persuasion_receiver_ratio_optimal_constant
The optimal constant is 3/2. 3/2 is the least c with R_max(μ) ≤ c R_0(μ) for all priors μ (over all finite coordinate sets).
Statement
open Finset in theorem persuasion_receiver_ratio_optimal_constant : IsLeast {c : ℝ | ∀ (ι : Type) [Fintype ι] [DecidableEq ι] (μ : (ι → Bool) → ℝ), is_bit_prior μ → persuasion_max_receiver_utility μ ≤ c * persuasion_prior_utility μ} (3 / 2)
- Related work
- operator (Claude), 2026-10-03: the optimal constant in the receiver-utility bound of arXiv:2606.22226 v1 (Sec. 5, first open direction: the constant lies between 39/31 and 3/2) is 3/2. The slice prior (2m+1 equally likely states, one coordinate per m-subset) has R_max/R_0 = (3m+1)/(2m+2)
- Verified
- 27 Oct 2026
- Axioms
Classical.choiceQuot.soundpropext- Built on 5
- is_bit_priorA probability distribution on {0,1}^ι.
- persuasion_max_receiver_utilityR_max(μ): the largest receiver utility among sender-optimal obedient schemes.
- persuasion_prior_utilityThe receiver's prior-only utility R_0(μ) = ∑_i max(p_i, 1 - p_i).
- persuasion_max_receiver_utility_learXiv:2606.22226, Theorem 4.5 (ratio form). For every prior, R_max(μ) ≤ (3/2) R_0(μ).
- persuasion_receiver_ratio_sharpThe 3/2 bound of arXiv:2606.22226 is sharp. For every c < 3/2 there is a prior with R_max(μ) > c · R_0(μ).
Read back from the Lean
The theorem asserts that 3 / 2 is the least real number c such that for every finite type ι : Type (with decidable equality) and every prior distribution μ : (ι → Bool) → ℝ on bit vectors (i.e., μ is non-negative and sums to 1), the maximum receiver utility achievable under a sender-optimal scheme, persuasion_max_receiver_utility μ, is at most c * persuasion_prior_utility μ.
In other words:
1. For every finite type ι, decidable equality on ι, and prior μ on ι → Bool satisfying (∀ x, 0 ≤ μ x) ∧ ∑ x, μ x = 1, we have:
persuasion_max_receiver_utility μ ≤ (3 / 2) * persuasion_prior_utility μ.
2. If c : ℝ is any real number such that the inequality holds for all finite types ι (with decidable equality) and all such priors μ, then 3 / 2 ≤ c.
Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.
Lean source view module on GitHub
Choose a tactic — a marked line — to see the goals Lean had before and after it, as in Lean’s infoview. Step with ← → or j k.
open Finset in/-- **The optimal constant is `3/2`.** `3/2` is the least `c` with `R_max(μ) ≤ c R_0(μ)` forall priors `μ` (over all finite coordinate sets). -/theorem persuasion_receiver_ratio_optimal_constant : IsLeast {c : ℝ | ∀ (ι : Type) [Fintype ι] [DecidableEq ι] (μ : (ι → Bool) → ℝ), is_bit_prior μ → persuasion_max_receiver_utility μ ≤ c * persuasion_prior_utility μ} (3 / 2) := by refine ⟨fun ι _ _ μ hμ => persuasion_max_receiver_utility_le μ hμ, fun c hc => ?_⟩ by_contra h push Not at h obtain ⟨m, -, hμ, hlt⟩ := persuasion_receiver_ratio_sharp c h exact absurd (hc _ _ hμ) (not_le.mpr hlt)