[Paper Review] A Solution to the Lonely Runner Conjecture for Almost All Points

C. Harold Horvat, Matthew Stoffregen|arXiv (Cornell University)|Mar 8, 2011

Mathematics and Applications4 citations

TL;DR

This paper proves the Lonely Runner Conjecture for almost all velocity vectors by leveraging ergodic theory and Diophantine approximation, showing that for linearly independent velocities, a time exists when one runner is at least 1/(k+1) away from all others. The set of exceptions has Lebesgue measure zero, establishing the conjecture holds for 'almost all' configurations.

ABSTRACT

The Lonely Runner Conjecture is a number theory problem, dating to 1964. Using dynamical systems theory, we show almost all sets of velocities solve the conjecture. Furthermore, any "traditional" approach of Diophantine approximation cannot solve the problem, and we offer a short list of reformulations of the problem.

Motivation & Objective

• To resolve the Lonely Runner Conjecture for the vast majority of velocity configurations on a circular track.
• To demonstrate that for linearly independent velocities, there exists a time when one runner is at least 1/(k+1) from all others.
• To show that the set of velocity vectors violating the conjecture has Lebesgue measure zero.
• To analyze the structure of rational-dependent velocity vectors and their measure-theoretic rarity.
• To provide a constructive framework for identifying 'good' velocity ratios near optimal configurations.

Proposed method

• Use of irrational rotation maps on the circle to model runner trajectories, ensuring dense orbits under linear independence.
• Application of the product of irrational rotations to achieve density on the torus T^k, guaranteeing approximation of any configuration.
• Employment of the Chinese Remainder Theorem to construct a time T where all runners are simultaneously at least 1/(n+1) from the start.
• Definition of a 'quality' function Q(D_i, D_{i+1}) to bound allowable velocity ratio deviations while preserving the lonely runner condition.
• Use of volume integration over rational approximations to estimate the measure of good velocity vectors.
• Backward construction via continuous maps F1 from rational to irrational points to locate exact or pseudo-exact solutions.

Experimental results

Research questions

• RQ1 Does the Lonely Runner Conjecture hold for all but a set of measure-zero velocity vectors?
• RQ2 Can a time be found such that one runner is at least 1/(k+1) from all others when velocities are linearly independent?
• RQ3 What is the measure of the set of velocity vectors for which the conjecture fails?
• RQ4 How can rational approximations to optimal runner configurations be used to construct good velocity ratios?
• RQ5 Can the conjecture be extended to rational points by continuity and density arguments?

Key findings

• The Lonely Runner Conjecture holds for all velocity vectors whose components are linearly independent over Q.
• The set of velocity vectors violating the conjecture has Lebesgue measure zero, as it lies within a countable union of lower-dimensional hyperplanes.
• For any coprime integer-based velocity set (1/D1, ..., 1/Dn), a time T < ∏Di exists such that each runner is at least 1/(n+1) from the start.
• A perturbation bound δvi is derived such that small changes to good velocity vectors preserve the lonely runner condition.
• The volume of good velocity ratios is bounded below by a sum over rational approximations, with intervals defined by δvi and Di.
• The existence of a point B on the line from a rational failure point to an irrational point with F1(B) = 1/(n+1) implies that no rational point can be a pseudo-exact failure case.