Keyboard shortcuts

Press ← or → to navigate between chapters

Press ? to show this help

Press Esc to hide this help


slug: chervov-2025-cayleypy-conjecture-one-refutation bibkey: chervov2025cayleypy doi: 10.48550/arXiv.2509.19162 url: https://arxiv.org/abs/2509.19162v2 triage: theorem motivation_gids:

  • D5/S0/CayleyGrowth/CayleyPyConjectureOneRefutation.cayleyPy_conjecture1_refuted

CayleyPy Growth Conjecture 1 is false

Problem

Chervov et al., CayleyPy Growth (arXiv:2509.19162v2), Section 3.2, conjecture that every generator family of S_n output by an algorithm of polynomial complexity in n has an eventually linear or quadratic quasipolynomial Cayley-graph diameter.

Motivation

The conjecture quantifies over every polynomial-time output family, so a single explicit family with non-quasipolynomial diameter settles its literal statement. The earlier repository result treats Conjecture 2’s marked-element distance, which does not imply a diameter counterexample.

Gap

The bounded literature check read arXiv:2509.19162v2, Section 3.2; the arXiv API still returned v2 on 2026-09-29. A public-source search of both arXiv versions, title and identifier queries, Conjecture 1 with diameter and transposition terms, CayleyPy-4, and accessible citing discussions found no published resolution in that scope. The Semantic Scholar citation query returned HTTP 429, so it supplied no citation evidence. This does not establish exhaustive worldwide priority or exclude an independent proof.

Route

Let T_n be all transpositions of S_n. At nonsquare sizes use T_n; at square sizes use every product of at most three elements of T_n. The second set can be output by enumerating lists of zero, one, two, or three pairs of indices and composing their swaps. There are at most 1 + C(n,2) + C(n,2)^2 + C(n,2)^3 such lists, and each output permutation has length n. The square test, list enumeration, permutation composition, and duplicate removal all take polynomial time in n. If generators are required to exclude the identity, omit it from the output; the Cayley graph is unchanged. Thus this family meets the paper’s explicit-output condition.

For the nonsquare graph, a full-support rotation moves all n points, while each transposition moves at most two. Every path to that rotation has length at least n/2, so the diameter is at least n/2. For the square graph, Mathlib’s constructive factorization writes each permutation as at most n transpositions. Grouping that list into blocks of at most three gives diameter at most ceil(n/3).

Suppose the diameter were eventually quasipolynomial with period p. Squares of sufficiently large multiples of p and the numbers obtained by adding p are in the same residue class, with the latter lying strictly between consecutive squares. The constituent polynomial is below 5n/12 on the square subsequence and above 5n/12 on the nonsquare subsequence. Its difference from the linear polynomial 5X/12 would then have both signs arbitrarily far out, impossible for a rational polynomial.

Falsifier

The refutation would fail if the paper restricted its quantified generator families beyond polynomial-time explicit output, if the square-size family could not be output in polynomial time, or if one residue-class polynomial could maintain the two eventual diameter inequalities. The source wording, enumeration count, and Lean bounds address these three points.

Evidence

The canonical Lean source is D5/S0/CayleyGrowth/CayleyPyConjectureOneRefutation.lean. It proves the transposition lower bound, the triple-product upper bound, the polynomial sign obstruction, and CayleyGrowth.cayleyPy_conjecture1_refuted. The polynomial-time output count is a source-level complexity argument, outside the Lean statement.

The frozen module statement identity is sha256:42d3e90e37b3c7b79ca0e957e1c2c87f8e4046d897fe13071318fbc9364de61f. The Freeze event is sha256:73539ed8251121e370dae79647338a6cfbbb0882a1df725d339f9b30f0c100b8. The public declarations are general diameter and quasipolynomial statements, not bounded enumeration, a checker, a numeric reduction, or a certified finite instance. Their utility classification is none.

The Section 3.9 source-bound escape audit is unfinished. Issue #11179 records the current registration obstacle and missing four-slot evidence for the three public theorems; it does not certify a registration.

Triage

Tier 1 external named conjecture. Resolution: refuted. The new general diameter and polynomial-sign arguments provide the escape-witness admission basis; the utility kind is none.

ASSUMED-UNVERIFIED

The result addresses the conjecture’s unrestricted polynomial-output class. It does not settle any restricted natural-generator version.