The number system in rational base 3 / 2 3/2 and the 3 x + 1 3x+1 problem
Shalom Eliahou and Jean-Louis Verger-Gaugry · 2025
Read The number system in rational base 3 / 2 3/2 and the 3 x + 1 3x+1 problem by Shalom Eliahou and Jean-Louis Verger-Gaugry (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-13716.txt.
Introduction
Shalom Eliahou and Jean-Louis Verger-Gaugry link writing numbers in the fractional base three-halves to the famous Collatz problem, in a 2025 note. Rational base systems, introduced in 2008, write numbers using a fraction as the base. The three-halves case drew special focus from the start.
The authors show that natural questions about digits in that base turn out to involve the Collatz three-x-plus-one function in a deep way. That function is simple to state but its long-term behavior is unproven. The note lays out the links in plain form. Its stated aim is to motivate more research into the connection.
Catalog record
| CREATOR | Shalom Eliahou and Jean-Louis Verger-Gaugry |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 18 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-13716
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
The number system in rational base 3 / 2 3/2 and the 3 x + 1 3x+1 problem Shalom Eliahou and Jean-Louis Verger-Gaugry Abstract . The representation of numbers in rational base p / q p/q was introduced in 2008 by Akiyama, Frougny and Sakarovitch, with a special focus on the case p / q = 3 / 2 p/q=3/2 . Unnoticed since then, natural questions related to representations in that specific base turn out to intimately involve the Collatz 3 x + 1 3x+1 function. Our purpose in this note is to expose these links and motivate further research into them. 2020 Mathematics Subject Classification 11B83, 68R15 1. Introduction The representation of numbers in rational base p/q with p,q coprime positive integers was introduced in 2008 in [ 1 ] , with a focus on the case p/q=3/2 to study Mahler’s Z -number problem and the Josephus problem. Given n\in\mathbb{N} , its representation in rational base 3/2 is the unique expression of the form n=\sum_{i=0}^{k}\frac{a_{i}}{2}\left(\frac{3}{2}\right)^{i} with digits a_{i}\in\{0,1,2\} for all i\geq 0 and a_{k}\neq 0 if n\geq 1 . We then denote the corresponding word in \{0,1,2\}^{*} by \langle n\rangle=a_{k}a_{k-1}\cdots a_{0}. We set \langle 0\rangle=\epsilon , the empty word, and we identify any word 0w\in\{0,1,2\}^{*} with w . Denoting \mathcal{L}=\{\langle n\rangle\mid n\in\mathbb{N}\}\subset\{0,1,2\}^{*} , here are two natural questions about \mathcal{L} : (1) If w\in\mathcal{L} , which extensions among w0,w1,w2 still belong to \mathcal{L} ? (2) Given k\geq 1 , what is the lexicographically largest word of length k in \mathcal{L} ? Equivalently, what is the largest n\in\mathbb{N} such that \langle n\rangle is of length k ? Apparently unnoticed so far, answers to these questions turn out to intimately involve the Collatz 3x+1 function, namely T:\mathbb{N}\to\mathbb{N} defined by (1) T(n)=\begin{cases}n/2&\textrm{ if }n\equiv 0\bmod 2,\\ (3n+1)/2&\textrm{ if }n\equiv 1\bmod 2.\end{cases} The purpose of this note is to expose these links and motivate further research into them. Its contents are as follows. In Section 2 , we recall basic properties of the numeration system in rational base 3/2 , in particular the recursive construction of the word \langle n\rangle in Section 2.1 and the odometer allowing to construct \langle n+1\rangle from \langle n\rangle in Section 2.2 . We also propose a variant of the Collatz conjecture in Section 2.3 . In Sections 3.1 and 3.2 , we address the above questions (1) and (2), respectively. Section 4 concludes the note with two questions and two conjectures closely related to the Collatz conjecture and possibly of the same order of difficulty. 2. Basics Following [ 1 ] , every n\in\mathbb{N} has a unique expression of the form n=\frac{1}{2}\big(a_{k}(3/2)^{k}+a_{k-1}(3/2)^{k-1}+\dots+a_{0}\big) with a_{i}\in\{0,1,2\} for all i and a_{k}\neq 0 if n\geq 1 . Accordingly, we associate to n the word \langle n\rangle=a_{k}a_{k-1}\cdots a_{0}\in\{0,1,2\}^{*}, called the representation of n in rational base 3/2 . We set \langle 0\rangle=\epsilon , the empty word, and we identify any word 0w\in\{0,1,2\}^{*} with w . By Lemma 4.1 in [ 5 ] , the length of the word \langle n\rangle is given by |\langle n\rangle|=\log_{3/2}(n)+O(1) . Definition 2.1 . We say that a word w\in\{0,1,2\}^{*} is admissible if w=\langle n\rangle for some n\in\mathbb{N} . We denote by \mathcal{L}\subset\{0,1,2\}^{*} the language of admissible words, i.e. \mathcal{L}=\{\langle n\rangle\mid n\in\mathbb{N}\}. As observed in [ 1 ] , the language \mathcal{L} is prefix-closed , i.e. stable under taking prefixes. That is, every prefix w^{\prime} of an admissible word w=w^{\prime}w^{\prime\prime} is itself admissible. Conversely, given an admissible word w\in\mathcal{L} , which extensions among w0,w1,w2 remain admissible? The answer involves the Collatz 3x+1 function. See Proposition 3.1 . 2.1. Construction of \langle n\rangle Given n\in\mathbb{N} , a recursive construction of \langle n\rangle\in\{0,1,2\}^{*} may be described as follows. Set \langle 0\rangle=\epsilon , the empty word. Proposition 2.2 . Let n_{0}\in\mathbb{N}\setminus\{0\} . Denote \langle n_{0}\rangle=a_{k}\cdots a_{0} with a_{i}\in\{0,1,2\} for all i . Then \begin{cases}\ a_{0}\equiv 2n_{0}\bmod 3,\\ \ \langle n_{0}\rangle=\langle n_{1}\rangle a_{0}\textrm{ where }n_{1}=(2n_{0}-a_{0})/3.\end{cases} Moreover, we have a_{0}=2n_{0}-3n_{1} and a_{0}\equiv n_{1}\bmod 2 . Proof. As \langle n_{0}\rangle=a_{k}\cdots a_{0} , we have \displaystyle n_{0} \displaystyle=\frac{1}{2}\big(a_{k}(3/2)^{k}+\dots+a_{1}(3/2)+a_{0}\big) \displaystyle=\left(\frac{1}{2}\big(a_{k}(3/2)^{k-1}+\dots+a_{1}\big)\right)\frac{3}{2}+\frac{a_{0}}{2}. Hence a_{0}\equiv 2n_{0}\bmod 3 , and setting n_{1}=(2n_{0}-a_{0})/3 , we have \langle n_{1}\rangle=a_{k}\cdots a_{1} . The claims follow. ∎ 2.2. The odometer The odometer constructs the word \langle n+1\rangle from the word \langle n\rangle . It is described as a finite automaton in [ 1 ] . Here is an equivalent description. Let \sigma=(2\ 1\ 0) be the cyclic permutation 2\mapsto 1\mapsto 0\mapsto 2 . For a word u\in\{0,1,2\}^{*} , let u^{\sigma}\in\{0,1,2\}^{*} denote the word obtained by applying \sigma to each letter of u . For instance, 12012^{\sigma}=01201 , then identified to 1201 . Proposition 2.3 . Let n\in\mathbb{N} . If \langle n\rangle=u0v\in\{0,1,2\}^{*} with v\in\{1,2\}^{*} , then \langle n+1\rangle=u2v^{\sigma} . Proof. See [ 1 , Section 3.2.3, pp. 66-67] . ∎ The rule applies even if u or v is the empty word. In particular, if \langle n\rangle=v\in\{1,2\}^{*} , then identifying v with 0v , we get \langle n+1\rangle=2v^{\sigma} . Table 1 below, listing \langle n\rangle for 1\leq n\leq 27 , was constructed by hand using this description of the odometer. The interested reader will effortlessly extend it much farther. \begin{array}[]{ccccc}\begin{array}[]{|r||r|}\hline\cr n&\langle n\rangle\\ \hline\cr 1&2\\ 2&21\\ 3&210\\ 4&212\\ 5&2101\\ 6&2120\\ 7&2122\\ 8&21011\\ 9&21200\\ \hline\cr\end{array}&&\begin{array}[]{|r||r|}\hline\cr n&\langle n\rangle\\ \hline\cr 10&21202\\ 11&21221\\ 12&210110\\ 13&210112\\ 14&212001\\ 15&212020\\ 16&212022\\ 17&212211\\ 18&2101100\\ \hline\cr\end{array}&&\begin{array}[]{|r||r|}\hline\cr n&\langle n\rangle\\ \hline\cr 19&2101102\\ 20&2101121\\ 21&2120010\\ 22&2120012\\ 23&2120201\\ 24&2120220\\ 25&2120222\\ 26&2122111\\ 27&21011000\\ \hline\cr\end{array}\end{array} Table 1. The words \langle n\rangle for 1\leq n\leq 27 2.3. The Collatz conjecture Dating back to the 1930’s, this conjecture deals with iterates T^{(k)} of the Collatz function T:\mathbb{N}\to\mathbb{N} defined in ( 1 ). Conjecture 2.4 . Let n_{0}\in\mathbb{N} , n_{0}\geq 1 . Then there exists k\geq 1 such that T^{(k)}(n_{0})=1 . Here is an equivalent version, as shown below. Conjecture 2.5 . Let n_{0}\in\mathbb{N} , n_{0}\geq 1 . Then the proportion of even numbers in the sequence of iterates T^{(k)}(n_{0}) tends to 1/2 as k tends to infinity. Proposition 2.6 . Conjectures 2.4 and 2.5 are equivalent. We thank the anonymous referee for insisting to formally state Proposition 6 and carefully prove it, as opposed to just a remark with informal arguments as in the original version of this article. Proof. If Conjecture 2.4 holds, then at some point the sequence of iterates T^{(k)}(n_{0}) collapses to 1,2,1,2,1,2,\dots , whence Conjecture 2.5 . Conversely, assume Conjecture 2.5 holds. Given an integer n\geq 1 , let \Omega(n)=(n,T(n),T^{(2)}(n),\dots) denote the trajectory of n under iteration of T . Moreover, given k\geq 1 , let \Omega_{k}(n)=(n,T(n),T^{(2)}(n),\dots,T^{(k-1)}(n)), and let k_{1} denote the number of odd integers in \Omega_{k}(n) . Assume for a contradiction that Conjecture 2.4 fails. Then there is an integer m\geq 10 such that \min\Omega(m)=m . (Note that \min\Omega(n)=1 for n\leq 9 .) We claim that for all k\geq 1 , we have (2) T^{(k)}(m)\leq\frac{(3.1)^{k_{1}}}{2^{k}}m. The proof is straightforward, by induction on k . Indeed, ( 2 ) holds for k=1 , since m is odd by minimality in \Omega(m) and T(m)=(3m+1)/2=(3+1/m)m/2\leq(3.1)m/2 since m\geq 10 . Let now k\geq 1 and assume that T^{(k)}(m)\leq(3.1)^{k_{1}}/2^{k}m . Denote n=T^{(k)}(m) . If n is even, then T^{(k+1)}(m)=T(n)=n/2\leq(3.1)^{k_{1}}/2^{k+1}m , as desired. If n is odd, then T^{(k+1)}(m)=T(n)=(3+1/n)n/2\leq(3+1/m)n/2\leq(3.1)n/2 since n\geq m\geq 10 . Hence T^{(k+1)}(m)\leq(3.1)^{k_{1}+1}/2^{k+1}m , as desired. The claim is proved. Now (3.1)^{k_{1}}/2^{k}\geq 1 if and only