Theorem 11 — the adaptive Naor–Yung scheme ; success given the sampler succeeds
OpenRestrictedDec.Supplement.theorem_11Let be a common message space, an observation with the two public conditions of §2.3 and label length , and any uniformly efficient map on decryption outputs. Let be an efficient, ordinarily correct residual scheme over ; an efficient label scheme over with (KC); an efficient, perfectly complete NIZK for (B.4); and an efficient, correct, strongly unforgeable one-time signature. Suppose a uniform PPT sampler, run independently of the generated keys, finds valid distinct equal-length with
with non-negligible probability on infinitely many security parameters. Then
- the signed scheme is not in ;
- there are a uniform PPT attacker and a negligible such that, for every nonnegative function bounding the honest decryption error of (Definition 3), every and every coin vector on which the sampler succeeds, the attacker run with as its sampler coins guesses the challenge bit of the adaptive experiment correctly with probability at least
The proof constrains only the labels, so a fresh, independently signed ciphertext can reuse the challenge's residual component; a finer oracle then reveals the residual message.
Differences from the source:
- encoding: keys, messages, ciphertexts, labels, M_⊥, coins, PPT adversaries and reductions → bit strings with tuple codes, Option Str (none = ⊥), explicit uniform coin vectors, uniform reactive machines with a polynomial-time step function (RestrictedDec.Adversary); reason: the model of the published base mission; same instances
- encoding: Sec^t(q) = Sec^t(F_q), F_q = {g ∘ q : g an admissible circuit / efficient public postprocessing} → SecM t M (famObs q); reason: the published famObs is the class of §2.3, p.14
- encoding: PPT primitives (R, L, I, P, S, N, q, samplers) → separate efficiency hypotheses (IsSchemeM, IsEfficient, IsPublic, IsObsMap); reason: the paper's PPT convention; the reductions run these algorithms
- generalization: conditional on the sampler succeeding, the attacker guesses correctly with probability at least 1 − δ_R − 2δ_L − negl(λ) → the bound for each successful coin vector ρ of the sampler, the attacker started with ρ as its state, with one negligible ν uniform in ρ, λ and δ_R; reason: averaging the pointwise bound over the successful ρ gives the conditional probability
- generalization: the standing assumptions of App. B.1 on the construction (CPA security of R and L, the NIZK's soundness, zero knowledge and simulation soundness, efficient recognition of plaintexts) → only the properties the attack uses: correctness of R, (KC) of L, NIZK completeness, signature correctness and strong unforgeability, and efficiency; reason: fewer hypotheses; the paper's setting satisfies them
- reading: δ_R (R's honest error) and δ_L in the bound → every nonnegative δ_R with ErrBound M R δ_R (Definition 3), and δ_L = kcFail L ℓ, the exact (KC) failure probability; reason: App. B.1, p.35 ('negligible fixed-message error bound δR'; (KC)); an error bound is a probability bound, and a negative δ_R (possible only where R never errs) would push the lower bound above 1
- encoding: a uniform PPT sampler finding the pair with nonnegligible probability on infinitely many security parameters → a uniform PPT Sampler whose success probability is not negligible; reason: a function is non-negligible iff it exceeds some inverse polynomial infinitely often
- encoding: the common message space → hypothesis MsgSpace.Found M; reason: standing assumption of §2.1, p.13
import Mathlib import Definitions.Def_RestrictedDec_famObs import Definitions.Def_RestrictedDec_Supplement_MsgSpace import Definitions.Def_RestrictedDec_Supplement_IsSchemeM import Definitions.Def_RestrictedDec_Supplement_SecM import Definitions.Def_RestrictedDec_Supplement_INDCPAM import Definitions.Def_RestrictedDec_Supplement_kcFail import Definitions.Def_RestrictedDec_Supplement_Sampler import Definitions.Def_RestrictedDec_Supplement_nyAd open QAlgorithms.Complexity set_option autoImplicit false
namespace RestrictedDec
namespace Supplement
/-- **Theorem 11** (adaptive finer-observation separation for the signed Naor–Yung scheme). Let `M`
be a common message space, `q` an observation with the two public conditions of §2.3 and label
length `ℓ(λ)`, and `r` any uniformly efficient map on decryption outputs. Let `R` be an efficient,
ordinarily correct residual scheme over `M`; `L` an efficient label scheme over `{0,1}^{ℓ(λ)}`
with (KC); `N` an efficient, perfectly complete NIZK for (B.4); and `S` an efficient, correct,
strongly unforgeable one-time signature. Suppose a uniform PPT sampler `W`, run independently of the
generated keys, finds valid distinct equal-length `m₀, m₁` with `q(m₀) = q(m₁)` and
`r(m₀) ≠ r(m₁)` with non-negligible probability on infinitely many security parameters. Then
1. the signed scheme `Π_q` is not in `Sec^ad(r)`;
2. there are a uniform PPT attacker `A`, which reads the sampler's coins `ρ` as its initial
state, and a negligible `ν` such that, for every nonnegative bound `δ_R` on the honest decryption
error of `R` (Definition 3), every `λ` and every `ρ` on which the sampler succeeds, `A` guesses the
challenge bit of `Exp^{IND,ad}_{Π_q,𝓕_r}` correctly with probability at least
`1 − δ_R(λ) − 2δ_L(λ) − ν(λ)`.
Nguyen–Susilo, STOC 2027 submission (manuscript sha256 6aeac40c…), Theorem 11, p.43; construction
`Π_q`, p.39.
**Formalization Note.** "Conditional on the sampler succeeding" is stated for each successful coin
vector `ρ` of the sampler (the attacker run with `ρ` as initial state); averaging over the
successful `ρ` gives the conditional probability. `δ_L` is the exact (KC) failure probability
`kcFail`. -/
theorem theorem_11 (M : ℕ → Str → Prop) (hM : MsgSpace.Found M)
(o : Obs) (ho : o.IsPublic M) (ℓ : ℕ → ℕ)
(hℓ : ∀ lam (x : Option Str), (∀ m, x = some m → M lam m) → (o.q lam x).length = ℓ lam)
(r : ℕ → Option Str → Str) (hr : IsEffMap M r)
(R : PKE) (hR : IsSchemeM M R) (hRcor : CorrectM M R)
(L : PKE) (hL : IsSchemeM (fun lam y => y.length = ℓ lam) L) (hLkc : KeywiseCorrect L ℓ)
(S : OTS) (hS : S.IsEfficient) (hSsuf : S.SUF) (hScor : S.Correct)
(N : NIZK) (hN : N.IsEfficient) (hNc : N.Complete (relB4 L o))
(W : Sampler) (hW : W.IsEfficient)
(hWs : ¬ NoisyOWMPC.Negligible (W.succ fun lam u v => M lam u ∧ M lam v ∧ u ≠ v ∧
u.length = v.length ∧ o.q lam (some u) = o.q lam (some v) ∧ r lam (some u) ≠ r lam (some v))) :
¬ SecM .ad M (famObs r) (nyAd M o R L N S) ∧
∃ A : Adversary, A.Valid ∧ ∃ ν : ℕ → ℝ, NoisyOWMPC.Negligible ν ∧
∀ δR : ℕ → ℝ, (∀ lam, 0 ≤ δR lam) → ErrBound M R δR → ∀ (lam : ℕ) (ρ : Fin (W.coins lam) → Bool),
(M lam (W.out lam ρ).1 ∧ M lam (W.out lam ρ).2 ∧ (W.out lam ρ).1 ≠ (W.out lam ρ).2 ∧
(W.out lam ρ).1.length = (W.out lam ρ).2.length ∧
o.q lam (some (W.out lam ρ).1) = o.q lam (some (W.out lam ρ).2) ∧
r lam (some (W.out lam ρ).1) ≠ r lam (some (W.out lam ρ).2)) →
ENNReal.ofReal (1 - δR lam - 2 * kcFail L ℓ lam - ν lam) ≤
expINDM .ad M (nyAd M o R L N S) (famObs r) A lam (List.ofFn ρ) true := by sorry
end Supplement
end RestrictedDec