Deletions, replacements and rescaled blocks in the lonely runner problem: Lean-verified results and an exact census

Preprint with Lean 4 formalization, programs and census data. For a finite set V of positive integers let LR(V) = max_t min_{v in V} ||vt||. We study sets obtained from {1,...,n-1} by deleting k speeds and inserting k speeds >= n, and ask when some time puts every runner at distance strictly greater than 1/n from the origin. Answering questions of Zhang (arXiv:2608.13599v3) we show: LR([n-1] \ {r}) > 1/(n-1) for all n >= 5 and 1 <= r <= n-2; for 8 <= n <= 18 the only set ([n-1] \ {2,3}) ∪ {u,v} without a strict witness is {1,4,5,6,7,11,13}, with no upper bound on u, v; whenever 1 is deleted and n >= 4k a strict witness exists (for k = 2 already when n >= 6, which is sharp); for two arbitrary deletions and n >= 8 every configuration without a strict witness has largest speed at most 6(n-1)^3; for each fixed k the existence of a strict witness is decided by inserted speeds below an explicit polynomial in n; and tight k-replacements with arbitrarily large speeds exist exactly when a set with fewer inserted speeds is tight at the same level. We also give polynomial bounds for three replacements (n prime, or all deletions above n/2) and a linear bound when all deleted speeds are units modulo n. For arbitrary finite speed sets we give a phase certificate for LR(S) >= d, and for two sets A, B we prove min(LR(A) - max(A)/(2M), LR(B)) <= LR(A ∪ MB) <= min(LR(A), LR(B)), with equality LR(B) for large M when LR(B) < LR(A). The same method gives an exact criterion for extra forbidden differences in an integer packing problem and the exact optimal radius for two-channel weighted phase fitting. All theorems are checked in Lean 4.24.0 with Mathlib (no sorry; axioms propext, Classical.choice, Quot.sound only). Separately, an exact-arithmetic census with unbounded inserted speeds finds that for 8 <= n <= 140 the only two-entry replacements without a strict witness are the two known tight sets, that every three-entry replacement for 8 <= n <= 50 has a strict witness, and that the only four-entry exception for 9 <= n <= 24 is {2,4,...,16}; the census is not certified in Lean. The proofs, formalization and programs were developed with substantial assistance from AI coding assistants.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-19
DOI
https://doi.org/10.5281/zenodo.24363089
Primary Topic
Analytic Number Theory Research
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Deletions, replacements and rescaled blocks in the lonely runner problem: Lean-verified results and an exact census

Jason Peng
Zenodo (CERN European Organization for Nuclear Research)
Analytic Number Theory Research
preprint

Deletions, replacements and rescaled blocks in the lonely runner problem: Lean-verified results and an exact census

Jason Peng
preprint en

Abstract

Preprint with Lean 4 formalization, programs and census data. For a finite set V of positive integers let LR(V) = max_t min_{v in V} ||vt||. We study sets obtained from {1,...,n-1} by deleting k speeds and inserting k speeds >= n, and ask when some time puts every runner at distance strictly greater than 1/n from the origin. Answering questions of Zhang (arXiv:2608.13599v3) we show: LR([n-1] \ {r}) > 1/(n-1) for all n >= 5 and 1 <= r <= n-2; for 8 <= n <= 18 the only set ([n-1] \ {2,3}) ∪ {u,v} without a strict witness is {1,4,5,6,7,11,13}, with no upper bound on u, v; whenever 1 is deleted and n >= 4k a strict witness exists (for k = 2 already when n >= 6, which is sharp); for two arbitrary deletions and n >= 8 every configuration without a strict witness has largest speed at most 6(n-1)^3; for each fixed k the existence of a strict witness is decided by inserted speeds below an explicit polynomial in n; and tight k-replacements with arbitrarily large speeds exist exactly when a set with fewer inserted speeds is tight at the same level. We also give polynomial bounds for three replacements (n prime, or all deletions above n/2) and a linear bound when all deleted speeds are units modulo n. For arbitrary finite speed sets we give a phase certificate for LR(S) >= d, and for two sets A, B we prove min(LR(A) - max(A)/(2M), LR(B)) <= LR(A ∪ MB) <= min(LR(A), LR(B)), with equality LR(B) for large M when LR(B) < LR(A). The same method gives an exact criterion for extra forbidden differences in an integer packing problem and the exact optimal radius for two-channel weighted phase fitting. All theorems are checked in Lean 4.24.0 with Mathlib (no sorry; axioms propext, Classical.choice, Quot.sound only). Separately, an exact-arithmetic census with unbounded inserted speeds finds that for 8 <= n <= 140 the only two-entry replacements without a strict witness are the two known tight sets, that every three-entry replacement for 8 <= n <= 50 has a strict witness, and that the only four-entry exception for 9 <= n <= 24 is {2,4,...,16}; the census is not certified in Lean. The proofs, formalization and programs were developed with substantial assistance from AI coding assistants.

Zenodo (CERN European Organization for Nuclear Research)
Analytic Number Theory Research
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