is absent from the small decoding set. This is the fixed-point step in Section 4.1. The table may be real valued and bounded by one, so the same statement applies after averaging over a side condition.

    Concept map
    18 concepts; 1 descendant hidden
    100%
    Proven claimThis conceptRelated conceptA → B: B builds on A
    Evidence

    Lean source view on GitHub

    1import Lax253009.FourierDecoding
    2import Lax253009.HighDegreeSoundness
    3import Lax253009.SmallCoefficientSoundness
    4import Lax253009.LargeCoefficientSoundness
    5
    6/-!
    7---
    8title: Quantitative soundness at a fixed evaluation point
    9type: theorem
    10---
    11The three Fourier estimates apply to the actual event that every balanced
    12query agrees with evaluation at a fixed point while its label is absent
    13from the small decoding set. This is the fixed-point step in Section 4.1.
    14The table may be real valued and bounded by one, so the same statement
    15applies after averaging over a side condition.
    16-/
    17
    18namespace Lax253009.CNAPointSoundness
    19
    20open BooleanFourier BalancedPredicates FiniteProbability
    21
    22noncomputable def decoding {ι : Type} [Fintype ι] [DecidableEq ι]
    23 (F : Cube ι → ℝ) (l : ℕ) (τ : ℝ) : Finset ι :=
    24 SmallSupport.decodingSet (coefficient F) l (τ ^ 2 / l)
    25
    26def Matches {ι κ : Type} [Fintype κ] [DecidableEq κ]
    27 (F : Cube ι → ℝ) (n : ℕ) (f : ι → κ) (y : ι) : Prop :=
    28 ∀ B : predicates κ n, F (fun i ↦ B.val (f i)) = sign (B.val (f y))
    29
    30def Avoids {ι κ : Type} (D : Finset ι) (f : ι → κ) (y : ι) : Prop :=
    31 ∀ x ∈ D, f x ≠ f y
    32
    33noncomputable def pointBound (l m r N : ℕ) (τ q : ℝ) : ℝ :=
    34 9 * (q ^ l + 2 * (N + 1 : ℝ) * Real.exp (-(N : ℝ) * q ^ 2 / 2)) +
    35 (3 : ℝ) ^ m * SmallCoefficientSoundness.boundValue l m r N τ q
    36
    37axiom decoding_card {ι : Type} [Fintype ι] [DecidableEq ι]
    38 (F : Cube ι → ℝ) (hF : ∀ x, |F x| ≤ 1)
    39 (l : ℕ) (hl : 0 < l) (τ : ℝ) (hτ : 0 < τ) :
    40 ((decoding F l τ).card : ℝ) ≤ (l : ℝ) / τ ^ 2
    41
    42axiom point_soundness {ι κ : Type} [Fintype ι] [DecidableEq ι]
    43 [Fintype κ] [DecidableEq κ] [Nonempty κ]
    44 (F : Cube ι → ℝ) (hF : ∀ x, |F x| ≤ 1)
    45 (n : ℕ) (hn : 0 < n) (hN : Fintype.card κ = 2 * n)
    46 (l m r : ℕ) (hl : 0 < l) (hlN : l < Fintype.card κ)
    47 (τ q : ℝ) (hτ : 0 < τ) (hq : 0 ≤ q) (hq1 : q ≤ 1)
    48 (hr : 2 * r ≤ m) (hm : Even m)
    49 (hlarge : (l : ℝ) / ((Fintype.card κ : ℝ) - l) / τ ≤ 1 / 3)
    50 (y : ι) :
    51 probability (fun f : ι → κ ↦ Matches F n f y ∧ Avoids (decoding F l τ) f y) ≤
    52 pointBound l m r (Fintype.card κ) τ q
    53
    54end Lax253009.CNAPointSoundness
    55
    Show ProofShow Proof

    Discussion

    Ask a question or add context. Endorsements and structured flags are kept in the review panel above.

    Loading discussion…