Beyond Gold and Silver: Half-Factorial Real Quadratic Orders, Independent of GRH
For a prime p, the position of the first positive Fibonacci number divisible by p is called the rank of apparition of p, and the same is defined for the Pell numbers. This paper is about how often that rank is as large as it can possibly be, which is p + 1 or p − 1 depending on p (apart from p = 5 for the Fibonacci numbers and p = 2 for the Pell numbers). Under the generalized Riemann hypothesis (GRH), a famous unproved conjecture, the share of primes for which this happens was known to exist and to be positive. The paper shows, without assuming GRH, that the share stays above a fixed positive amount. It does so by carrying over a method of Bhargava, and it depends on three results from manuscripts of September and October 2026 that have not yet been peer reviewed. The same is shown for the analogous rank in every real quadratic field whose fundamental unit has norm −1. The golden ratio, behind the Fibonacci numbers, and the silver ratio, behind the Pell numbers, are the two best-known such units. Given the same results, a criterion of Coykendall then shows that every such field of class number at most two contains infinitely many half-factorial orders. An order here is a ring of numbers such as a + bp√2, with a and b whole numbers and p a fixed prime, and it is half-factorial when any two factorizations of an element into irreducible elements have the same number of factors. The two conditions on the field are known to be necessary, so the same results determine exactly which real quadratic fields contain infinitely many half-factorial orders. For ℚ(√2) this is Coykendall's Conjecture B, which Pollack (2024) proved under GRH. The abstract of the paper follows.Let K be a real quadratic field whose fundamental unit μ has norm −1, and let n(p) be the order of γ = −μ2 modulo a prime p; it divides p + 1 if p is inert in K and p − 1 if p splits. For K = ℚ(√5) and K = ℚ(√2), n(p) is the rank of apparition of p in the Fibonacci and the Pell sequence. Under the generalized Riemann hypothesis (GRH), n(p) takes its largest value for a positive density of primes (Roskam; Chen). The main result is that, without GRH, n(p) = p + 1 for a positive lower density of inert primes and n(p) = p − 1 for a positive lower density of split primes. It is conditional on three results of manuscripts of September and October 2026 that have not yet been peer reviewed. These are the zero-free half-plane of Cureton and Thorner for Dedekind zeta functions and the two counts of candidate primes of Bhargava, which rest on Type II estimates in manuscripts of OpenAI. The method is Bhargava's, carried from an integer base to the unit. The Kummer fields of γ, the fields of the proofs under GRH, are Galois over ℚ with a central element of order two, so the inert case needs only Dedekind zeta functions of Galois fields, as Bhargava's splitting bound requires. Given those results, a criterion of Coykendall then shows that every real quadratic field of class number at most two whose fundamental unit has norm −1 contains infinitely many half-factorial orders. These conditions are known to be necessary, so the same results determine the real quadratic fields with infinitely many half-factorial orders. For ℚ(√2) the statement is Coykendall's Conjecture B, which Pollack (2024) proves under GRH. Without GRH, Pollack (2024) proves the statement for all but finitely many of these fields, without identifying any one of them.In the interest of full disclosure: Fable 5.2 was used as a tool to search within literature, run computations, and help draft and check the arguments. Two of the lemmas were tested in exact arithmetic on eleven fields, with no exception found. I directed the work and take full responsibility for its content.
Authors
- A. Down (ORCID: https://orcid.org/0009-0004-8555-4797)
Publication Details
- Journal
- Figshare
- Published
- 2026-11-14
- DOI
- https://doi.org/10.6084/m9.figshare.35009296
- Primary Topic
- Analytic Number Theory Research
- Type
- preprint