betaFeedback

WorksMathematics

Common values of linear recurrences related to Ankeny-Brauer-Chowla polynomials

Abstract

Let Pu(X)=XQ(X)(X−u)+1P_u(X)=XQ(X)(X-u)+1, where Q(X)∈Z[X]Q(X)\in \mathbb{Z}[X] is monic, separabel, has constant term ±1\pm 1, and degree q≥1q\ge 1. Denote (An(u))(A_n(u)) the linear recursive sequence having characteristic polynomial Pu(X)P_u(X) and initial terms A0,…,Aq+1∈ZA_0,\ldots,A_{q+1}\in \mathbb{Z} such that at least one of them is non-zero. Then there exist effectively computable constants n0,u0n_0,u_0 depending only on Q(X)Q(X) and max⁡{∣A0∣,…,∣Aq+1∣}\max\{|A_0|,\ldots, |A_{q+1}|\} such that if ∣An(u)∣=∣Am(u)∣|A_n(u)| = |A_m(u)| holds for some integers u,n,mu,n,m with u>u0u>u_0 and n≠mn\not=m then $\max{

Introduction

Let Pu(X)=XQ(X)(X−u)+1P_u(X)=XQ(X)(X-u)+1 with Q(X)∈Z[X]Q(X)\in\mathbb{Z}[X] and with the non-negative integer parameter uu. Such kind of polynomials seems to be investigated at the first time by Ankeny, Brauer and Chowla in 1956 [2]. They considered only such QQ's, which have solely simple, integer roots. They proved that Pu(X)P_u(X) is irreducible and the number fields generated by one of its roots have as large as possible class number for infinitely many uu. To the honor of the authors I will call Pu(X)P_u(X) the family of ABC polynomials.

One can associate to any integer polynomial a Thue equation. Denote the degree of Q(X)Q(X) by qq then its associated Thue equation is

Yq+2Pu(X/Y)=X(YqQ(X/Y))(X−uY)+Yq+2=1.(1)Y^{q+2}P_u(X/Y)=X(Y^qQ(X/Y))(X-uY)+Y^{q+2}=1. \tag*{(1)}

If Q(X)=∏j=1q(X−aj)Q(X)=\prod_{j=1}^{q}(X-a_j) with a1,…,aqa_1,\ldots,a_q pairwise distinct integers, then (1) has the solutions (1,0),(0,1),(aj,1)(1,0),(0,1),(a_j,1), j=1,…,qj=1,\ldots,q and (0,−1)(0,-1) if qq is even. We call them trivial. Bombieri and Schmidt [4] proved that a Thue equation of degree nn admits at most O(n)O(n) solutions. Moreover they pointed out that Thue equations associated to ABC polynomials are with this respect up to a constant factor extremal, because they admit, at least n+1n+1 solutions.

For q=1q=1, a1=1a_1=1 and with −(u+1)-(u+1) instead of uu, Mignotte and Tzanakis [20] proved that if ∣u∣>3.67⋅1032|u|>3.67\cdot10^{32}, then (1) has only one non-trivial solution, namely (x,y)=(−u−1,−1)(x,y)=(-u-1,-1). Mignotte [17] proved the same for ∣u∣≥5|u|\ge5 and settled the cases u=0,1,2,3,4u=0,1,2,3,4 too.

For q=2q=2, a1=1a_1=1, a2=−1a_2=-1, equation (1) was completely solved for ∣a∣>1028|a|>10^{28} by Pethő [22]. His result was extended to ∣a∣≥2|a|\ge2 by Mignotte, Pethő and Roth [19]. In this case, there is again exactly one (parametrized) non-trivial solution, namely (x,y)=(1,−a)(x,y)=(1,-a). Marzenta [12] gave an essentially new proof of the theorem of Pethő. His proof is based on a method developed for the solution of parametrized families of norm form equations and presented Amozo, Masser and Zannier in [1].

Thomas [29] investigated equation (1) in the case, when the factor X−uX-u disappear, but the aia_i denote integer polynomials of distinct degrees in an integer variable uu, ai=ai(u)∈Z[u]a_i=a_i(u)\in\mathbb{Z}[u], i=1,2i=1,2. He conjectured that (1) has only the trivial solutions if uu is large enough, and pointed out that the conjecture might also be true in specific cases of polynomials of equal degrees. In the case n=3n=3, a1=uka_1=u^k, a2=ula_2=u^l, where 0<k<l0<k<l are integers, Thomas proved in [29], that for u≥(2⋅106(k+2l))4.85/(l−k)u\geq\left(2\cdot10^6(k+2l)\right)^{4.85/(l-k)}, (1) has only the trivial solutions. Heuberger [8] settled Thomas conjecture even if Q(X)Q(X) is the product of any number of factors X−ai(u)X-a_i(u), but only under certain assumptions.

The above examples show that not only the trivial, but most of the non-trivial solutions are polynomials in uu. Zigler [32] reformulated Thomas’ conjecture for solutions (x,y)∈C[u](x,y)\in\mathbb{C}[u] of (1) and proved it even in the case when a1(u),a2(u)∈C[u]a_1(u),a_2(u)\in\mathbb{C}[u].

The proofs of most of the mentioned results consist of three main steps; the first is the reformulation of (1) to a weighted unit equation in two variables. In the second step it is shown that writing a solution of the weighted unit equation in an appropriate system of fundamental units, one of the exponents must be larger than culog⁡ucu\log u with a positive constant cc. In the final step some result on logarithmic forms imply that there is no solution provided uu is large enough. Halter-Koch, Lettl, Pethő and Tichy [7] implemented this program, when Pu(X)P_u(X) denotes a classical ABC polynomial, i.e. when Q(X)=∏j=1q(X−aj)Q(X)=\prod_{j=1}^{q}(X-a_j) with a1,…,aqa_1,\ldots,a_q pairwise distinct integers. They finished the first two steps successfully, but failed with the third one, because the available lower bounds for linear forms in logarithms of algebraic numbers were not sharp enough. They were able to finish the work by invoking the Lang-Waldschmidt conjecture, see [10]. Despite of the 25 years development the Lang-Waldschmidt conjecture is still far from a proof.

The present paper was motivated by [7] and by the paper of Pethő and Tengely [24]. They investigated common values of linear recursive sequences associated to Shanks’ simplest cubics. The Thue equation associated to the same family of polynomials was solved by Thomas [28] and Mignotte [16] more than thirty years ago. Taking heart I tried the same game with ABC polynomials. More precisely write Pu(X)=∑j=0q+2pj(u)XjP_u(X)=\sum_{j=0}^{q+2}p_j(u)X^j, then pj(u)p_j(u), j=1,…,q+1j=1,\ldots,q+1 are linear polynomials with integer coefficients of uu and p0=pq+2=1p_0=p_{q+2}=1. Let A0(u),…,Aq+1(u)∈ZA_0(u),\ldots,A_{q+1}(u)\in\mathbb{Z} and define An(u)A_n(u), n≥q+2n\geq q+2 by the linear recursion

An(u)=−pq+1(u)An−1(u)−⋯−p0(u)An−q−2(u).(2)A_n(u)=-p_{q+1}(u)A_{n-1}(u)-\cdots-p_0(u)A_{n-q-2}(u). \tag*{(2)}

For simplicity we denote the initial terms by A0,…,Aq+1A_0,\ldots,A_{q+1}, and to exclude the zero sequence we assume that at least one of them is not zero. Set L=max⁡{∣A1∣,…,∣Aq+1∣}L=\max\{|A_1|,\ldots,|A_{q+1}|\}.

Plainly An(u)∈Z[u]A_n(u)\in\mathbb{Z}[u] for all n≥0n\geq0. The polynomial Pu(X)P_u(X) is called the characteristic polynomial of the sequence (An(u))(A_n(u)). As p0(u)=1p_0(u)=1 this sequence can be extended to negative indexes too such that all members of the two sided sequence belong to Z[u]\mathbb{Z}[u]. Hence the search for the integers, which appear multiple times in the two-parametric sequence (An(u))(A_n(u)) is meaningful and non-trivial. In the language of Diophantine equations this means to find solutions of the equation

∣An(u)∣=∣Am(u)∣(3)|A_n(u)|=|A_m(u)| \tag*{(3)}

in n,m,u∈Zn,m,u\in\mathbb{Z}, n≠mn\ne m. Our aim is to prove the following.

Theorem 1. Let Pu(X)=XQ(X)(X−u)+1P_u(X)=XQ(X)(X-u)+1, where Q(X)∈Z[X]Q(X)\in\mathbb{Z}[X] is monic, separable, has constant term ±1\pm1, and degree q≥1q\ge1. Denote (An(u))(A_n(u)) the linear recursive sequence having characteristic polynomial Pu(X)P_u(X) and initial terms A0,…,Aq+1∈ZA_0,\ldots,A_{q+1}\in\mathbb{Z} such that at least one of them is non-zero. Then there exist effectively computable constants n0,u0n_0,u_0 depending only on Q(X)Q(X) and L=max⁡{∣A0∣,…,∣Aq+1∣}L=\max\{|A_0|,\ldots,|A_{q+1}|\} such that if (3) holds for some integers u,n,mu,n,m with u>u0u>u_0 and n≠mn\ne m then

∣n∣,∣m∣<n0.|n|,|m|<n_0.

After submitting the manuscript we we made acquainted with the paper of Noubissie and Tichy [21]. They investigate common values of linear recursive sequences associated to generalized ABC-polynomials, πU(X)=X(X−P1(U))⋯(X−Pq−1(U))+1\pi_U(X)=X(X-P_1(U))\cdots(X-P_{q-1}(U))+1, where P1(U),…,Pq−1(U)∈Z[U]P_1(U),\ldots,P_{q-1}(U)\in\mathbb{Z}[U] are pairwise distinct and such that deg⁡P1≤⋯≤deg⁡Pq−3<deg⁡Pq−2<deg⁡Pq−1\deg P_1\le\cdots\le\deg P_{q-3}<\deg P_{q-2}<\deg P_{q-1}. They proved that these sequences have at most two sufficiently large common values. Their result is not effective. Unfortunately it say nothing on common values of linear recursive sequences associated to our version of ABC-polynomials, but with ∣Q(0)∣>1|Q(0)|>1.

The strategy of our proof is similar that were used for the investigation of parametric families of Thue equations. In Section 3 we carefully investigate the roots of Pu(X)P_u(X). It turns out that if uu is large enough then they behave regularly. It admits two real roots, one αq+1(u)\alpha_{q+1}(u) is as large as uu, and the other α0(u)\alpha_0(u) lies near to zero, moreover Q(0)α0(u)αq+1(u)Q(0)\alpha_0(u)\alpha_{q+1}(u) is asymptotically one.

Section 4 is devoted to the ABC sequence with special attention to the weights of α0(u)\alpha_0(u) and αq+1(u)\alpha_{q+1}(u) appearing in the Binet formula of (An(u))(A_n(u)). In both sections the machinery of Puiseux expansions of algebraic functions play important role. In Section 5 we collected necessary tools from transcendental number theory.

In the last section we prove Theorem 1. First we deal with the easy cases, i.e. if nn and mm have the same signs. The hard case is, when nm<0nm<0. Then we derive a lower bound for max⁡{∣n∣,∣m∣}\max\{|n|,|m|\} in term of uu, which have the shape gufgu^f with positive constants g,fg,f. Having it we derive the desired upper bound for max⁡{∣n∣,∣m∣}\max\{|n|,|m|\} by using Matveev’s theorem.

The symbols cx,nx,uxc_x,n_x,u_x denote constants, which depend only on Q(X)Q(X) and LL, the uxu_x are such that the corresponding predicate holds only for u≤uxu\le u_x, the nxn_x play similar role with respect to max⁡{∣n∣,∣m∣}\max\{|n|,|m|\}.

References

  1. [1]F. Amoroso, D. Masser and U. Zannier, Pencils of norm form equations and a conjecture of Thomas. Mathematika 67 (2021), no. 4, 897–916.
  2. [2]N.C. Ankeny, R. Brauer and S. Chowla, A note on the class numbers of algebraic number fields, Amer. J. Math. 78 (1956), 51–61.
  3. [4]E. Bombieri and W.M. Schmidt, On Thue’s equation, Invent. Math. 109 (1992), 445–472.
  4. [6]J-H. Evertse and K. Győry, Unit Equations in Diophantine Number Theory, Cambridge Studies in Advanced Mathematics, 146, 2015.DOI
  5. [7]F. Halter-Koch, G. Lettl, A. Pethő and R.F. Tichy, Thue equations associated with Ankeny-Brauer-Chowla number fields. J. London Math. Soc. (2) 60 (1999), 1–20.
  6. [8]C. Heuberger, On explicit bounds for the solutions of a class of parametrized Thue equations of arbitrary degree. Monatsh. Math. 132 (2001), 325–339.DOI
  7. [9]H.W.E. Jung, Einführung in die Theorie der algebraischen Funktionen einer Veränderlichen, Vereiniging wissenschaftlicher Verleger, Walter de Gruyter & Co.
  8. [10]S. Lang, Elliptic Curves: Diophantine Analysis, Springer Verlag, 1978.DOI
  9. [12]G. Marzenta, Effective resolution of families of norm form equations, Rend. Lincei Mat. Appl. 36 (2025), 285–313.DOI
  10. [13]E.M. Matveev, An explicit lower bound for a homogeneous rational linear form in logarithms of algebraic numbers, II, . Ross. Akad. Nauk Ser. Mat. 64 (2000), 125–180. (English translation in Izv. Math. 64 (2000), 1217–1269.DOI
  11. [14]M. Mignotte, Intersection des images de certaines suites récurrentes linéaires. Theor. Comput. Sci., 7 (1978), 117–121.DOI
  12. [16]M. Mignotte, Verification of a conjecture of E. Thomas. J. Number Theory 44 (1993), 172–177.DOI
  13. [17]M. Mignotte, Pethő’s cubics, Dedicated to Professor Kálmán Győry on the occasion of his 60th birthday, Publ. Math. Debrecen 56 (2000), 481–505.
  14. [19]M. Mignotte, A. Pethő and R. Roth, Complete solutions of quartic Thue and index form equations, Math. Comp., 65 (1996), 341–354.
  15. [20]M. Mignotte and N. Tzanakis, On a family of cubics, J. Number Theory 39 (1991), 41–49.DOI
  16. [21]A. Noubissie and R.F. Tichy, Large common values of generalized Ankeny-Brauer-Chowla recurrences, arXiv:2606.27885v1.
  17. [22]A. Pethő, Complete solutions to a family of quartic diophantine equations, Math. Comp. 57 (1991), 777–798.
  18. [24]A. Pethő and Sz. Tengely, Common values of linear recurrences related to Shanks’ simplest cubics, J. Number Theory 267 (2025), 34 - 79.DOI
  19. [25]G. Sanchez and F. Luca, Linear combinations of factorials and S-units in a binary recurrences sequence, Ann. Math. Québec 38 (2014), 169-188.DOI
  20. [26]W.M. Schmidt, Einstein’s theorem on power series expansions of algebraic functions, Acta Arith. 56 (1990), 161–179.DOI
  21. [27]T.N. Shorey and R. Tijdeman, Exponential Diophantine Equations, Cambridge Tracts in Mathematics, 87. Cambridge University Press, 1986.DOI
  22. [28]E. Thomas, Complete solutions to a family of cubic diophantine equations, J. Number Theory 34 (1990), 235-250.DOI
  23. [29]E. Thomas, Solutions to Certain Families of Thue Equations, J. Number Theory 43 (1993), 319–369.DOI
  24. [30]M. Waldschmidt, Diophantine Approximation on Linear Algebraic Groups, Springer Verlag, 2000.DOI
  25. [31]B.M.M. de Weger, Algorithms for Diophantine equations. CWI Tract, 65. Stichting Mathematisch Centrum, Centrum voor Wiskunde en Informatica, Amsterdam, 1989.
  26. [32]V. Ziegler, Thomas’ conjecture over function fields. J. Théor. Nombres Bordeaux 19 (2007), 289–309.DOI

Paper details

vlnc:W_ctvnhk6e
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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