Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

The weights of the two blocks outside Λ+2\Lambda^2_+Λ+2​

Proved
BunkbedAux.xc_and_A_weight

by burkh4rt · Sep 12, 2026 · Mathlib 0df444a (Lean v4.33.1)

bunkbedcombinatoricspercolationprobability

Once the configurations meeting Λ+2\Lambda^2_+Λ+2​ have been cancelled by the involution, the robustness lemma reduces to the complementary set: those in which every hyperedge avoids Λ+2\Lambda^2_+Λ+2​. This theorem computes the total weight of that set, and of the sub-block that carries the refutation.

Write W=PabcPa∣b∣c−Pab∣cPac∣bW = P_{abc}P_{a|b|c} - P_{ab|c}P_{ac|b}W=Pabc​Pa∣b∣c​−Pab∣c​Pac∣b​ and let PPP be a probability distribution on the five Wierman--Ziff states. Then:

  1. the configurations all of whose hyperedges avoid Λ+2\Lambda^2_+Λ+2​ have total weight
Z6,Z=2W+2Pa∣bc−Pa∣bc2;Z^{6}, \qquad Z = 2W + 2P_{a|bc} - P_{a|bc}^{2};Z6,Z=2W+2Pa∣bc​−Pa∣bc2​;
  1. the configurations all of whose hyperedges are in a −-− state have total weight
(2W)6=64 W6.(2W)^{6} = 64\,W^{6}.(2W)6=64W6.

Both follow from the per-hyperedge weights by independence across hyperedges: a set of configurations specified by a condition on each coordinate separately is a product, so its total weight is the product of the per-coordinate weights, and here that product is a sixth power since the condition is the same at every hyperedge. The per-hyperedge figure ZZZ counts the two −-− variants, of weight WWW each, together with the nine pairs involving the state a∣bca|bca∣bc, whose weight is 2Pa∣bc−Pa∣bc22P_{a|bc} - P_{a|bc}^22Pa∣bc​−Pa∣bc2​ by inclusion--exclusion; the normalisation ∑sPs=1\sum_s P_s = 1∑s​Ps​=1 is used exactly there.

Mathematical role. The ratio of the two is the conditional probability that every hyperedge is in a −-− state, given that all avoid Λ+2\Lambda^2_+Λ+2​. Those 26=642^6 = 6426=64 all-minus configurations are precisely the alternative bunkbed hypergraph model, and by Hollom's computation they contribute a net surplus of one cross-level connection — the source of the refutation. All other configurations in the complement are error terms, bounded in absolute value by their total weight Z6−(2W)6Z^6 - (2W)^6Z6−(2W)6. Under Eq. [1], which reads 400Pa∣bc≤W400P_{a|bc} \le W400Pa∣bc​≤W, one has Z≤2W⋅401400Z \le 2W \cdot \frac{401}{400}Z≤2W⋅400401​, and the refutation survives because

64⋅4016  <  65⋅4006,64 \cdot 401^{6} \;<\; 65 \cdot 400^{6},64⋅4016<65⋅4006,

which is exactly the inequality Z6<65 W6Z^6 < 65\,W^6Z6<65W6 needed for the surplus W6W^6W6 to dominate the error Z6−64W6Z^6 - 64W^6Z6−64W6. The margin is about one part in two thousand, which is why the constant in Eq. [1] has to be as large as it is.

Preamble
import Definitions.Def_BunkbedWZRefined
open Bunkbed Bunkbed.Hyper Finset
Formal statement
namespace BunkbedAux

/-- **The weights of the two blocks outside `Λ²₊`.** The configurations all of whose hyperedges
avoid `Λ²₊` have total weight `Z ^ 6` with `Z = 2W + 2 P_a|bc - P_a|bc ^ 2`, and those all of
whose hyperedges are in a `-` state have total weight `(2W) ^ 6`, where
`W = P_abc * P_a|b|c - P_ab|c * P_ac|b`. -/
theorem xc_and_A_weight (P : WZ → ℚ)
    (hsum : P .abc + P .a_b_c + P .a_bc + P .ab_c + P .ac_b = 1) :
    (∑ Ψ ∈ Finset.univ.filter (fun Ψ : Fin 6 → WZExt =>
        ∀ h, Ψ h ∈ Finset.univ.filter
          (fun q : WZExt => WZExt.legal q = true ∧ WZExt.inLamSq q = false)),
        ∏ h, WZExt.wt P (Ψ h))
      = (2 * (P .abc * P .a_b_c - P .ab_c * P .ac_b) + 2 * P .a_bc - P .a_bc * P .a_bc) ^ 6
    ∧ (∑ Ψ ∈ Finset.univ.filter (fun Ψ : Fin 6 → WZExt =>
        ∀ h, Ψ h ∈ ({Sum.inr 1, Sum.inr 3} : Finset WZExt)), ∏ h, WZExt.wt P (Ψ h))
      = (2 * (P .abc * P .a_b_c - P .ab_c * P .ac_b)) ^ 6 := by
  sorry

end BunkbedAux
Source
N. Gladkov, I. Pak, A. Zimin, The bunkbed conjecture is false, PNAS 122 (2025) no. 24, e2420725122, https://doi.org/10.1073/pnas.2420725122, Section 3.6 (p. 6): the normalizing constant Z and the event A.

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me