Lemma 15, p. 57 — on G, levels ℓ ≥ ℓ⋆ never hold an incorrect edge and hold edge (i, j) once η̂_i, η̂_j ≥ γ_{j,i}
ProvedFakeUserRanking.FORC.lemma_15Throughout, products with click probabilities in (condition (1)) and exit probabilities are shown to users, at most of whom are fake, controlled by an adaptive randomized policy that sees only the past history and puts no mass on fake users once rounds were fake. FORC (Algorithm 3) runs levels with ; it does not know , or . Its unspecified choices (tie-breaking in GraphRankSelect, the "arbitrary ranking", and Step 6 when no level qualifies) are arbitrary functions of its past, and the statement holds for all of them.
Lemma 15 (Properties of Graph under Event ). Assume that the event of Definition 1 holds. Then for any two products with (so ) and any level :
- at the end of every round , the graph does not contain the incorrect edge ;
- if at the end of round
then contains the correct edge at the end of every round .
The statement is pathwise: it holds on every realization in . It shows that the high levels are never eliminated and learn every pair once both products have enough cross-learned feedback; it is the basis of Lemmas 7 and 10.
Formalization Note. Products, positions, levels and rounds are 0-based in Lean (Fin n, Fin L, Fin T): Lean product is paper product , and Lean level is paper level , so the paper's condition is written . "Round " means FORC's state after the first rounds (Steps 8–11 included); clause 1 holds for and clause 2 for . j < i in Fin n is the paper's .
Differences from the source:
- encoding: products, positions, levels and rounds indexed from 1; paper level ℓ; ℓ ≥ ℓ⋆ = ⌈log₂F⌉ → Fin n, Fin L, Fin T, all 0-based; Lean level k is paper level ℓ = k + 1; ℓ ≥ ⌈log₂F⌉ is written F ≤ 2^(k+1); reason: Lean convention; same instances (F ≤ 2^ℓ is exactly ℓ ≥ ⌈log₂F⌉, and holds for every level when F = 0)
- encoding: expectations and probabilities over the random history H_T → finite sums over trajectories Fin T → Step, weighted by the product of the level weights, the policy's PMF and the real customer's cascade law; reason: every space of the model is finite, so this is the law of the history exactly
- reading: L = log₂(T) levels → L = ⌈log₂T⌉ (Nat.clog 2 T); reason: the number of levels must be an integer; the ceiling keeps 2^L ≥ T (paper.md Slip 1)
- reading: Step 8, no click at the last examined position: r ← r(η − 1)/(η + 1) → r^(ℓ)_i = (clicks of i at level ℓ)/η^(ℓ)_i, the average reward; reason: Step 3 calls r the average reward; FAR's Step 6 has r(η − 1)/η and the proofs use r·η = sum of samples (paper.md Slip 2)
- encoding: Step 9 updates η̂, r̂ incrementally, only for levels ℓ ≥ ℓ_t and examined products → η̂, r̂ in closed form (17)–(18) from the current counts, for every level; reason: η^(g) with g ≤ ℓ changes only in rounds with ℓ_t ≤ ℓ, which trigger the update, so the stored values always equal the closed form
- reading: ties among the smallest η in GraphRankSelect, its 'arbitrary ranking' on a cyclic graph, and Step 6 when no non-eliminated level above ℓ_t exists are left unspecified → universally quantified functions tb (a member of the tied set) and fb of FORC's past and the sampled level; reason: the claim is made for the algorithm as printed, so it must hold for every resolution of its unspecified choices (paper.md Slip 8)
- encoding: n ≥ 1 products and T rounds, unrestricted → 1 ≤ n and 2 ≤ T; reason: δ = 1/(n³T) is undefined at n = 0, and T ≤ 1 gives L = ⌈log₂T⌉ = 0 levels, where FORC draws no level and the statement is vacuous
- encoding: 'Assume that event G holds … in any round t' → for every trajectory in goodEvent; round t is FORC's state after t rounds, t ≤ T (clause 1), 1 ≤ t ≤ t' ≤ T (clause 2); reason: the lemma is a deterministic consequence of G; the horizon has rounds 1..T
import Mathlib import Definitions.Def_FakeUserRanking_FORC_Model import Definitions.Def_FakeUserRanking_FORC_Algorithm import Definitions.Def_FakeUserRanking_FORC_GoodEvent
namespace FakeUserRanking.FORC
/-- Lemma 15 (Properties of Graph G under Event G), p. 57, pathwise: on every trajectory in the
good event, for products `j < i` (paper: `i > j`, so `μ_j > μ_i`) and every level `ℓ ≥ ℓ⋆`
(Lean: `F ≤ 2^(k+1)`), (1) `G^(ℓ)` never holds the incorrect edge `(j, i)`, and (2) once
`η̂^(ℓ)_{i,t}, η̂^(ℓ)_{j,t} ≥ γ_{j,i}` at the end of round `t`, `G^(ℓ)` holds the edge `(i, j)` at the
end of every round `t' ≥ t`. -/
theorem lemma_15 (n T F : ℕ) (hn : 1 ≤ n) (hT : 2 ≤ T)
(inst : Instance n) (hμ : StrictAnti inst.μ)
(tb : List (Fin (numLevels T) × Obs n) → Fin (numLevels T) → Finset (Fin n) → Fin n)
(htb : ∀ h k S, S.Nonempty → tb h k S ∈ S)
(fb : List (Fin (numLevels T) × Obs n) → Fin (numLevels T) → Equiv.Perm (Fin n))
(τ : Fin T → Step n (numLevels T)) (hG : goodEvent inst F tb fb τ)
(j i : Fin n) (hji : j < i) (k : Fin (numLevels T)) (hk : F ≤ 2 ^ ((k : ℕ) + 1)) :
(∀ t : ℕ, t ≤ T → (j, i) ∉ (stateAt tb fb τ t).graph k) ∧
(∀ t : ℕ, 1 ≤ t → t ≤ T →
gamma T inst j i ≤ etaHat (stateAt tb fb τ t) k i →
gamma T inst j i ≤ etaHat (stateAt tb fb τ t) k j →
∀ t' : ℕ, t ≤ t' → t' ≤ T → (i, j) ∈ (stateAt tb fb τ t').graph k) := by sorry
end FakeUserRanking.FORC
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.