Affine lifts of cyclic Legendre pairs

A normalized cyclic Legendre pair is an ordered pair of ±1 sequences, each of sum one, whose periodic autocorrelations add to −2 at every nonzero shift. We ask when one complex affine map sends every normalized pair of odd length n > 1 to a normalized pair of length N and commutes with shifts by the subgroup of order n in the target cyclic group. Suppose the source family is nonempty and satisfies coordinate rigidity and character separation. Then the index N/n is congruent to 1 modulo n + 1. For n ≥ 5, a lift at the first allowed proper index n + 2 exists exactly when there is a cyclic affine difference set of order n + 1. Up to independent translations and exchange of the outputs, each output copies its corresponding input along such a set. For primes p ≡ 1 (mod 4), there is no lift from p to p(p + 2). For odd primes p, a lift from p to 9p exists exactly when p is 3 or 7. Among the odd lengths from 3 through 69 and at 75, a lift from n to n(n + 2) exists exactly for n = 3, 7, 15, 31 and 63. At fixed length with the same shift action, each output of an equivariant affine self-map of the full nonempty family is a fixed translate of a fixed input. All ten numbered results have proofs in Lean 4 with Mathlib, including both directions of the difference-set classification and the counterexample at length seven. The claim map links each result to its formal statement and proof. The replay compiles the 87 modules in the archive from source, audits their axiom dependencies and checks the declarations in a fresh instance of the Lean 4.33.1 kernel. The consequence for shifts twisted by a unit follows by reindexing in the text. This record holds the article and its source archive, supplement.zip. The archive contains the LaTeX source and bibliography, the Lean sources and replay instructions, and 38 witness pairs at 35 lengths. A table identifies the witnesses used at each length. A separate verifier runs on Python 3.11 or later using only the standard library. It checks both row sums and every nonzero shift for each pair, the modular and character tests behind the finite classification, and the example at length 255. The article, its LaTeX text, the documentation, the coverage table and the separate length-27 witness are licensed under CC BY 4.0. The original Lean and Python code and reusable TeX files use the MIT License. The 29 modules from the Legendre Pairs miner retain their Apache 2.0 license, and the 38 catalogue pairs retain CC0 1.0. LICENSES.md gives the terms for each part, including the fonts embedded in the PDF. Lean and Mathlib are obtained separately. Use of generative AI. Generative AI tools, including ChatGPT, Codex and Claude, assisted with the mathematics, proofs, programs and text. The author is responsible for all content.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-18
DOI
https://doi.org/10.5281/zenodo.24329584
Primary Topic
Coding theory and cryptography
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Affine lifts of cyclic Legendre pairs

Benedek Blackthorn
7 links
Zenodo (CERN European Organization for Nuclear Research)
Coding theory and cryptography
preprint

Affine lifts of cyclic Legendre pairs

Benedek Blackthorn
preprint en

Abstract

A normalized cyclic Legendre pair is an ordered pair of ±1 sequences, each of sum one, whose periodic autocorrelations add to −2 at every nonzero shift. We ask when one complex affine map sends every normalized pair of odd length n > 1 to a normalized pair of length N and commutes with shifts by the subgroup of order n in the target cyclic group. Suppose the source family is nonempty and satisfies coordinate rigidity and character separation. Then the index N/n is congruent to 1 modulo n + 1. For n ≥ 5, a lift at the first allowed proper index n + 2 exists exactly when there is a cyclic affine difference set of order n + 1. Up to independent translations and exchange of the outputs, each output copies its corresponding input along such a set. For primes p ≡ 1 (mod 4), there is no lift from p to p(p + 2). For odd primes p, a lift from p to 9p exists exactly when p is 3 or 7. Among the odd lengths from 3 through 69 and at 75, a lift from n to n(n + 2) exists exactly for n = 3, 7, 15, 31 and 63. At fixed length with the same shift action, each output of an equivariant affine self-map of the full nonempty family is a fixed translate of a fixed input. All ten numbered results have proofs in Lean 4 with Mathlib, including both directions of the difference-set classification and the counterexample at length seven. The claim map links each result to its formal statement and proof. The replay compiles the 87 modules in the archive from source, audits their axiom dependencies and checks the declarations in a fresh instance of the Lean 4.33.1 kernel. The consequence for shifts twisted by a unit follows by reindexing in the text. This record holds the article and its source archive, supplement.zip. The archive contains the LaTeX source and bibliography, the Lean sources and replay instructions, and 38 witness pairs at 35 lengths. A table identifies the witnesses used at each length. A separate verifier runs on Python 3.11 or later using only the standard library. It checks both row sums and every nonzero shift for each pair, the modular and character tests behind the finite classification, and the example at length 255. The article, its LaTeX text, the documentation, the coverage table and the separate length-27 witness are licensed under CC BY 4.0. The original Lean and Python code and reusable TeX files use the MIT License. The 29 modules from the Legendre Pairs miner retain their Apache 2.0 license, and the 38 catalogue pairs retain CC0 1.0. LICENSES.md gives the terms for each part, including the fonts embedded in the PDF. Lean and Mathlib are obtained separately. Use of generative AI. Generative AI tools, including ChatGPT, Codex and Claude, assisted with the mathematics, proofs, programs and text. The author is responsible for all content.

Zenodo (CERN European Organization for Nuclear Research)
Coding theory and cryptography
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon