betaFeedback

WorksMathematics

Estimates of Lucas series and expansions in non-integer bases

Abstract

Special Fibonacci and Lucas type expansions proved useful to ``correct'' some theorems on the number of expansions in non-integer bases \cite{BaiKomLor2021, KomLorPed2024, KomLorPed2025}. The proof of a key lemma was non-constructive. It was made constructive in the Fibonacci case by giving a precise estimate of the growth of the partial sums of the series of the inverses of the Fibonacci numbers \cite{KomPed2025}. In this paper we extend this result to the general case.

Introduction

In integer bases every real number has at most two expansions. In non-integer bases the situation is completely different: if MM is a positive integer and M<q<M+1M < q < M+1, then Lebesgue almost every x∈[0,Mq−1]x \in[0,\frac{M}{q-1}] has 2ℵ02^{\aleph_0} expansions of the form

x=∑i=1∞ciqi,(ci)∈{0,1,…,M}N(1)x = \sum_{i=1}^{\infty} \frac{c_i}{q^i}, \qquad(c_i) \in\{0,1,\ldots,M\}^{\mathbb{N}} \tag*{(1)}

[14, 3]. Moreover, there exists a largest base qMq_M such that if q<qMq < q_M, then every x∈(0,Mq−1)x \in(0,\frac{M}{q-1}) has 2ℵ02^{\aleph_0} expansions of the form (1.1) [4, 6, 5, 2]. For example, q1q_1 is the Golden ratio. More generally,

q2p−1=p+p2+4p2q_{2p-1} = \frac{p+\sqrt{p^2+4p}}{2}

and q2p=p+1q_{2p}=p+1 for p=1,2,…p=1,2,\ldots. Similar results hold for arbitrary ternary alphabets, too [7, 10, 15].

If M=2p−1M=2p-1 is odd, then the last property no longer holds for q=qMq=q_M. However, the property still holds for a closely related expansion with integer denominators. Indeed, introducing the Lucas sequence

F0=0,F1=1,andFk=p(Fk−1+Fk−2)for k=2,3,…(2)F_0=0,\qquad F_1=1,\qquad\text{and}\qquad F_k=p(F_{k-1}+F_{k-2})\quad\text{for }k=2,3,\ldots \tag*{(2)}

(for simplicity we do not indicate their dependence on pp), we have Fk+1/Fk→qMF_{k+1}/F_k \to q_M as k→∞k \to\infty, and every x∈(0,∑i=1∞MFi)x \in(0,\sum_{i=1}^{\infty}\frac{M}{F_i}) has 2ℵ02^{\aleph_0} expansions of the form

x=∑i=1∞ciFi,(ci)∈{0,1,…,M}Nx = \sum_{i=1}^{\infty} \frac{c_i}{F_i}, \qquad(c_i) \in\{0,1,\ldots,M\}^{\mathbb{N}}

[1, 8, 9].

The convergence relation Fk+1/Fk→qMF_{k+1}/F_k \to q_M easily follows from the explicit formula

Fk=Ak−BkA−B,k=0,1,…,F_k=\frac{A^k-B^k}{A-B},\qquad k=0,1,\ldots,

where

A:=qM=p+p2+4p2andB:=p−p2+4p2=−pAA:=q_M=\frac{p+\sqrt{p^2+4p}}{2} \qquad\text{and}\qquad B:=\frac{p-\sqrt{p^2+4p}}{2}=-\frac{p}{A}

are the solutions of the equation x2=px+px^2=px+p, and from the obvious inequalities

p<A<p+1and−1<B<0.p<A<p+1\qquad\text{and}\qquad-1<B<0.

A main idea of the proof in [1] was that in case M=1M=1 we may remove from the sequence (1/Fi)(1/F_i) infinitely many elements so that the remaining sequence (pi)(p_i) satisfies the inequalities

pn≤∑i=n+1∞pip_n \le\sum_{i=n+1}^{\infty} p_i

for every nn. The proof of this lemma was not constructive. The following more precise estimates, obtained in [11], imply that the removal of the elements 1/F2⋅3n1/F_{2\cdot3^n} from (1/Fi)(1/F_i) yields such a sequence (pn)(p_n):

Theorem 1.1. Let (Fi)(F_i) denote the Fibonacci sequence. If kk is a positive odd integer, then

∑i=k+23k+11Fi<1Fk<∑i=k+23k+21Fi.\sum_{i=k+2}^{3k+1}\frac{1}{F_i}<\frac{1}{F_k}<\sum_{i=k+2}^{3k+2}\frac{1}{F_i}.

The object of the present work is to extend Theorem 1.1 to the Lucas sequences (1.2); they similarly lead to a constructive proof of the main results of [8, 9].

The method of our proof allows us to consider a larger class of Lucas sequences. We fix two integers p≥q≥1p\ge q\ge1, and we define the sequence

F0=0,F1=1,andFk=pFk−1+qFk−2for k=2,3,…;(3)F_0=0,\qquad F_1=1,\qquad\text{and}\qquad F_k=pF_{k-1}+qF_{k-2}\quad\text{for }k=2,3,\ldots; \tag*{(3)}

for simplicity we do not indicate their dependence on p,qp,q.

Binet’s formula readily extends to these sequences: introducing the solutions

A:=p+p2+4q2andB:=p−p2+4q2=−qAA:=\frac{p+\sqrt{p^2+4q}}{2} \qquad\text{and}\qquad B:=\frac{p-\sqrt{p^2+4q}}{2}=-\frac{q}{A}

of the equation x2=px+qx^2=px+q, we have p<A<p+1p<A<p+1, −1<B<0-1<B<0, and

Fk=Ak−BkA−BF_k=\frac{A^k-B^k}{A-B}

for all k≥0k\ge0.

First we consider the special case where q=1q=1.

Theorem 1.2. If p≥2p\ge2 and q=1q=1, then

p−1Fk+1+∑i=k+23kpFi<1Fk<p−1Fk+1+∑i=k+23k+1pFi.(4)\frac{p-1}{F_{k+1}}+\sum_{i=k+2}^{3k}\frac{p}{F_i}<\frac{1}{F_k}<\frac{p-1}{F_{k+1}}+\sum_{i=k+2}^{3k+1}\frac{p}{F_i}. \tag*{(4)}

for all positive odd integers kk.

Observe the difference between the cases p=1p=1 and p≥2p\ge2: the summation limits are 3k+13k+1 and 3k+23k+2 in Theorem 1.1, and 3k3k and 3k+13k+1 in Theorem 1.2.

Our next result shows that the summation limits are more complicated for q≥2q\ge2.

Theorem 1.3. If p≥q≥2p\ge q\ge2, then we have

p−1Fk+1+∑i=k+2K1(k)p+q−1Fi<1Fk<p−1Fk+1+∑i=k+2K2(k)p+q−1Fi.(5)\frac{p-1}{F_{k+1}}+\sum_{i=k+2}^{K_1(k)}\frac{p+q-1}{F_i}<\frac{1}{F_k}<\frac{p-1}{F_{k+1}}+\sum_{i=k+2}^{K_2(k)}\frac{p+q-1}{F_i}. \tag*{(5)}

for all positive odd integers kk, where

K1(k)=⌊(3−ln⁡qln⁡A)k+ln⁡(p+q−1)ln⁡A−1⌋K_1(k)=\left\lfloor\left(3-\frac{\ln q}{\ln A}\right)k+\frac{\ln(p+q-1)}{\ln A}-1\right\rfloor

and

K2(k)=⌈(3−ln⁡qln⁡A)k+ln⁡(p+q−1)−2ln⁡qln⁡A+1⌉.K_2(k)=\left\lceil\left(3-\frac{\ln q}{\ln A}\right)k+\frac{\ln(p+q-1)-2\ln q}{\ln A}+1\right\rceil.

Remark 1.4. It follows from the definition that

K1(k)=max⁡{K∈Z:qkA−3k+1≤p+q−1}K_1(k)=\max\left\{K\in\mathbb{Z}:q^kA^{-3k+1}\le p+q-1\right\}

and

K2(k)=min⁡{K∈Z:qk+2A−3k−1≥p+q−1}.K_2(k)=\min\left\{K\in\mathbb{Z}:q^{k+2}A^{-3k-1}\ge p+q-1\right\}.

References

  1. [1]C. Baiocchi, V. Komornik, P. Loreti, Fibonacci expansions, Rend. Lincei Mat. Appl. 32 (2021), 379–389.arxiv.org/abs/2102.12410
  2. [2]S. Baker, Generalized golden ratios over integer alphabets, Integers 14 (2014), Paper no. A15, 28 pp.
  3. [3]K. Dajani, M. de Vries, Invariant densities for random β-expansions, J. Eur. Math. Soc. 9 (2007), no. 1, 157–176.DOI
  4. [4]L. C. Eggan and C. L. Vanden Eynden, “Decimal” expansions to nonintegral bases, Amer. Math. Monthly 73 (1966), 576–582.DOI
  5. [5]P. Erdős, M. Horváth, I. Joó, On the uniqueness of the expansions 1 = ∑ q⁻ⁿⁱ, Acta Math. Hungar. 58 (1991), 333–342.
  6. [6]P. Erdős, I. Joó, V. Komornik, Characterization of the unique expansions 1 = ∑ q⁻ⁿⁱ and related problems. Bull. Soc. Math. France 118 (1990), 377–390.DOI
  7. [7]V. Komornik, A. C. Lai, M. Pedicini, Generalized golden ratios of ternary alphabets. J. Eur. Math. Soc. 13 (2011), 4, 1113–1146.
  8. [8]V. Komornik, P. Loreti, M. Pedicini, A quasi-ergodic approach to non-integer base expansions, Journal of Number Theory, 254 (2024) 146–168.DOI
  9. [9]V. Komornik, P. Loreti, M. Pedicini, Multiple Kakeya expansions, Ann. Sc. Norm. Super. Pisa Cl. Sci. (to appear).DOI
  10. [10]V. Komornik, M. Pedicini, Critical bases for ternary alphabets, Acta Math. Hungar., 152 (1) (2017), 25–57.arxiv.org/abs/1605.08990
  11. [11]V. Komornik, M. Pedicini, Estimates of Fibonacci series, submitted.
  12. [14]N. Sidorov, Almost every number has a continuum of β-expansions, Amer. Math. Monthly 110 (2003), 9, 838–842.DOI
  13. [15]W. Steiner, Thue–Morse–Sturmian words and critical bases for ternary alphabets, Bull. Soc. Math. France 148 (2020), no. 4, 597–611.arxiv.org/abs/1902.07078

Paper details

vlnc:W_ranbkp7k
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY-SA 4.0
Usage

views · 5

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents