Disproving the quasi-uniformity of the Halton sequences and some of Halton-type sequences
Takashi Goda, Roswitha Hofer, Kosuke Suzuki
TL;DR
The paper proves that Halton sequences in dimensions $d\ge 2$ with pairwise coprime bases are not quasi-uniform by establishing that the separation radius $q(P_N)$ can decay faster than the optimal $N^{-1/d}$ along infinitely many $N$, specifically $q(P_N) \lesssim N^{-1/d} (\log N)^{-1/(d(d-1))}$. It introduces a constructive number-theoretic framework to generate close pairs $(n,m)$ across all bases, using a divisibility lemma for $p\equiv 1 \pmod q$ and Dirichlet simultaneous approximation to balance digits across bases. The work extends to Halton-type sequences over polynomial bases in $\mathbb{F}_p[X]$, providing analogous non-quasi-uniformity results in several concrete cases (including Faure-related constructions) and a general covering-radius bound for these digital constructions via a polynomial-analytic approach. These findings clarify the geometric limitations of Halton and Halton-type sequences beyond their discrepancy properties, with implications for their use in numerical integration and scattered-data approximation.
Abstract
In this short article, we prove that the Halton sequence, one of the most well-known low-discrepancy sequences, is not quasi-uniform in any dimension $d \ge 2$ with any pairwise relatively prime bases. We further disprove the quasi-uniformity of some Halton-type sequences, including the Faure sequence, which provides an alternative proof of the known results.
