Documentation

MIPRE.Background.Tailored.Intro.Scale

The output at C λ, read at λ #

The tailored presentation of Introspection.seven at λ runs seven's construction at C λ, where C is the constant of TailoredVerifier.ofTNFVT_isBounded (the input's normal form verifier is C λ-bounded). This file collects what that rescaling needs:

Multiplication by a power of two #

Budgets at C λ #

2^{(C λ n + 1)^{C'}} ≤ 2^{(λ n + 1)^{C C'}}.

theorem MIPRE.Tailored.Intro.Scale.LenBound.mono {L : Decider} {n B B' : ℕ} (h : LenBound L n B) (hB : B ≤ B') :
LenBound L n B'
theorem MIPRE.Tailored.Intro.Scale.lenIntro_within :
∃ (C' : ℕ), ∀ (C : ℕ), 1 ≤ C → ∀ (lam n : ℕ), (LenIntro.lenIntro (C * lam)).TimeBoundAt n (Introspection.ansBound (C * C') lam n) (C * C') ∧ LenBound (LenIntro.lenIntro (C * lam)) n (Introspection.ansBound (C * C') lam n)

The answer-length calculator at C λ is within the budget at λ, with the exponent multiplied by C.

Index zero #

theorem MIPRE.Tailored.Intro.Scale.hasPerfectZPC_of_zero {ℓ : ℕ} (W : TailoredVerifier ℓ) (n : ℕ) (hlen : ∀ (x : Cost.BitStr) (κ : Bool), LenIs W.len n x κ 0) (hlp : ∀ (x y aR bR : Cost.BitStr), LpIs W.lp n x y aR bR []) :

An output that is empty at index n has a perfect ZPC strategy there: when the calculator outputs 0 and the processor no constraint, every pair of empty answers is accepted, and the trivial strategy is perfect.