Tournament score sequences, Erdős-Ginzburg-Ziv numbers, and the Lévy-Khintchine method
Michal Bassan, Serte Donderwinkel, Brett Kolesnik
TL;DR
The paper develops a concise probabilistic approach to the asymptotic enumeration of score sequences $S_n$ in tournaments, revealing a deep link to Erdős–Ginzburg–Ziv numbers $N_n$ via the Lévy–Khintchine method. By leveraging the lattice-path representation of score sequences and Kleitman’s cyclic-shift ideas, the authors derive a simple proof that $S_n^*=N_n$ and use renewal-sequence theory to connect $S_n$ to $N_n$, ultimately confirming Moser's conjecture that $S_n\sim C 4^n/n^{5/2}$ with $C$ expressed in terms of $N_n$. The main innovations are the geometric diamond-area interpretation of bridges, the renewal-structure framework for analyzing cyclic shifts, and a clean bijection that ties the two counting problems together. The results illustrate the power of the Lévy–Khintchine paradigm in combinatorial asymptotics and suggest broader applicability to renewal-like structures in discrete geometry and permutation-based polytopes.
Abstract
We give a short proof of a recent result of Claesson, Dukes, Franklín and Stefánsson, connecting the number $S_n$ of score sequences and the Erdős-Ginzburg-Ziv numbers $N_n$ from additive number theory. Our proof utilizes the lattice path representation of score sequences by Erdős and Moser, and remarks by Kleitman added to an article of Moser regarding cyclic shifts of such paths. The connection between $S_n$ and $N_n$ is an instance of the Lévy-Khintchine formula from probability theory. We highlight the utility of such formulas, by giving a short proof of Moser's conjecture that $S_n\sim C4^n/n^{5/2}$, where $C$ is described in terms of $N_n$.
