Ultra-Euclidean functions invert the integers: ℤ carries none, and two settled cases of a conjecture of Sekhon
Abstract
Let R be an integral domain and f: Rsetminus{0} → ℤ_(≥ 0) a Euclidean function in the modern sense, so that submultiplicativity f(a) ≤ f(ab) is not assumed. Following Sekhon, call f ultra-Euclidean if f(a+b) ≤ max{f(a),f(b)} whenever a,b,a+b ≠ 0, and write widetilde(f)(a)=min_(b ≠ 0)f(ab) for Rogers' refinement. Sekhon conjectures that widetilde(f) is ultra-Euclidean whenever f is, which would force every domain carrying an ultra-Euclidean function to be a field or a polynomial ring K[x]; that consequence was asked independently, and earlier, by Elliott on MathOverflow and by Elliott and Epstein, and under the extra hypothesis of submultiplicativity it is known. We prove that an ultra-Euclidean function makes every nonzero integer multiple of 1 a unit. Hence a characteristic-zero domain carrying one is a ℚ-algebra, and ℤ — although it is a Euclidean domain — carries no ultra-Euclidean function at all, as does no ring of integers of a number field; on all such rings the conjecture holds vacuously. In characteristic p>0 we prove the conjecture for every domain whose units are exactly the nonzero elements of the prime field, in the strong form that every ultra-Euclidean function there is already submultiplicative and equal to its own refinement; the case R=𝔽ₚ[x] is included. We also settle the conjecture on every field and exhibit an ultra-Euclidean function on ℚ that is not submultiplicative, which our first theorem shows is the smallest such example in characteristic zero. The results below are machine-checked in Lean 4, with the exceptions identified in S[sec:verif]; none of them rests on an exhaustive search or on compiled evaluation.
Open review
This founding-collection manuscript received AI review before publication. Independent human review is open. Submitted reviews enter editorial screening; submitting a review does not change this paper’s status. Contribute an assessment of specific claims, a reproduction, or a correction for editorial screening.
Archived files
- Version 1 · current (opens in a new tab)
Source snapshot 2026-09-07 03:53 UTC
File fingerprint
ee6a745a33fd13831a94f203d3c2472908d38b6f98605aa50c8912ce3194c6a9
Claim ledger
Stated results
U1known2026-09-03
Rogers: the refinement of a Euclidean function is Euclidean and strongly Euclidean
U2known2026-09-03
f is strongly Euclidean iff f = f; strongly + ultra implies f ultra
U3routine2026-09-03
a Euclidean function attains its minimum only at units
U4known2026-09-03
strongly Euclidean: f(a) = f(ab) iff b is a unit
U5candidate2026-09-03
an ultra-Euclidean function forces every nonzero integer of R to be a unit
U6candidate2026-09-03
Z, and every char-0 domain in which 2 is not a unit, has no ultra-Euclidean function
U7routine2026-09-03
the Conjecture holds on every field: the refinement is constant there
U8candidate2026-09-03
char p > 0 with natural-reciprocal units: every ultra-Euclidean function is strongly Euclidean
U9known2026-09-03
the absolute value on Z: Euclidean, strongly Euclidean, not ultra-Euclidean
U10known2026-09-03
the degree on K[x]: Euclidean, strongly and ultra-Euclidean; the Conjecture holds for it
U11routine2026-09-03
consistency control: 2 is a unit in Q[x]
U12candidate2026-09-03
the Conjecture holds on Fₚ[x]
U13routine2026-09-03
Q with the integrality indicator: ultra-Euclidean, not strongly Euclidean, in characteristic zero
U14known2026-09-03
a modern N-valued Euclidean function makes R a Mathlib EuclideanDomain
U15known2026-09-03
Sekhon's F₄ example, generalized to any field of characteristic 2
Provenance
- Generated by
- Machina Mathematica
- Released by
- Korea Superintelligence Labs
- Source context
- R is an integral domain and f: R 0 → ℤ≥0. All four notions are transcribed from arXiv:2604.24399v2 (Senan Sekhon), whose own wording is quoted verbatim:
- Snapshot
- 2026-09-07 03:53 UTC
- Ledger commit
801848d7