Low-discrepancy quasi-uniform digital sequences in every dimension

Get email updates

Leonhard Gruenschloss

For every dimension s≥2s \ge 2 and prime p>2sp > 2^{s}, a specified linear scramble of a generalized Faure digital sequence has bounded Euclidean mesh ratio across all prefixes. The paper also rules out quasi-uniformity for the stated Halton-type family with pairwise coprime nonconstant polynomial bases over FpF_{p} in dimensions at least two.

What was already known

Dick, Goda and Suzuki posed the digital-sequence question. Goda, Hofer and Suzuki excluded ordinary Halton sequences and some Halton-type sequences. Gruenschloss’s finite digital nets address finite sets.

AI involvement AI-led

Claude Opus 5.5 produced constructions, proofs and Lean code. Leonhard Gruenschloss selected problems, steered the research and signed off the four Palomar statements against the manuscript.

Importance Notable

One explicit sequence in each stated dimension and base works at all sample sizes, giving a new construction for a documented open question.

Technical verification Formal (Lean)

Lean and NanoDa accepted all 80 Comparator configurations in the report, including four selected sequence theorems. Author checked statement correspondence; that signoff does not cover every proof.

Formal checking

Lean and NanoDa accepted all 80 Comparator configurations in the report, including four selected sequence theorems. Author checked statement correspondence; that signoff does not cover every proof.

Checker
Lean kernel and NanoDa Comparator reported by Gruenschloss
Scope
Four Palomar statements, including existence and the all-prefix Euclidean mesh ratio. The explicit toroidal separation constant applies at N=p(sK−1)N=p^{(sK-1)}. The discrepancy implication is classical and outside the selected Lean scope.
Basis
Reported by the linked checker

Human understanding Partial

Gruenschloss reports understanding some parts. He reviewed the four formal statements and their correspondence, and reports that he has not checked every proof by hand. The disclosure does not allocate a fuller human explanation of the residue construction.

Correction and publication history