Lemma 3.3 — the robust hyperedge lemma
ProvedBunkbedFalse.robust_hyperedgeLet be Hollom's hypergraph with transversal set , and consider Wierman–Ziff percolation on the associated bunkbed hypergraph, in which each hyperedge independently takes one of five states with probabilities , , , , summing to . Assume the model's symmetry , the nondegeneracy condition , and the robustness hypothesis
Then
This is the robust hyperedge lemma and the technical heart of the refutation. Hollom's computation applies only to the exact hypergraph model; this lemma shows the strict inequality survives in the far more flexible WZ model, and therefore survives the inexact simulation of a hyperedge by a gadget graph. The source proves it by constructing a weight-preserving involution on a refined configuration space, built from a reflection involution and a half-reflection involution.
Formalization Note The hypothesis that the five probabilities are nonnegative and sum to is stated explicitly. The source notes (§8.2) that this lemma is a finite problem — positivity of degree- polynomials in five variables with at most nonzero coefficients on a region of the unit cube — and that it was independently verified by computer; a formal proof would replace that check with a certificate.
⚠️ Deviation from the printed statement. The published Lemma 3.3 does not include the nondegeneracy hypothesis , and is false without it. Take
These are nonnegative, sum to , satisfy the symmetry, and satisfy Eq. [1] as . But this is precisely the case in which each hyperedge degenerates into two independent ordinary edges – and –. Every edge of the resulting graph is then incident to a transversal vertex; the posts identify across the two levels; and and each hang off a single transversal vertex by one edge. Both probabilities therefore factor as and are exactly equal, so the strict conclusion fails. (At both equal .)
The source's own proof needs this hypothesis: it divides by , which vanishes on exactly this family, and it concludes strictly only when the conditioning event has positive probability. The hypothesis is satisfied in the intended application, since Lemma 4.1 supplies a strictly positive value.
import Definitions.Def_BunkbedHypergraph open Bunkbed Bunkbed.Hyper SimpleGraph Finset
namespace BunkbedFalse
/-- **Lemma 3.3** (robust hyperedge lemma, p. 3), with the nondegeneracy hypothesis `hpos`
added: the printed statement is false without it. -/
theorem robust_hyperedge (P : WZ → ℚ)
(hnn : ∀ s, 0 ≤ P s)
(hsum : P .abc + P .a_b_c + P .a_bc + P .ab_c + P .ac_b = 1)
(hsym : P .ab_c = P .ac_b)
(hpos : 0 < P .abc * P .a_b_c - P .ab_c * P .ac_b)
(h400 : 400 * P .a_bc ≤ P .abc * P .a_b_c - P .ab_c * P .ac_b) :
wzProb hollomTriple hollomT P ((0 : Fin 10), (0 : Fin 2)) ((9 : Fin 10), (0 : Fin 2))
< wzProb hollomTriple hollomT P ((0 : Fin 10), (0 : Fin 2)) ((9 : Fin 10), (1 : Fin 2)) := by
sorry
end BunkbedFalseRead-back
What the Lean code literally says, in plain math · claude-opus-5
Read-back — BunkbedFalse.robust_hyperedge
The objects the statement is built from
The statement lives on the vertex set (written Fin 10), doubled into two levels indexed by , so the ambient vertex set of every graph below is
twenty vertices in all. Level is the "lower" copy and level the "upper" copy.
The six ordered triples. The statement uses the fixed list hollomTriple, a function given by the ordered triples
The order inside each triple is load-bearing: writing , the first entry is distinguished from the other two. Note , , ; vertex occurs only in (as ) and vertex only in (as ).
The transversal set and the posts. hollomT is the three-element set , which is exactly . The posts are the three vertical edges
and they are present in every configuration, with probability — they are never randomised. They are the only edges joining level to level .
The five states. WZ is a five-element inductive type with constructors
For a triple and a level , writing , , , a state contributes the following edge set, all inside level :
| state | edges contributed |
|---|---|
| (a triangle) | |
| only | |
| only | |
| only | |
| none |
The random graph. A configuration is a function assigning a state to each of the pairs (hyperedge index, level): . The graph on has edge set
read as a simple graph (loops, were there any, are discarded).
The weight functional. For an arbitrary function (rationals, not reals — and with no positivity or normalisation built in), and for two vertices ,
a sum over all configurations. The twelve states are drawn independently: one per hyperedge per level, so the two levels are i.i.d. copies of each other, coupled only through the three always-present posts. (Connectedness is Reachable, which is reflexive; the two vertices compared below are distinct, so this is immaterial.)
What the theorem asserts
Abbreviate
For every function satisfying all five of the following hypotheses:
- (nonnegativity) for each of the five states, i.e. ;
- (normalisation) ;
- (symmetry of the two one-edge states through the distinguished vertex) , that is ;
- (strict positivity of a determinant) , i.e. ;
- (the factor-400 bound) , i.e. — note this is non-strict , while (4) is strict ;
the conclusion is the strict inequality
Both probabilities have the same source : vertex on level . The left-hand side is the same-level connection probability, from to vertex on level ; the right-hand side is the opposite-level connection probability, from to vertex on level . The claim is that the same-level side is strictly smaller — i.e. the target in the other level is claimed strictly more likely to be connected. Neither nor belongs to the transversal set , so and are genuinely distinct, unposted vertices.
Immediate consequences of the hypotheses, made explicit. Hypotheses (1)+(2) are precisely the conditions under which is a probability (the definition carries no internal normalisation; the total mass is , here ). Hypothesis (4) together with (3) reads , which forces and : neither nor may vanish. Hypotheses (3)+(4) force . Hypothesis (5) forces : the state (the one joining the two non-distinguished vertices of a triple) is squeezed to essentially zero. Hypothesis (5) permits , and (1)+(3) permit .
Note on the artifact. The declaration's proof is sorry — the statement is asserted, not proved, in this file.
Testimony on satisfiability (question (a)): the hypothesis set is SATISFIABLE
The hypotheses are not vacuous. A concrete witness meeting every one of the five simultaneously, verified exactly:
Check: all five values ; sum ; so (3) holds; so (4) holds; so (5) holds.
A second witness with all five states strictly positive and (5) tight ( exactly):
for which and exactly. The feasible region is in fact a full-dimensional open-ish set: it is the set of with , , , , .
Testimony on truth (question (b)): NO COUNTEREXAMPLE FOUND — the conclusion held on every feasible point tested
I reimplemented the model from the two definition files in Python and computed the two probabilities exactly over . The reimplementation exploits the fact that the only level-to-level edges are the three posts at , so each level's state-vectors need only be reduced to the partition it induces on the five terminals (47 distinct partitions arise); the two levels are then glued through the posts. This reduction was cross-validated against brute-force union-find on the full 20-vertex graph for 20,000 random configurations: 0 mismatches. Total level mass came out to exactly on a test distribution, and the level-weight polynomial summed to exactly .
Writing (the conclusion says ), I obtained as an exact integer polynomial, homogeneous of degree , with 231 monomials in — and it satisfies the identity
for an integer polynomial of degree with mixed signs. On the slice it collapses to exactly .
Search performed. Exhaustive-in-practice sweeps of the feasible region, parametrised so that hypotheses (1)–(3) and (5) hold exactly in rational arithmetic (for fixed rational and , the level of (5)-tightness gives the rational with , and ):
- points in double precision over , the whole admissible -interval, and ;
- points re-evaluated in exact rational arithmetic, including a dedicated fine sweep of the region and of the regime (down to , where ), plus 15,000 randomised exact samples.
Result: no hypothesis-satisfying assignment was found on which the two probabilities are equal or reversed. Across all exact points, and the ratio stayed in the narrow band ; the global exact minimum of encountered was (at a near-degenerate point), always strictly positive. The double-precision sweep did flag 706 apparently-negative points, but every one had against — pure catastrophic cancellation — and exact rational re-evaluation of that entire neighbourhood returned at all 7,080 exact points, with minimum ratio . So there is no critical finding to report: I found no violation.
Exact values at the witnesses.
- At , :
so : the conclusion holds, with the opposite-level probability larger by exactly one part in .
- At the all-five-positive, (5)-tight witness above: , , .
- At , , , (hypothesis (5) tight, ): , , .
Boundary behaviour of the numerical hypotheses
- Hypothesis (4) is exactly tight. At , — which satisfies (1), (2), (3) and (5) but makes , so (4) fails by the barest possible margin — the two probabilities are exactly equal:
Thus the strict inequality in (4) cannot be weakened to without making the strict conclusion false; and generally as (4) degenerates, so the conclusion's margin becomes arbitrarily small while remaining positive.
- Hypothesis (5) is essential, not cosmetic. Dropping only (5) while keeping (1)–(4), a random sample of 40,000 normalised assignments produced 17,256 with (worst observed ), e.g. , , , . So the factor is doing real work: without it the stated inequality reverses on a large part of the simplex.
- Zero-probability states. and are both permitted and occur in the witnesses above. or is impossible under (4). No state may have probability .
- Range of . Feasibility forces ; as the admissible -window shrinks to the single point and .
Confirmed by the mission captain (proposal self-audit).