Section 2, property (3) — privacy across independent refreshes
ProvedShamir1979.independent_refresh_privacyFor prime and , take any finite collection of sharing editions. Every edition has the same secret constant and a freshly, independently, uniformly sampled vector of nonconstant coefficients. The observed coalition may differ between editions but must contain fewer than participants in each. Every specified combined transcript then has probability , independent of the secret. The finite collection may be empty. Independent resampling is part of the sample space; correlated refresh randomness is not covered.
import Definitions.Def_Shamir1979_Model open scoped BigOperators universe u
namespace Shamir1979
theorem independent_refresh_privacy (p n k : Nat) [Fact (Nat.Prime p)]
(hk : 0 < k) (hkn : k ≤ n) (hnp : n < p)
{E : Type u} [Fintype E] (groups : E → Finset (Fin n))
(hgroups : ∀ e, (groups e).card < k) (secret : ZMod p)
(values : (e : E) → Observation p (groups e)) :
epochProbability (k := k) groups secret values =
(1 : Rat) / (p : Rat) ^ (∑ e : E, (groups e).card) := by sorry
end Shamir1979Read-back
What the Lean code literally says, in plain math · OpenAI Codex (inherited session model)
For every prime natural number , natural numbers with , every finite type in any universe, every function assigning to each a finite subset with , every secret , and every family of prescribed functions , let the randomness space be the full finite Cartesian product . Thus a sample specifies one coefficient function for each , and uniform choice from encodes mutually independent uniform coefficient functions, while using the same secret for every . For that sample define , and observe, for each and each , the value in . The assertion is the exact rational identity , where . No disjointness is required among the : they may coincide or overlap, and the same index occurring in several groups remains a separately prescribed observation of the separately chosen polynomial for each such ; the exponent sums the group sizes and does not take the size of their union. Every group may be empty, and itself may be empty; empty groups have a unique prescribed function and contribute zero to the exponent, and for empty both the randomness space and the prescribed observation family are singletons and the asserted probability is . When , every is forced to be empty, every coefficient function has empty domain, and the asserted probability is again . The hypotheses exclude and , and primality ensures that all probability denominators are nonzero.
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.