betaFeedback

WorksMathematics

On the exponential Diophantine equation (Pn+1(k))x−(Pn−1(k))x=2Pm(k)\left(P_{n+1}^{(k)}\right)^{x}-\left(P_{n-1}^{(k)}\right)^{x}=2P_{m}^{(k)}

Abstract

For the Pell numbers one has Pn+1−Pn−1=2PnP_{n+1}-P_{n-1}=2P_n and Pn+12−Pn−12=2P2nP_{n+1}^2-P_{n-1}^2=2P_{2n}. We look for all the solutions of (Pn+1(k))x−(Pn−1(k))x=2Pm(k)\big(P_{n+1}^{(k)}\big)^x-\big(P_{n-1}^{(k)}\big)^x=2P_m^{(k)} in positive integers k≥2k\ge2, nn, mm, xx, where P(k)P^{(k)} is the kk-generalized Pell sequence. We prove that for k=2k=2 the two identities above give all the solutions, and that for k≥3k\ge3 the only solutions are (n,m,x)=(1,1,1),(1,2,2),(2,2,1)(n,m,x)=(1,1,1),(1,2,2),(2,2,1). In contrast with the analogous equation for kk-generalized Fibonacci numbers, there are no parametric families of solutions. The proof uses an auxiliary equation with Fibonacci numbers which does not depend on kk, Matveev's lower bound for linear forms in logarithms, the reduction of Dujella and Pethő, and, for large kk, the fact that the dominant root of P(k)P^{(k)} is very close to φ2φ^2.

Introduction

The Fibonacci numbers satisfy Fn+12−Fn−12=F2nF_{n+1}^2-F_{n-1}^2=F_{2n}. Patel and Chaves [12] asked when Fn+1x−Fn−1xF_{n+1}^x-F_{n-1}^x is again a Fibonacci number and solved the equation Fn+1x−Fn−1x=FmF_{n+1}^x-F_{n-1}^x=F_m completely. Apart from trivial cases, the only solutions come from x=2x=2, m=2nm=2n. Gómez, Luca and Gómez [9] treated Fn+1x−Fn−1x=FmyF_{n+1}^x-F_{n-1}^x=F_m^y. The same question for the kk-generalized Fibonacci numbers was studied by Bensella, Patel and Behloul [2], and solved completely by Gómez, Gómez and Luca [8]; in that case there are two infinite parametric families of solutions, with n∈{4,5}n\in\{4,5\} and kk growing exponentially with the parameter. These families come from the fact that the kk-generalized Fibonacci numbers Fr(k)F_r^{(k)} are powers of 22 for 2≤r≤k+12\le r\le k+1.

In this paper, we are interested in the Pell analogue because of the two identities

Pn+1−Pn−1=2Pn,Pn+12−Pn−12=(Pn+1−Pn−1)(Pn+1+Pn−1)=2P2n,(1)P_{n+1}-P_{n-1}=2P_n,\qquad P_{n+1}^2-P_{n-1}^2=(P_{n+1}-P_{n-1})(P_{n+1}+P_{n-1})=2P_{2n}, \tag*{(1)}

where PnP_n is the nn-th Pell number (the second one follows from the first and from Pn+1+Pn−1=QnP_{n+1}+P_{n-1}=Q_n, PnQn=P2nP_nQ_n=P_{2n}, with QnQ_n the Pell–Lucas numbers). So, for Pell numbers, the natural question is when Pn+1x−Pn−1xP_{n+1}^x-P_{n-1}^x is twice a Pell number. Let k≥2k\ge2 and let P(k)=(Pn(k))nP^{(k)}=(P_n^{(k)})_n be the kk-generalized Pell sequence. We study the equation

(Pn+1(k))x−(Pn−1(k))x=2Pm(k)(2)\left(P_{n+1}^{(k)}\right)^x-\left(P_{n-1}^{(k)}\right)^x=2P_m^{(k)} \tag*{(2)}

in positive integers k≥2k\ge2, nn, mm and xx. In a companion paper [7] we treat, in the same spirit, (Pn(k))x+(Pn+1(k))x=Pm(k)\left(P_n^{(k)}\right)^x+\left(P_{n+1}^{(k)}\right)^x=P_m^{(k)}, which generalizes the Pell identity Pn2+Pn+12=P2n+1P_n^2+P_{n+1}^2=P_{2n+1} to arbitrary indices.

In a first version of this work we tried to follow the method of [8] step by step, and we obtained two parametric families similar to the families for kk-Fibonacci numbers. A numerical check showed that they are not solutions. For example, the first member (k,n,m,x)=(7,4,11,3)(k,n,m,x)=(7,4,11,3) gives

343−53=39179while2P11(7)=21854.34^3-5^3=39179\qquad\text{while}\qquad2P_{11}^{(7)}=21854.

The reason is that the property used in [8] has no analogue here. By a result of Kılıç [10], the first k+1k+1 non-zero terms of P(k)P^{(k)} are not powers of 22 but their are equal to the Fibonacci numbers with odd index, Pr(k)=F2r−1P_r^{(k)}=F_{2r-1} for 1≤r≤k+11\le r\le k+1. Hence, when n≤kn\le k and m≤k+1m\le k+1, (2) becomes an equation which does not depend on kk, namely

F2n+1−F2n−3=2F2m−1F_{2n+1}-F_{2n-3}=2F_{2m-1}

and this equation has no solution (see Theorem 1.2). This is the main reason why the final answer is so short. In fact, in the range where the kk-Fibonacci numbers produce whole families of solutions, the kk-Pell numbers produce none.

Theorem 1.1. Let k≥2k\ge2, n≥1n\ge1, m≥1m\ge1 and x≥1x\ge1 be integers satisfying (2).

(i) If k=2k=2, then (m,x)=(n,1)(m,x)=(n,1) or (m,x)=(2n,2)(m,x)=(2n,2); conversely, these are solutions for every n≥1n\ge1.

(ii) If k≥3k\ge3, then (n,m,x)∈{(1,1,1),(1,2,2),(2,2,1)}(n,m,x)\in\{(1,1,1),(1,2,2),(2,2,1)\}, and these three triples are solutions for every k≥3k\ge3.

Theorem 1.2. The equation

F2n+1−F2n−3=2F2m−1(3)F_{2n+1}-F_{2n-3}=2F_{2m-1} \tag*{(3)}

has no solution in integers n≥2n\ge2, m≥1m\ge1 and x≥2x\ge2.

We use the following strategy for the proof. The elementary cases (x=1x=1, n=1n=1, and k=2k=2 with x=2x=2) are settled in § 3, and Theorem 1.2 is proved in § 4, with three linear forms in logarithms in Q(5)\mathbb{Q}(\sqrt{5}). For an arbitrary kk we use the same three forms, now in Q(α)\mathbb{Q}(\alpha), where α\alpha is the dominant root of P(k)P^{(k)}. This bounds xx in terms of kk (Section 5). This gives first k<1.05×1045k<1.05\times10^{45}, and six rounds of reduction bring this down to k≤257k\le257 (§ 6). The remaining values 2≤k≤2572\le k\le257 are treated one by one in Section 7.

Preliminaries

The generalized Pell sequence

Let k≥2k\ge2 be an integer. The kk-generalized Pell sequence P(k):=(Pn(k))n≥2−kP^{(k)}:=(P_n^{(k)})_{n\ge2-k} is defined by

Pn(k)=2Pn−1(k)+Pn−2(k)+⋯+Pn−k(k)for all n≥2,P_n^{(k)}=2P_{n-1}^{(k)}+P_{n-2}^{(k)}+\cdots+P_{n-k}^{(k)}\quad\text{for all }n\ge2,

with the initial conditions P−(k−2)(k)=P−(k−3)(k)=⋯=P−1(k)=P0(k)=0P_{-(k-2)}^{(k)}=P_{-(k-3)}^{(k)}=\cdots=P_{-1}^{(k)}=P_0^{(k)}=0 and P1(k)=1P_1^{(k)}=1. For k=2k=2 we obtain the Pell sequence 0,1,2,5,12,29,70,…0,1,2,5,12,29,70,\ldots. The terms of this sequence are called kk-generalized Pell numbers. When kk is fixed we simply write Pn:=Pn(k)P_n:=P_n^{(k)}.

Kılıç [10] showed that the first k+1k+1 non-zero terms of P(k)P^{(k)} are Fibonacci numbers with odd index, namely

Pn(k)=F2n−1for all 1≤n≤k+1.(4)P_n^{(k)}=F_{2n-1}\quad\text{for all }1\le n\le k+1. \tag*{(4)}

The characteristic polynomial of P(k)P^{(k)},

Ψk(x):=xk−2xk−1−xk−2−⋯−x−1,\Psi_k(x):=x^k-2x^{k-1}-x^{k-2}-\cdots-x-1,

is irreducible over Q\mathbb{Q} and has just one root outside the unit circle [4]. Let α:=α(k)\alpha:=\alpha(k) denote that root. It was proved in [4] that

φ2(1−φ−k)<α<φ2,(5)\varphi^2(1-\varphi^{-k})<\alpha<\varphi^2, \tag*{(5)}

where φ=(1+5)/2\varphi=(1+\sqrt{5})/2 is the golden section. We write α1:=α,α2,…,αk\alpha_1:=\alpha,\alpha_2,\ldots,\alpha_k for the roots of Ψk\Psi_k; thus ∣αi∣<1|\alpha_i|<1 for 2≤i≤k2\le i\le k. A direct computation gives

(x−1)Ψk(x)=xk+1−3xk+xk−1+1,so thatα2−3α+1=−α1−k.(6)(x-1)\Psi_k(x)=x^{k+1}-3x^k+x^{k-1}+1,\qquad\text{so that}\qquad\alpha^2-3\alpha+1=-\alpha^{1-k}. \tag*{(6)}

For x0=1+2x_0=1+\sqrt{2} we have x02−3x0+1=1−2x_0^2-3x_0+1=1-\sqrt{2}, so (6) gives (x0−1)Ψk(x0)=1−(2−1)x0k−1=1−x0k−2≤0(x_0-1)\Psi_k(x_0)=1-(\sqrt{2}-1)x_0^{k-1}=1-x_0^{k-2}\le0. Since Ψk\Psi_k has exactly one positive root, we get α≥1+2\alpha\ge1+\sqrt{2}, with equality for k=2k=2.

For an integer k≥2k\ge2 we consider the function

gk(z)=z−1(k+1)z2−3kz+k−1=z−1k(z2−3z+1)+z2−1(z∈C),(7)g_k(z)=\frac{z-1}{(k+1)z^2-3kz+k-1}=\frac{z-1}{k(z^2-3z+1)+z^2-1}\qquad(z\in\mathbb{C}), \tag*{(7)}

and we put g:=gk(α)g := g_k(\alpha).

Lemma 2.1 (Bravo, Herrera and Luca [4]). Let k≥2k \ge2. Then:

(a) αn−2≤Pn(k)≤αn−1\alpha^{n-2} \le P_n^{(k)} \le\alpha^{n-1} for all n≥1n \ge1;

(b) Pn(k)=∑i=1kgk(αi)αinP_n^{(k)} = \sum_{i=1}^{k} g_k(\alpha_i)\alpha_i^n for all n≥2−kn \ge2-k;

(c) en:=Pn(k)−gαne_n := P_n^{(k)} - g\alpha^n satisfies ∣en∣<1/2|e_n| < 1/2 for all n≥2−kn \ge2-k;

(d) 0.276<g<0.50.276 < g < 0.5.

Lemma 2.2 (Bravo and Herrera [3]; see also [5]). For every k≥2k \ge2 and 2≤i≤k2 \le i \le k we have ∣gk(αi)∣<1|g_k(\alpha_i)| < 1.

We shall also use the following elementary facts. For n≥2n \ge2 we have Pn≥2Pn−1P_n \ge2P_{n-1}, since Pn=2Pn−1+P_n = 2P_{n-1} + (non-negative terms); hence, for n≥2n \ge2,

Pn+1≥2Pn+Pn−1≥5Pn−1.(8)P_{n+1} \ge2P_n + P_{n-1} \ge5P_{n-1}. \tag*{(8)}

The sequence (Pn)n≥1(P_n)_{n\ge1} is strictly increasing, with P1=1P_1 = 1, P2=2P_2 = 2 and P3=5P_3 = 5 for every k≥2k \ge2.

Lemma 2.3. Let k≥2k \ge2 and n≥1n \ge1. If Pn(k)=2aP_n^{(k)} = 2^a with a≥1a \ge1, then n=2n = 2 and a=1a = 1.

Proof. See Lemma 2.3 in [7]. □

Approximation for large k

For large kk the dominant root α\alpha is very close to φ2\varphi^2, and Pn(k)P_n^{(k)} remains very close to F2n−1F_{2n-1} (in relative terms) also for n>k+1n > k + 1, as long as nn is small compared with φ2k\varphi^{2k}. We put

c:=1φ2+1=15φ=5−510=0.2763…c := \frac{1}{\varphi^2+1} = \frac{1}{\sqrt{5}\varphi} = \frac{5-\sqrt{5}}{10} = 0.2763\ldots

For instance F2n−1=cφ2n+φ−2n+1/5F_{2n-1} = c\varphi^{2n} + \varphi^{-2n+1}/\sqrt{5}, since cφ=1/5c\varphi= 1/\sqrt{5}.

Lemma 2.4. Let k≥30k \ge30 and write θ:=φ2−α\theta:= \varphi^2 - \alpha, δ:=θ/φ2\delta:= \theta/\varphi^2 and g=c(1+η)g = c(1+\eta). Then

0<θ<1.1711φ−2k,0<δ<0.4474φ−2k,∣η∣≤(0.45k+1.07)φ−2k.0 < \theta< 1.1711\varphi^{-2k}, \qquad0 < \delta< 0.4474\varphi^{-2k}, \qquad|\eta| \le(0.45k+1.07)\varphi^{-2k}.

Consequently, for every j≥0j \ge0,

gαj=cφ2j(1+ζj)with∣ζj∣≤(0.45(k+j)+1.07)φ−2k.(9)g\alpha^j = c\varphi^{2j}(1+\zeta_j) \qquad\text{with} \qquad|\zeta_j| \le(0.45(k+j)+1.07)\varphi^{-2k}. \tag*{(9)}

Proof. See the proof of Lemma 2.5 in [7]. □

Notations and terminology from algebraic number theory

Let η\eta be an algebraic number of degree dd with minimal primitive polynomial over the integers

a0xd+a1xd−1+⋯+ad=a0∏i=1d(x−η(i)),a_0x^d + a_1x^{d-1} + \cdots+ a_d = a_0\prod_{i=1}^{d}(x-\eta^{(i)}),

where the leading coefficient a0a_0 is positive and the η(i)\eta^{(i)}'s are the conjugates of η\eta. Then the logarithmic height of η\eta is given by

h(η):=1d(log⁡a0+∑i=1dlog⁡(max⁡{∣η(i)∣,1})).h(\eta) := \frac{1}{d}\left(\log a_0 + \sum_{i=1}^{d}\log\left(\max\{|\eta^{(i)}|,1\}\right)\right).

In particular, if η=p/q\eta= p/q is a rational number with gcd⁡(p,q)=1\gcd(p,q) = 1 and q>0q > 0, then h(η)=log⁡max⁡{∣p∣,q}h(\eta) = \log\max\{|p|,q\}. The following properties of h(⋅)h(\cdot) will be used without reference:

h(η±γ)≤h(η)+h(γ)+log⁡2,h(ηγ±1)≤h(η)+h(γ),h(ηs)=∣s∣h(η)(s∈Z).\begin{aligned} h(\eta\pm\gamma) &\le h(\eta) + h(\gamma) + \log2,\\ h(\eta\gamma^{\pm1}) &\le h(\eta) + h(\gamma),\\ h(\eta^s) &= |s|h(\eta) \qquad(s \in\mathbb{Z}). \end{aligned}

Since Ψk\Psi_k is monic, irreducible, and has exactly one root of modulus greater than 1, we have h(α)=(log⁡α)/kh(\alpha) = (\log\alpha)/k. For gg we use the bound of Bravo, Herrera and Luca [5 Lemmas 3 and 6]:

h(g)<H(k):=4log⁡φ+log⁡(k+1)(k≥2).(10)h(g) < H(k) := 4\log\varphi+ \log(k+1) \qquad(k \ge2). \tag*{(10)}

In Q(5)\mathbb{Q}(\sqrt{5}) we shall use h(2)=log⁡2h(2)=\log2, h(5)=12log⁡5h(\sqrt{5})=\frac{1}{2}\log5, h(φ)=12log⁡φh(\varphi)=\frac{1}{2}\log\varphi, h(2/5)=12log⁡5h(2/\sqrt{5})=\frac{1}{2}\log5 (minimal polynomial 5y2−45y^2-4), and h(2c)=12log⁡(5(1+1/5))=0.98953…h(2c)=\frac{1}{2}\log(5(1+1/\sqrt{5}))=0.98953\ldots, since 2c=1−1/52c=1-1/\sqrt{5} has minimal polynomial 5y2−10y+45y^2-10y+4.

Linear forms in logarithms and the reduction lemma

We use the following result of Matveev [11].

Theorem 2.5 (Matveev). Let γ1,…,γt\gamma_1,\ldots,\gamma_t be positive real algebraic numbers in a real algebraic number field K\mathbb{K} of degree DD, b1,…,btb_1,\ldots,b_t be nonzero integers, and assume that

Λ:=γ1b1⋯γtbt−1\Lambda:=\gamma_1^{b_1}\cdots\gamma_t^{b_t}-1

is nonzero. Then

log⁡∣Λ∣>−1.4×30t+3×t4.5×D2(1+log⁡D)(1+log⁡B)A1⋯At,\log|\Lambda|>-1.4\times30^{t+3}\times t^{4.5}\times D^2(1+\log D)(1+\log B)A_1\cdots A_t,

where B≥max⁡{∣b1∣,…,∣bt∣}B\ge\max\{|b_1|,\ldots,|b_t|\} and Ai≥max⁡{Dh(γi),∣log⁡γi∣,0.16}A_i\ge\max\{Dh(\gamma_i),|\log\gamma_i|,0.16\} for all i=1,…,ti=1,\ldots,t.

We write C(D,t):=1.4×30t+3×t4.5×D2(1+log⁡D)C(D,t):=1.4\times30^{t+3}\times t^{4.5}\times D^2(1+\log D).

For a real number rr we denote by ∥r∥\lVert r\rVert the distance from rr to the nearest integer. The following is the reduction lemma of Dujella and Pethő [6 Lemma 5(a)]; we state it with a real exponent ww, the proof being the same.

Lemma 2.6 (Dujella and Pethő). Let MM be a positive integer, let τ\tau be an irrational number and let p/qp/q be a convergent of the continued fraction of τ\tau with q>6Mq>6M. Let AA, BB, μ\mu be real numbers with A>0A>0 and B>1B>1, and put ε:=∥μq∥−M∥τq∥\varepsilon:=\lVert\mu q\rVert-M\lVert\tau q\rVert. If ε>0\varepsilon>0, then there is no solution of

0<∣uτ−v+μ∣<AB−w0<|u\tau-v+\mu|<AB^{-w}

in integers uu, vv and a real number ww with 1≤u≤M1\le u\le M and w≥log⁡(Aq/ε)/log⁡Bw\ge\log(Aq/\varepsilon)/\log B.

In practice we take the first convergent with q>6Mq>6M, and the next ones if ε≤0\varepsilon\le0; we never needed more than one extra convergent. We shall also use, without further mention, that if ∣Λ∣<1/2|\Lambda|<1/2 and Λ=eΓ−1\Lambda=e^\Gamma-1 with Γ\Gamma real, then ∣Γ∣<2∣Λ∣|\Gamma|<2|\Lambda|. When ∣Λ∣≥1/2|\Lambda|\ge1/2, the upper bound for ∣Λ∣|\Lambda| that we use already gives a bound for the exponent ww which is smaller than the bounds obtained with Lemma 2.6.

Elementary cases

Let (k,n,m,x)(k,n,m,x) be a solution of (2) in positive integers with k≥2k\ge2.

Lemma 3.1 (The case x=1x=1). If x=1x=1, then m=nm=n when k=2k=2, and (n,m)∈{(1,1),(2,2)}(n,m)\in\{(1,1),(2,2)\} when k≥3k\ge3.

Proof. For k=2k=2, Pn+1−Pn−1=2PnP_{n+1}-P_{n-1}=2P_n by (1), so Pm=PnP_m=P_n and m=nm=n. Let k≥3k\ge3. For n=1n=1, P2−P0=2=2P1P_2-P_0=2=2P_1, and for n=2n=2, P3−P1=4=2P2P_3-P_1=4=2P_2. Let n≥3n\ge3. Then

Pn+1−Pn−1=2Pn+Pn−2+Pn−3+⋯+Pn+1−k≥2Pn+Pn−2>2Pn,P_{n+1}-P_{n-1}=2P_n+P_{n-2}+P_{n-3}+\cdots+P_{n+1-k}\ge2P_n+P_{n-2}>2P_n,

since k≥3k\ge3 and Pn−2≥P1=1P_{n-2}\ge P_1=1, and Pn+1−Pn−1<2Pn+1P_{n+1}-P_{n-1}<2P_{n+1}. Since (Pr)r≥1(P_r)_{r\ge1} is strictly increasing, 2Pm2P_m cannot lie strictly between 2Pn2P_n and 2Pn+12P_{n+1}.

Lemma 3.2 (The case n=1n=1). If n=1n=1, then (m,x)∈{(1,1),(2,2)}(m,x)\in\{(1,1),(2,2)\}.

Proof. Since P2=2P_2=2 and P0=0P_0=0, (2) becomes Pm=2x−1P_m=2^{x-1}. For x=1x=1 we get m=1m=1; for x≥2x\ge2, Lemma 2.3 gives m=2m=2 and x=2x=2.

Lemma 3.3 (The case k=2k=2, x=2x=2). If k=2k=2 and x=2x=2, then m=2nm=2n.

Proof. By (1), Pn+12−Pn−12=2P2nP_{n+1}^2-P_{n-1}^2=2P_{2n}, so Pm=P2nP_m=P_{2n} and m=2nm=2n.

The triples listed in Theorem 1.1 are the solutions. This follows from (1) for k=2k=2, and from P2−P0=2P1P_2-P_0=2P_1, P22−P02=2P2P_2^2-P_0^2=2P_2 and P3−P1=2P2P_3-P_1=2P_2 for k≥3k\ge3. Therefore it remains to show that (2) has no solution with

n≥2,x≥2,(k,x)≠(2,2).(11)n\ge2,\qquad x\ge2,\qquad(k,x)\ne(2,2). \tag*{(11)}

Lemma 3.4 (Size of mm). If n≥2n \ge2 and x≥2x \ge2, then (n−1)x+1≤m≤nx+1(n-1)x+1 \le m \le nx+1. In particular m>xm>x, and the integer

j:=(n+1)x−mj := (n+1)x-m

satisfies x−1≤j≤2x−1x-1 \le j \le2x-1.

Proof. By Lemma 2.1(a), 2αm−2≤2Pm<Pn+1x≤αnx2\alpha^{m-2} \le2P_m < P_{n+1}^{x} \le\alpha^{nx}, so m−2<nxm-2<nx and m≤nx+1m \le nx+1. By (8), Pn−1x≤5−xPn+1xP_{n-1}^{x} \le5^{-x}P_{n+1}^{x}, hence 2αm−1≥2Pm≥(1−5−x)Pn+1x≥2425α(n−1)x2\alpha^{m-1} \ge2P_m \ge(1-5^{-x})P_{n+1}^{x} \ge\frac{24}{25}\alpha^{(n-1)x}. Thus

m−1≥(n−1)x−log⁡(25/12)log⁡α>(n−1)x−1m-1 \ge(n-1)x-\frac{\log(25/12)}{\log\alpha}>(n-1)x-1

because log⁡(25/12)<log⁡(1+2)\log(25/12)<\log(1+\sqrt{2}). Hence m≥(n−1)x+1m \ge(n-1)x+1.

Proof of Theorem 1.2

We begin with the case in which all the indices are at most k+1k+1, which reduces to (3). Everything takes place in Q(5)\mathbb{Q}(\sqrt{5}), the heights are small, and the argument shows on a simple example the steps that we shall repeat later for an arbitrary kk.

In this section (n,m,x)(n,m,x) is a solution of (3) with n≥2n \ge2 and x≥2x \ge2. We use Fr=(φr−(−φ)−r)/5F_r=(\varphi^r-(-\varphi)^{-r})/\sqrt{5} and φr−2≤Fr≤φr−1\varphi^{r-2}\le F_r\le\varphi^{r-1} for r≥1r\ge1.

Since Fr+4=3Fr+1+2Fr≥5FrF_{r+4}=3F_{r+1}+2F_r\ge5F_r, we have F2n−3≤F2n+1/5F_{2n-3}\le F_{2n+1}/5. Exactly as in Lemma 3.4 (with φ2\varphi^2 in place of α\alpha) we get (n−1)x+1≤m≤nx+1(n-1)x+1\le m\le nx+1. We put

e:=2m−1−(2n+1)x,e:=2m-1-(2n+1)x,

so that −3x+1≤e≤−x+1-3x+1\le e\le-x+1.

Write a:=φ−4n−2a:=\varphi^{-4n-2} and b:=φ−4n+6b:=\varphi^{-4n+6}, so that F2n+1=φ2n+15(1+a)F_{2n+1}=\frac{\varphi^{2n+1}}{\sqrt{5}}(1+a) and F2n−3=φ2n−35(1+b)F_{2n-3}=\frac{\varphi^{2n-3}}{\sqrt{5}}(1+b), and 2F2m−1=25(φ2m−1+φ−2m+1)2F_{2m-1}=\frac{2}{\sqrt{5}}(\varphi^{2m-1}+\varphi^{-2m+1}). We consider

Φ1:=25φ2m−1F2n+1−x−1,\Phi_1:=\frac{2}{\sqrt{5}}\varphi^{2m-1}F_{2n+1}^{-x}-1,
Φ2:=2⋅5 x−1φe−1,\Phi_2:=2\cdot\sqrt{5}^{\,x-1}\varphi^e-1,
Φ3:=2⋅5 x−1φe(1−φ−4x)−1−1.\Phi_3:=2\cdot\sqrt{5}^{\,x-1}\varphi^e(1-\varphi^{-4x})^{-1}-1.

Dividing (3) by F2n+1xF_{2n+1}^{x}, we get Φ1=−(F2n−3/F2n+1)x−25φ−2m+1F2n+1−x\Phi_1=-(F_{2n-3}/F_{2n+1})^x-\frac{2}{\sqrt{5}}\varphi^{-2m+1}F_{2n+1}^{-x}, hence

∣Φ1∣<1.4⋅5−x.(12)|\Phi_1|<1.4\cdot5^{-x}. \tag*{(12)}

Dividing (3) by (φ2n+1/5)x(\varphi^{2n+1}/\sqrt{5})^x, we get

Φ2=[(1+a)x−1]−φ−4x(1+b)x−r,r:=25φ−2m+1(5φ2n+1)x.\Phi_2=[(1+a)^x-1]-\varphi^{-4x}(1+b)^x-r,\qquad r:=\frac{2}{\sqrt{5}}\varphi^{-2m+1}\left(\frac{\sqrt{5}}{\varphi^{2n+1}}\right)^x.

Assume that

xφ−4n+6≤0.05.(13)x\varphi^{-4n+6}\le0.05. \tag*{(13)}

Then (1+a)x−1≤1.026xa(1+a)^x-1\le1.026xa, (1+b)x≤e0.05(1+b)^x\le e^{0.05}, and, using m≥(n−1)x+1m\ge(n-1)x+1, r≤0.014xar\le0.014xa. This gives

∣Φ2∣<2.7max⁡{xφ−4n−2,φ−4x}.(14)|\Phi_2|<2.7\max\{x\varphi^{-4n-2},\varphi^{-4x}\}. \tag*{(14)}

In the same way, the numerator of Φ3\Phi_3 is [(1+a)x−1]−φ−4x[(1+b)x−1]−r[(1+a)^x-1]-\varphi^{-4x}[(1+b)^x-1]-r. Note that φ−4xb≤a\varphi^{-4x}b\le a for x≥2x\ge2, and the denominator is 1−φ−4x≥1−φ−81-\varphi^{-4x}\ge1-\varphi^{-8}; hence, under (12a),

∣Φ3∣<2.66xφ−4n−2.(15)|\Phi_3|<2.66x\varphi^{-4n-2}. \tag*{(15)}

The norms from Q(5)\mathbb{Q}(\sqrt{5}) to Q\mathbb{Q} satisfy ∣N(φ)∣=1|N(\varphi)|=1, ∣N(5)∣=5|N(\sqrt{5})|=5, ∣N(2/5)∣=4/5|N(2/\sqrt{5})|=4/5. If Φ1=0\Phi_1=0, then ∣N(2φ2m−1/5)∣=4/5|N(2\varphi^{2m-1}/\sqrt{5})|=4/5 would be equal to F2n+12∈ZF_{2n+1}^{2}\in\mathbb{Z}, which is impossible. If Φ2=0\Phi_2=0, then 1=∣N(25 x−1φe)∣=4⋅5x−11=|N(2\sqrt{5}^{\,x-1}\varphi^e)|=4\cdot5^{x-1}, which is impossible. If Φ3=0\Phi_3=0, then

4⋅5x−1=∣N(1−φ−4x)∣=∣(1−φ−4x)(1−φ4x)∣=L4x−2=5F2x2,4\cdot5^{x-1}=|N(1-\varphi^{-4x})|=|(1-\varphi^{-4x})(1-\varphi^{4x})|=L_{4x}-2=5F_{2x}^{2},

that is, F2x2=4⋅5x−2F_{2x}^{2}=4\cdot5^{x-2}. This is false for x=2x=2 (F4=3F_4=3); for x≥3x\ge3 it would give 5∣F2x5\mid F_{2x}, hence 5∣x5\mid x, and then 2v5(x)=2v5(F2x)=x−22v_5(x)=2v_5(F_{2x})=x-2, which is impossible since x−2>2log⁡x/log⁡5x-2>2\log x/\log5 for x≥5x\ge5. Hence Φ1Φ2Φ3≠0\Phi_1\Phi_2\Phi_3\ne0.

We apply Theorem 2.5 in K=Q(5)\mathbb{K} = \mathbb{Q}(\sqrt{5}), D=2D = 2:

  • to Φ2\Phi_{2}, with t=3t = 3, (γ1,γ2,γ3)=(2,5,φ)(\gamma_{1}, \gamma_{2}, \gamma_{3}) = (2, \sqrt{5}, \varphi), (b1,b2,b3)=(1,x−1,e)(b_{1}, b_{2}, b_{3}) = (1, x - 1, e), B=3xB = 3x, and (A1,A2,A3)=(2log⁡2,log⁡5,log⁡φ)(A_{1}, A_{2}, A_{3}) = (2 \log2, \log5, \log\varphi). We get log⁡∣Φ2∣>−K2(1+log⁡3x)\log|\Phi_{2}| > -K_{2}(1 + \log3x), with

K2<1.0413×1012K_{2} < 1.0413 \times10^{12}
  • to Φ3\Phi_{3}, with t=4t = 4, the same data and γ4=1−φ−4x\gamma_{4} = 1 - \varphi^{-4x}, b4=−1b_{4} = -1, A4=log⁡(φ4x−1)<4xlog⁡φA_{4} = \log(\varphi^{4x} - 1) < 4x \log\varphi where the conjugate of γ4\gamma_{4} is 1−φ4x1 - \varphi^{4x}. More,

log⁡∣Φ3∣>−4xlog⁡φ⋅K3(1+log⁡3x)\log|\Phi_{3}| > -4x \log\varphi\cdot K_{3}(1 + \log3x)
K3:=C(2,4)<1.140×1014K_{3} := C(2, 4) < 1.140 \times10^{14}
  • to Φ1\Phi_{1}, with t=3t = 3, (γ1,γ2,γ3)=(2/5,φ,F2n+1)(\gamma_{1}, \gamma_{2}, \gamma_{3}) = (2/\sqrt{5}, \varphi, F_{2n+1}), (b1,b2,b3)=(1,2m−1,−x)(b_{1}, b_{2}, b_{3}) = (1, 2m - 1, -x), B=2mB = 2m, (A1,A2,A3)=(log⁡5,log⁡φ,2log⁡F2n+1)(A_{1}, A_{2}, A_{3}) = (\log5, \log\varphi, 2 \log F_{2n+1}) with 2log⁡F2n+1≤4nlog⁡φ2 \log F_{2n+1} \le4n \log\varphi where

log⁡∣Φ1∣>−4nlog⁡φ⋅K1(1+log⁡2m)\log|\Phi_{1}| > -4n \log\varphi\cdot K_{1}(1 + \log2m)
K1:=C(2,3)log⁡5log⁡φ<7.511×1011.K_{1} := C(2, 3) \log5 \log\varphi< 7.511 \times10^{11}.

Suppose first that (13) holds and that φ−4x≥xφ−4n−2\varphi^{-4x} \ge x\varphi^{-4n-2}. Then (14) and the bound for Φ2\Phi_{2} give 4xlog⁡φ<log⁡2.7+K2(1+log⁡3x)4x \log\varphi< \log2.7 + K_{2}(1 + \log3x), hence x<X0:=1.7634×1013x < X_{0} := 1.7634 \times10^{13}; then (15) and the bound for Φ3\Phi_{3} give n<6.553×1028n < 6.553 \times10^{28}. Otherwise, either (13) fails, and then 4n<6+log⁡(20x)/log⁡φ4n < 6 + \log(20x)/\log\varphi, or xφ−4n−2>φ−4xx\varphi^{-4n-2} > \varphi^{-4x}, and then (4n+2)log⁡φ<log⁡(2.7x)+K2(1+log⁡3x)(4n + 2)\log\varphi< \log(2.7x) + K_{2}(1 + \log3x). In both cases nn is bounded by an explicit function of xx; inserting this and m≤nx+1m \le nx + 1 in (12) and the bound for Φ1\Phi_{1}, we get x<3.0554×1027x < 3.0554 \times10^{27}. Therefore, in all cases,

x<3.06×1027(16)x < 3.06 \times10^{27} \tag*{(16)}

For the reduction phase, we put Γ2:=log⁡2+(x−1)log⁡5+elog⁡φ\Gamma_{2} := \log2 + (x - 1)\log\sqrt{5} + e\log\varphi, so that Φ2=eΓ2−1\Phi_{2} = e^{\Gamma_{2}} - 1. With τ:=log⁡5/log⁡φ\tau:= \log\sqrt{5}/\log\varphi and μ:=log⁡2/log⁡φ\mu:= \log2/\log\varphi, (14) gives (if ∣Φ2∣<1/2|\Phi_{2}| < 1/2)

0<∣(x−1)τ−(−e)+μ∣<5.4log⁡φe−w,w:=min⁡{4xlog⁡φ,(4n+2)log⁡φ−log⁡x}.0 < |(x - 1)\tau- (-e) + \mu| < \frac{5.4}{\log\varphi}e^{-w}, \qquad w := \min\{4x\log\varphi, (4n + 2)\log\varphi- \log x\}.

Lemma 2.6 with u=x−1≤Mu = x - 1 \le M gives w<Ww < W for an explicit WW. If w=4xlog⁡φw = 4x\log\varphi, then x<W/(4log⁡φ)x < W/(4\log\varphi), and for each such xx and each e∈[−3x+1,−x+1]e \in[-3x + 1, -x + 1] the number Φ3\Phi_{3} is explicit and non-zero, so that (15) bounds nn. If ww is the other term, or if (13) fails, nn is bounded directly. Once n≤Nn \le N, we apply Lemma 2.6 to Γ1/log⁡F2n+1\Gamma_{1}/\log F_{2n+1}, where

Γ1=log⁡(2/5)+(2m−1)log⁡φ−xlog⁡F2n+1\Gamma_{1} = \log(2/\sqrt{5}) + (2m - 1)\log\varphi- x\log F_{2n+1}

with u=2m−1≤2(NM+1)u = 2m - 1 \le2(NM + 1), A=2.8/log⁡F2n+1A = 2.8/\log F_{2n+1}, B=5B = 5, separately for each 2≤n≤N2 \le n \le N; this bounds xx. Table 1 shows the three passes we made.

passMMWWxx if w=4xlog⁡φw = 4x\log\varphinn from Φ3\Phi_{3}NNx≤x \le
13.06×10273.06 \times10^{27}69.18353.86846
24711.0052.377
387.4231.847

Table 1. Reduction for equation (3). Here MM is the bound for xx used in the pass, WW comes from Φ2\Phi_{2}, NN is the resulting bound for nn and the last column is the bound for xx given by Φ1\Phi_{1}.

As usual, almost all the work is done by the first pass, which brings the bound for xx from 3×10273 \times10^{27} down to 46. Thus n≤4n \le4 and x≤7x \le7. A direct search with 2≤n≤602 \le n \le60 and 2≤x≤602 \le x \le60 shows that (3) has no solution. This proves Theorem 1.2.

Linear forms in Q(α\alpha) and a bound for xx

The forms below are those of Section 4, with gg and α\alpha in the place of 1/51/\sqrt{5} and φ2\varphi^2. We then obtained that we work is now in a field of degree kk and that the height of gg grows like log⁡k\log k, so that the bound for xx will depend on kk.

In this section k≥2k \ge2 and (k,n,m,x)(k,n,m,x) is a solution of (2) satisfying (11). We use the notation of Lemma 2.1, and j=(n+1)x−mj=(n+1)x-m is as in Lemma 3.4. We consider

Λ1:=2gαmPn+1−x−1,Λ2:=2g1−xα−j−1,Λ3:=2g1−xα−j(1−α−2x)−1−1.\begin{aligned} \Lambda_1&:=2g\alpha^mP_{n+1}^{-x}-1,\\ \Lambda_2&:=2g^{1-x}\alpha^{-j}-1,\\ \Lambda_3&:=2g^{1-x}\alpha^{-j}(1-\alpha^{-2x})^{-1}-1. \end{aligned}

and the condition

36.4x≤αn−1.(17)36.4x\le\alpha^{n-1}. \tag*{(17)}

Lemma 5.1. (a) 0<∣Λ1∣<2⋅5−x0<|\Lambda_1|<2\cdot5^{-x}.

(b) If (15a) holds, then 0<∣Λ2∣<2.6max⁡{α−2x,xα−n}0<|\Lambda_2|<2.6\max\{\alpha^{-2x},x\alpha^{-n}\} and ∣Λ3∣<1.5xα−n|\Lambda_3|<1.5x\alpha^{-n}.

Proof. (a) Since 2gαm=2Pm−2em2g\alpha^m=2P_m-2e_m, dividing (2) by Pn+1xP_{n+1}^x gives

Λ1=−(Pn−1/Pn+1)x−2emPn+1−x.\Lambda_1=-(P_{n-1}/P_{n+1})^x-2e_mP_{n+1}^{-x}.

By (8) and Pn+1≥5P_{n+1}\ge5 we get ∣Λ1∣<2⋅5−x|\Lambda_1|<2\cdot5^{-x}. If Λ1=0\Lambda_1=0, then 2gαm=Pn+1x2g\alpha^m=P_{n+1}^x. Applying an automorphism of the Galois closure of Q(α)\mathbb{Q}(\alpha) sending α\alpha to α2\alpha_2 and using Lemma 2.2, we get

Pn+1x=2∣gk(α2)∣∣α2∣m<2,P_{n+1}^x=2|g_k(\alpha_2)||\alpha_2|^m<2,

a contradiction.

(b) Put U:=gαn+1U:=g\alpha^{n+1} and V:=gαn−1V:=g\alpha^{n-1}, and write Pn+1=U(1+ε1)P_{n+1}=U(1+\varepsilon_1), Pn−1=V(1+ε2)P_{n-1}=V(1+\varepsilon_2). By Lemma 2.1(c),(d), ∣ε1∣<1.812α−n−1|\varepsilon_1|<1.812\alpha^{-n-1} and ∣ε2∣<1.812α−n+1|\varepsilon_2|<1.812\alpha^{-n+1}, so that, by (15a), x∣ε2∣≤0.05x|\varepsilon_2|\le0.05; moreover (15a) implies αn−1≥72.8\alpha^{n-1}\ge72.8, hence n≥6n\ge6 and V>20V>20. Dividing (2) by UxU^x and using V/U=α−2V/U=\alpha^{-2},

Λ2=[(1+ε1)x−1]−α−2x(1+ε2)x−2emU−x.\Lambda_2=[(1+\varepsilon_1)^x-1]-\alpha^{-2x}(1+\varepsilon_2)^x-2e_mU^{-x}.

Now

∣(1+ε1)x−1∣≤ex∣ε1∣−1≤1.026x∣ε1∣<0.78xα−n|(1+\varepsilon_1)^x-1|\le e^{x|\varepsilon_1|}-1\le1.026x|\varepsilon_1|<0.78x\alpha^{-n}

as α≥1+2\alpha\ge1+\sqrt{2}, (1+∣ε2∣)x<1.052(1+|\varepsilon_2|)^x<1.052, and U−x=α−2xV−x<0.0025α−2xU^{-x}=\alpha^{-2x}V^{-x}<0.0025\alpha^{-2x}. This gives the bound for Λ2\Lambda_2. In the same way the numerator of Λ3\Lambda_3 is

[(1+ε1)x−1]−α−2x[(1+ε2)x−1]−2emU−x,[(1+\varepsilon_1)^x-1]-\alpha^{-2x}[(1+\varepsilon_2)^x-1]-2e_mU^{-x},

of absolute value less than 1.026x(∣ε1∣+α−2x∣ε2∣)+U−x<0.91xα−n1.026x(|\varepsilon_1|+\alpha^{-2x}|\varepsilon_2|)+U^{-x}<0.91x\alpha^{-n}, and its denominator is 1−α−2x≥1−α−4>0.971-\alpha^{-2x}\ge1-\alpha^{-4}>0.97; hence ∣Λ3∣<1.5xα−n|\Lambda_3|<1.5x\alpha^{-n}. Finally, if Λ2=0\Lambda_2=0, then gx−1αj=2g^{x-1}\alpha^j=2; conjugating as in (a) we get 2=∣gk(α2)∣x−1∣α2∣j<12=|g_k(\alpha_2)|^{x-1}|\alpha_2|^j<1, since j≥x−1≥1j\ge x-1\ge1, a contradiction. □

Lemma 5.2. (a) For k=2k=2, x≥2x\ge2 and x−1≤j≤2x−1x-1\le j\le2x-1, we have Λ3=0\Lambda_3=0 if and only if (x,j)=(2,2)(x,j)=(2,2).

(b) For 2≤k≤2572\le k\le257, 2≤x≤642\le x\le64, x−1≤j≤2x−1x-1\le j\le2x-1 and (k,x,j)≠(2,2,2)(k,x,j)\ne(2,2,2), we have ∣Λ3∣>1.22×10−3|\Lambda_3|>1.22\times10^{-3}.

Proof. (a) For k=2k=2 we have α=1+2\alpha=1+\sqrt{2} and g=1/(22)g=1/(2\sqrt{2}). For (x,j)=(2,2)(x,j)=(2,2),

2g−1α−2=42α−2=1−α−4,2g^{-1}\alpha^{-2}=4\sqrt{2}\alpha^{-2}=1-\alpha^{-4},

because α2−α−2=(α−α−1)(α+α−1)=2⋅22\alpha^2-\alpha^{-2}=(\alpha-\alpha^{-1})(\alpha+\alpha^{-1})=2\cdot2\sqrt{2}. This correspond to the identity

Pn+12−Pn−12=2P2nP_{n+1}^2-P_{n-1}^2=2P_{2n}

seen in Q(2)\mathbb{Q}(\sqrt{2}). Conversely, if Λ3=0\Lambda_3=0, taking norms from Q(2)\mathbb{Q}(\sqrt{2}) to Q\mathbb{Q} gives 23x−1=∣N(1−α−2x)∣=Q2x−22^{3x-1}=|N(1-\alpha^{-2x})|=Q_{2x}-2, where Qr=αr+(−α)−rQ_r=\alpha^r+(-\alpha)^{-r} are the Pell–Lucas numbers. For xx even, Q2x−2=8Px2Q_{2x}-2=8P_x^2, so PxP_x is a power of 2. Thus x=2x=2 by Lemma 2.3, and then j=2j=2. For xx odd, Q2x−2=Qx2Q_{2x}-2=Q_x^2, and QxQ_x is twice an odd number larger than 1 for x≥3x\ge3, which is impossible.

(b) This is a direct computation (Appendix A). The minimum is attained for k=4k=4, x=62x=62, j=80j=80, where ∣Λ3∣>1.22×10−3|\Lambda_3|>1.22\times10^{-3}. □

Part (a) explains the only zero we have got. Since (k,x)=(2,2)(k,x)=(2,2) is excluded by (11), it causes no problem. Part (b) covers all the pairs (x,j)(x,j) for which Λ3\Lambda_3 is used in Section 7, since there x≤64x\le64.

Now we can apply Matveev for Λ1\Lambda_1 and Λ2\Lambda_2. Let K=Q(α)K=\mathbb{Q}(\alpha), D=kD=k. For Λ1\Lambda_1 we take t=3t=3, (γ1,γ2,γ3)=(2g,α,Pn+1)(\gamma_1,\gamma_2,\gamma_3)=(2g,\alpha,P_{n+1}), (b1,b2,b3)=(1,m,−x)(b_1,b_2,b_3)=(1,m,-x) and B=mB=m (since m>xm>x). By (10) we may take A1=k(log⁡2+H(k))A_1=k(\log2+H(k)), A2=log⁡αA_2=\log\alpha and A3=klog⁡Pn+1≤knlog⁡αA_3=k\log P_{n+1}\le kn\log\alpha. For Λ2\Lambda_2 we take (γ1,γ2,γ3)=(2,g,α)(\gamma_1,\gamma_2,\gamma_3)=(2,g,\alpha), (b1,b2,b3)=(1,1−x,−j)(b_1,b_2,b_3)=(1,1-x,-j), B=2xB=2x, A1=k(log⁡2+H(k))A_1=k(\log2+H(k)), A2=kH(k)A_2=kH(k) and A3=log⁡αA_3=\log\alpha. Using log⁡α<2log⁡φ\log\alpha<2\log\varphi, we get

xlog⁡5−log⁡2<E1(k)n(1+log⁡m),(18)x\log5-\log2<E_1(k)n(1+\log m), \tag*{(18)}
min⁡{2xlog⁡α, nlog⁡α−log⁡x}−log⁡2.6<E2(k)(1+log⁡2x),(19)\min\{2x\log\alpha,\ n\log\alpha-\log x\}-\log2.6<E_2(k)(1+\log2x), \tag*{(19)}

where

E1(k):=C(k,3)k2(log⁡2+H(k))(2log⁡φ)2,E2(k):=C(k,3)k2(log⁡2+H(k))H(k) 2log⁡φ,E_1(k):=C(k,3)k^2(\log2+H(k))(2\log\varphi)^2,\qquad E_2(k):=C(k,3)k^2(\log2+H(k))H(k)\,2\log\varphi,

the second one under (15a).

Proposition 5.3. Let X(k)X(k) be the supremum of the real numbers x≥2x\ge2 such that either

2xlog⁡(1+2)<log⁡2.6+E2(k)(1+log⁡2x),2x\log(1+\sqrt{2})<\log2.6+E_2(k)(1+\log2x),

or

xlog⁡5<log⁡2+E1(k)Ns(x)(1+log⁡(Ns(x)x+1)),x\log5<\log2+E_1(k)N_s(x)(1+\log(N_s(x)x+1)),

where

Ns(x):=1+log⁡(36.4x)+log⁡2.6+E2(k)(1+log⁡2x)log⁡(1+2).N_s(x):=1+\frac{\log(36.4x)+\log2.6+E_2(k)(1+\log2x)}{\log(1+\sqrt{2})}.

Then every solution of (2) satisfying (11) has x<X(k)x<X(k).

Proof. If (15a) holds and the minimum in (17) is 2xlog⁡α2x\log\alpha, then xx satisfies the first inequality. Otherwise, either (15a) fails or the minimum is nlog⁡α−log⁡xn\log\alpha-\log x; in both cases n<Ns(x)n<N_s(x). Then m≤nx+1m\le nx+1 by Lemma 3.4, and (16) gives the second inequality. □\square

Large values of kk

The idea of this section is to exploit the error made by replacing α\alpha by φ2\varphi^2. This is of size about kφ−2kk\varphi^{-2k}, which is negligible as soon as kk is large, and the forms we obtain no longer depend on kk.

In this section k≥30k\ge30. Lemma 2.4 allows us to replace gg by c=1/(5φ)c=1/(\sqrt{5}\varphi) and α\alpha by φ2\varphi^2 in Λ1,Λ2,Λ3\Lambda_1,\Lambda_2,\Lambda_3. The new forms live in Q(5)\mathbb{Q}(\sqrt{5}), and their coefficients do not depend on kk. Since c1−x=(5φ)x−1c^{1-x}=(\sqrt{5}\varphi)^{x-1}, the forms are

L1:=2cφ2mF2n+1−x−1,L_1:=2c\varphi^{2m}F_{2n+1}^{-x}-1,
L2:=25 x−1φx−1−2j−1,L_2:=2\sqrt{5}^{\,x-1}\varphi^{x-1-2j}-1,
L3:=25 x−1φx−1−2j(1−φ−4x)−1−1.L_3:=2\sqrt{5}^{\,x-1}\varphi^{x-1-2j}(1-\varphi^{-4x})^{-1}-1.

Note that L2L_2 and L3L_3 are the forms Φ2\Phi_2 and Φ3\Phi_3 of § 4, with e=x−1−2je=x-1-2j. In particular they are non-zero, and Matveev gives the same lower bounds as for Φ2,Φ3\Phi_2,\Phi_3. Note that here ∣x−1−2j∣≤3x|x-1-2j|\le3x. The form L1L_1 is non-zero because ∣N(2cφ2m)∣=4/5|N(2c\varphi^{2m})|=4/5 is not an integer. Put

Ek(x):=1.011x(0.45k+2.02)φ−2k.E_k(x):=1.011x(0.45k+2.02)\varphi^{-2k}.

Lemma 6.1. Let k≥30k\ge30, and let (k,n,m,x)(k,n,m,x) be a solution of (2) which satisfies (11) and 2.022Ek(x)≤0.012.022E_k(x)\le0.01. Put λ∗:=−log⁡(2.022Ek(x))\lambda_*:=-\log(2.022E_k(x)).

(a) If (15a) holds, then ∣L2∣<e−λ2|L_2|<e^{-\lambda_2} and ∣L3∣<e−λ3|L_3|<e^{-\lambda_3}, where

λ2:=min⁡{2xlog⁡α−log⁡5.26, nlog⁡α−log⁡(5.26x), λ∗},\lambda_2:=\min\{2x\log\alpha-\log5.26,\ n\log\alpha-\log(5.26x),\ \lambda_*\},
λ3:=min⁡{nlog⁡α−log⁡(3.04x), λ∗}.\lambda_3:=\min\{n\log\alpha-\log(3.04x),\ \lambda_*\}.

:::`

(b) If n≤kn \le k and ∣ζm∣≤0.01|\zeta_m| \le0.01, then ∣L1∣<e−λ1|L_1| < e^{-\lambda_1}, where

λ1:=min⁡{xlog⁡5−log⁡4.1, −log⁡(2.1(0.45(k+m)+1.07)φ−2k)}.\lambda_1 := \min\left\{x\log5-\log4.1,\ -\log\left(2.1(0.45(k+m)+1.07)\varphi^{-2k}\right)\right\}.

Proof. (a) By Lemma 2.4, 2g1−xα−j=(L2+1)ρ2g^{1-x}\alpha^{-j}=(L_2+1)\rho with ρ=(1+η)1−x(1−δ)−j\rho=(1+\eta)^{1-x}(1-\delta)^{-j}, and

∣log⁡ρ∣≤1.0001((x−1)∣η∣+jδ)≤1.0001x(0.45k+1.965)φ−2k.|\log\rho| \le1.0001((x-1)|\eta|+j\delta)\le1.0001x(0.45k+1.965)\varphi^{-2k}.

We recall j≤2x−1j\le2x-1 and ∣ρ−1∣≤Ek(x)|\rho-1|\le E_k(x). Thus

∣L2∣=∣Λ2+1−ρ∣/ρ≤1.0102(∣Λ2∣+Ek(x))≤2.0204max⁡{∣Λ2∣,Ek(x)}|L_2|=|\Lambda_2+1-\rho|/\rho\le1.0102\left(|\Lambda_2|+E_k(x)\right)\le2.0204\max\{|\Lambda_2|,E_k(x)\}

and Lemma 5.1(b) gives the claim. For L3L_3 the argument is the same, with the additional factor (1−φ−4x)/(1−α−2x)(1-\varphi^{-4x})/(1-\alpha^{-2x}), which differs from 1 by less than 4xδφ−84x\delta\varphi^{-8} and is absorbed in Ek(x)E_k(x).

(b) If n≤kn\le k, then Pn+1=F2n+1P_{n+1}=F_{2n+1} by (4), and by (9), Λ1+1=(L1+1)(1+ζm)\Lambda_1+1=(L_1+1)(1+\zeta_m). Hence

∣L1∣≤(∣Λ1∣+∣ζm∣)/(1−∣ζm∣)≤2.0204max⁡{2⋅5−x,∣ζm∣}|L_1|\le(|\Lambda_1|+|\zeta_m|)/(1-|\zeta_m|)\le2.0204\max\{2\cdot5^{-x},|\zeta_m|\}

which gives the claim.

For L1L_1 we apply Theorem 2.5 in Q(5)\mathbb{Q}(\sqrt{5}) with (γ1,γ2,γ3)=(2c,φ,F2n+1)(\gamma_1,\gamma_2,\gamma_3)=(2c,\varphi,F_{2n+1}), (b1,b2,b3)=(1,2m,−x)(b_1,b_2,b_3)=(1,2m,-x), B=2mB=2m, (A1,A2,A3)=(log⁡(5+5),log⁡φ,2log⁡F2n+1)(A_1,A_2,A_3)=(\log(5+\sqrt{5}),\log\varphi,2\log F_{2n+1}) and we get

log⁡∣L1∣>−4nlog⁡φ⋅K1′(1+log⁡2m)\log|L_1|>-4n\log\varphi\cdot K'_1(1+\log2m)

with K1′:=C(2,3)log⁡(5+5)log⁡φ<9.236×1011K'_1:=C(2,3)\log(5+\sqrt{5})\log\varphi<9.236\times10^{11}.

Lemma 6.2. For every k≥258k\ge258 we have 2.022Ek(X(k))<10−552.022E_k(X(k))<10^{-55}.

Proof. Let k≥258k\ge258 and λ:=log⁡k\lambda:=\log k. We have C(k,3)k2=ck4(1+λ)C(k,3)k^2=ck^4(1+\lambda) with c:=1.4⋅306⋅34.5c:=1.4\cdot30^6\cdot3^{4.5}, and 2clog⁡φ<1.379⋅10112c\log\varphi<1.379\cdot10^{11}. Moreover H(k)≤λ+1.93H(k)\le\lambda+1.93, log⁡2+H(k)≤λ+2.627\log2+H(k)\le\lambda+2.627 and (1+λ)(λ+2.627)(λ+1.93)≤k1.1(1+\lambda)(\lambda+2.627)(\lambda+1.93)\le k^{1.1} (at k=258k=258 the left side is <401.2<401.2, the right side is >449>449, and k1.1k^{1.1} divided by the left side increases with kk). Since 2log⁡φ<H(k)2\log\varphi<H(k), we get

E1(k)<E2(k)≤M:=1.379⋅1011k5.1,M≥2.7⋅1023.E_1(k)<E_2(k)\le M:=1.379\cdot10^{11}k^{5.1},\qquad M\ge2.7\cdot10^{23}.

Put N(x):=1+(log⁡(94.64x)+M(1+log⁡2x))/log⁡(1+2)N(x):=1+(\log(94.64x)+M(1+\log2x))/\log(1+\sqrt{2}), and let x≥2x\ge2 satisfy one of the two inequalities of Proposition 5.3. Since 2log⁡(1+2)>log⁡52\log(1+\sqrt{2})>\log5, Ns(x)≤N(x)N_s(x)\le N(x) and E2(k)(1+log⁡2x)≤N(x)E_2(k)(1+\log2x)\le N(x), in both cases

xlog⁡5<log⁡2.6+MN(x)(1+log⁡(N(x)x+1)).x\log5<\log2.6+MN(x)(1+\log(N(x)x+1)).

Now let x≥M3x\ge M^3 and u:=log⁡xu:=\log x, so that u≥160u\ge160 and log⁡M≤u/3\log M\le u/3. Then N(x)≤1.135M(u+4.7)N(x)\le1.135M(u+4.7) and, since log⁡(u+4.7)≤0.035u\log(u+4.7)\le0.035u, 1+log⁡(N(x)x+1)≤1.377u1+\log(N(x)x+1)\le1.377u. Hence the right-hand side of the preceding inequality is at most 1.62M2u2≤1.62u2e2u/3<eulog⁡5=xlog⁡51.62M^2u^2\le1.62u^2e^{2u/3}<e^u\log5=x\log5, so the inequality fails. Therefore

X(k)≤M3<2.63⋅1033k15.3X(k)\le M^3<2.63\cdot10^{33}k^{15.3}

and

2.022Ek(X(k))≤G(k):=5.38⋅1033k15.3(0.45k+2.02)φ−2k.2.022E_k(X(k))\le G(k):=5.38\cdot10^{33}k^{15.3}(0.45k+2.02)\varphi^{-2k}.

Since 15.3/k+0.45/(0.45k+2.02)<2log⁡φ15.3/k+0.45/(0.45k+2.02)<2\log\varphi, the function GG is decreasing for k≥258k\ge258, and G(309)<10−55G(309)<10^{-55}. This proves the claim for k≥309k\ge309. For 258≤k≤308258\le k\le308 we computed X(k)X(k) directly. The largest value of log⁡10(2.022Ek(X(k)))\log_{10}(2.022E_k(X(k))) is −55.40-55.40, attained at k=258k=258.

Proposition 6.3. (2) has no solution satisfying (11) with k≥Kabs:=1.0427×1045k\ge K_{\mathrm{abs}}:=1.0427\times10^{45}.

Proof. Let k≥Kabsk\ge K_{\mathrm{abs}} and x<X:=X(k)x<X:=X(k); at k=Kabsk=K_{\mathrm{abs}} we have X<2.99×10398X<2.99\times10^{398}. By Lemma 6.2, Lemma 6.1 applies. Comparing with Matveev’s bound for L2L_2, the third term λ∗\lambda_* cannot be the minimum in λ2\lambda_2, because

2klog⁡φ>K2(1+log⁡3X)+log⁡(2.022⋅1.011X(0.45k+2.02)).2k\log\varphi>K_2(1+\log3X)+\log(2.022\cdot1.011X(0.45k+2.02)).

Hence, if (15a) holds, either 2xlog⁡α−log⁡5.26<K2(1+log⁡3x)2x\log\alpha-\log5.26<K_2(1+\log3x), which gives x<X0′=1.7634×1013x<X'_0=1.7634\times10^{13} as in Section 4, or

n<Nb:=(K2(1+log⁡3X)+log⁡(5.26X))/log⁡α>9.949×1014n<N_b:=(K_2(1+\log3X)+\log(5.26X))/\log\alpha>9.949\times10^{14}

if (15a) fails, n<1+log⁡(36.4X)/log⁡α<Nbn<1+\log(36.4X)/\log\alpha<N_b.

Case x<X0′x < X'_0. Then Matveev’s bound for L3L_3 gives, again because λ∗\lambda_* cannot be the minimum, n<Na:=1.32×1029n < N_a := 1.32 \times10^{29}. Hence n≤kn \le k and m≤nx+1<NaX0′+1<k+1m \le nx + 1 < N_aX'_0 + 1 < k + 1, so that all the terms in (2) are odd-indexed Fibonacci numbers by (4), and (2) becomes (3), which has no solution by Theorem 1.2.

Case n<Nbn < N_b. Then n≤kn \le k and m≤NbX+1m \le N_bX + 1, and ∣ζm∣|\zeta_m| is extremely small. Matveev’s bound for L1L_1 and Lemma 6.1(b) give x<XG=1.048×1030x < X_G = 1.048 \times10^{30}. Then m≤NbXG+1≤Kabs≤km \le N_bX_G + 1 \le K_{\mathrm{abs}} \le k. This is the condition which determines KabsK_{\mathrm{abs}}. We conclude with Theorem 1.2 as before.

All the inequalities used are of the form k>F(log⁡k)k > F(\log k), where FF is a polynomial with positive coefficients. Since they hold at k=Kabsk = K_{\mathrm{abs}}, they hold for all k≥Kabsk \ge K_{\mathrm{abs}}. We also checked them numerically for k=Kabs⋅10tk = K_{\mathrm{abs}} \cdot10^t, t=0,5,10,…,180t = 0, 5, 10, \ldots, 180.

We now lower KabsK_{\mathrm{abs}} by reduction. Let KK be an upper bound for kk, with K≥258K \ge258, and consider 258≤k≤K258 \le k \le K; then x<X(K)x < X(K) and, by Lemma 6.2, Lemma 6.1 applies. One round of reduction consists of the following steps, with log⁡α≥2log⁡φ+log⁡(1−φ−30)\log\alpha\ge2 \log\varphi+ \log(1 - \varphi^{-30}) and x≤Mx \le M:

(R1) Lemma 2.6 applied to L2L_2, with τ=log⁡5/log⁡φ\tau= \log\sqrt{5}/\log\varphi, μ=log⁡2/log⁡φ\mu= \log2/\log\varphi, u=x−1≤Mu = x - 1 \le M, A=2/log⁡φA = 2/\log\varphi, B=eB = e, gives λ2<W\lambda_2 < W. Hence either x≤Xa:=(W+log⁡5.26)/(2log⁡α)x \le X_a := (W + \log5.26)/(2\log\alpha), or n<(W+log⁡(5.26M))/log⁡αn < (W + \log(5.26M))/\log\alpha, or

k<Kc:=W+log⁡(2.022⋅1.011M(0.45K+2.02))2log⁡φk < K_c := \frac{W + \log(2.022 \cdot1.011M(0.45K + 2.02))}{2\log\varphi}

or (15a) fails.

(R2) For each 2≤x≤Xa2 \le x \le X_a and x−1≤j≤2x−1x - 1 \le j \le2x - 1 the number L3L_3 is explicit and non-zero; with ℓ:=−log⁡min⁡j∣L3∣\ell:= -\log\min_j |L_3|, Lemma 6.1(a) gives either n<(ℓ+log⁡(3.04x))/log⁡αn < (\ell+ \log(3.04x))/\log\alpha, or

2klog⁡φ<ℓ+log⁡(2.022⋅1.011x(0.45K+2.02)).2k\log\varphi< \ell+ \log(2.022 \cdot1.011x(0.45K + 2.02)).

(R3) Steps (R1)–(R2) give n≤Nn \le N unless kk is smaller than an explicit bound. For each 2≤n≤N2 \le n \le N, Lemma 2.6 applied to L1L_1, with τ=2log⁡φ/log⁡F2n+1\tau= 2\log\varphi/\log F_{2n+1}, μ=log⁡(2c)/log⁡F2n+1\mu= \log(2c)/\log F_{2n+1}, u=m≤NM+1u = m \le NM + 1, A=2/log⁡F2n+1A = 2/\log F_{2n+1}, B=eB = e, gives x≤XGx \le X_G unless kk is smaller than an explicit bound.

So, if kk is larger than a bound on K1K_1, let us say the maximum of the bounds met in (R1)–(R3) and of NN, we get n≤Nn \le N and x≤max⁡{Xa,XG}x \le\max\{X_a, X_G\}. We repeat (R1)–(R3) with this new bound for xx until it stabilizes, and we get final bounds n≤N′n \le N', x≤X′x \le X', valid for k>K2k > K_2. Finally, if k≥N′X′+1k \ge N'X' + 1, then n≤kn \le k and m≤N′X′+1≤k+1m \le N'X' + 1 \le k + 1, so that (2) becomes (3), which has no solution. Thus a round gives the new bound k≤max⁡{K1,K2,N′X′+1}k \le\max\{K_1, K_2, N'X' + 1\}.

Table 2. Rounds of reduction for large kk. Here KK is the bound for kk at the start of the round. For the first pass of each round we give ⌊log⁡10X(K)⌋\lfloor\log_{10} X(K)\rfloor, the value WW of (R1), XaX_a, the bound NN for nn, and the bound XGX_G of (R3). The last pass of every round gives W=6.40W = 6.40, N′=10N' = 10, X′=6X' = 6, hence N′X′+1=61N'X' + 1 = 61.

roundKKlog⁡10X(K)\log_{10} X(K)WWXaX_aNNXGX_Gnew KK
11.0427×10451.0427 \times10^{45}398923.6948019145802021
2202257138.327228392297
329850122.136424980261
426250120.606324679258
525950120.566324679257
625850120.556324679257

Table 2.

Proposition 6.4. Equation (2) has no solution satisfying (11) with k≥258k \ge258.

Proof. By Proposition 6.3 we may assume k<Kabsk < K_{\mathrm{abs}}. Table 2 shows the rounds. Each round starts from the bound given by the previous one plus one, and the last round gives again k≤257k \le257.

The cases 2≤k≤2572 \le k \le257

Now, it remains is a finite, but very long, computation. We have 256 values of kk, each with its own α\alpha, its own gg and a bound X(k)X(k) up to 105010^{50}. The steps are those of § 4, in Q(α)\mathbb{Q}(\alpha) instead of Q(5)\mathbb{Q}(\sqrt{5}).

Let 2≤k≤2572 \le k \le257. We start with x<X(k)≤1.08×1050x < X(k) \le1.08 \times10^{50} (Proposition 5.3) and make three passes of the following three steps, with x≤Mx \le M.

(S1) With Γ2=log⁡2+(1−x)log⁡g−jlog⁡α\Gamma_2 = \log2 + (1-x)\log g - j\log\alpha, Lemma 5.1(b) gives, under (15a),

0<∣xτ−j+μ∣<5.2log⁡αe−w,w=min⁡{2xlog⁡α,nlog⁡α−log⁡x},0 < \lvert x\tau-j+\mu\rvert< \frac{5.2}{\log\alpha}e^{-w}, \qquad w=\min\{2x\log\alpha,n\log\alpha-\log x\},

with τ=−log⁡g/log⁡α\tau=-\log g/\log\alpha and μ=log⁡(2g)/log⁡α\mu=\log(2g)/\log\alpha. Lemma 2.6 with u=x≤Mu=x\le M gives w<Ww<W. Hence either x<W/(2log⁡α)x<W/(2\log\alpha), or n<(W+log⁡M)/log⁡αn<(W+\log M)/\log\alpha, or (15a) fails and n<1+log⁡(36.4M)/log⁡αn<1+\log(36.4M)/\log\alpha.

(S2) For each 2≤x<W/(2log⁡α)2\le x<W/(2\log\alpha) and x−1≤j≤2x−1x-1\le j\le2x-1 we computed Λ3\Lambda_3 (it is nonzero, except for (k,x,j)=(2,2,2)(k,x,j)=(2,2,2), see Lemma 5.2) and n<log⁡(1.5x/∣Λ3∣)/log⁡αn<\log(1.5x/\lvert\Lambda_3\rvert)/\log\alpha from Lemma 5.1(b). Together with (S1) this gives n≤Nn\le N.

(S3) For each 2≤n≤N2\le n\le N, Lemma 2.6 applied to Γ1/log⁡Pn+1\Gamma_1/\log P_{n+1}, where Γ1=log⁡(2g)+mlog⁡α−xlog⁡Pn+1\Gamma_1=\log(2g)+m\log\alpha-x\log P_{n+1}, with τ=log⁡α/log⁡Pn+1\tau=\log\alpha/\log P_{n+1}, μ=log⁡(2g)/log⁡Pn+1\mu=\log(2g)/\log P_{n+1}, u=m≤NM+1u=m\le NM+1, A=4/log⁡Pn+1A=4/\log P_{n+1} and B=5B=5, gives a new bound for xx.

Proposition 7.1. Let 2≤k≤2572\le k\le257. Every solution of (2) with n≥2n\ge2 and x≥2x\ge2 satisfies n≤12n\le12 and x≤8x\le8. For k≥3k\ge3 there is no such solution, and for k=2k=2 all of them have x=2x=2 and m=2nm=2n.

Proof. The bounds follow from Table 3. For each kk, a search in exact integer arithmetic over 2≤n≤Nk2\le n\le N_k, 2≤x≤Xk2\le x\le X_k, where Nk≤12N_k\le12 and Xk≤8X_k\le8 are the final bounds of the third pass for this kk, comparing (Pn+1x−Pn−1x)/2(P_{n+1}^x-P_{n-1}^x)/2 with the list of the PmP_m, m≤NkXk+3m\le N_kX_k+3, gives the result. □

passMMWW in (S1)xx in (S2)NNxx from (S3)
11.08×10501.08 \times10^{50}124.456424882
28214.9971911
3119.635128

Table 3. Maximum over 2≤k≤2572 \le k \le257 of the quantities met in the three passes.

Proof of Theorem 1.1

Let (k,n,m,x)(k,n,m,x) be a solution of (2). If x=1x=1 or n=1n=1, or if k=2k=2 and x=2x=2, we conclude with Lemmas 3.1, 3.2 and 3.3. Otherwise (11) holds, that is, n≥2n\ge2, x≥2x\ge2 and (k,x)≠(2,2)(k,x)\ne(2,2). By Proposition 6.4 we have k≤257k\le257. By Proposition 7.1, every solution with 2≤k≤2572\le k\le257, n≥2n\ge2 and x≥2x\ge2 has k=2k=2 and x=2x=2, which is excluded by (11). Hence no solution satisfies (11), and this proves Theorem 1.1. □

Appendix A. Computational details

All computations were done in Python 3. Real numbers were handled with the decimal module; the precision was chosen in each step as twice the number of digits of the bound MM plus 60 to 120 digits, so that the products τq\tau q and μq\mu q in Lemma 2.6, where qq is about 6M⋅10206M\cdot10^{20}, are correct. The dominant root α\alpha was computed by Newton’s method on the equation x2−3x+1+x1−k=0x^2-3x+1+x^{1-k}=0 from (6), which is well conditioned for large kk, starting from φ2\varphi^2, and gg from (7). The kk-Pell numbers were computed exactly. In all the applications of Lemma 2.6 the first convergent with q>6Mq>6M or the next one gave ε>0\varepsilon>0.

As a consistency check, independent of the proof, we also verified by a direct search that, for 3≤k≤1003\le k\le100, 1≤n≤1501\le n\le150, 1≤x≤601\le x\le60, and for 2≤k≤602\le k\le60, 1≤n≤801\le n\le80, 1≤x≤301\le x\le30, the only solutions of (2) are those of Theorem 1.1. The minimum in Lemma 5.2(b) was computed with 150 significant digits.

For k≥30k \ge30 the minimum of ∣L3∣|L_3| over 2≤x≤4802 \le x \le480 and x−1≤j≤2x−1x-1 \le j \le2x-1 is greater than 8.43×10−48.43 \times10^{-4}, attained at (x,j)=(457,610)(x,j)=(457,610); so (R2)(R_2) never meets a small value.

Declarations

Data availability. The Python scripts used in this paper are available from the second author on request.

Use of AI tools. Claude (Anthropic) was used to find the error in an earlier version of this work, to write and run the Python code described in Appendix A. The authors take full responsibility for the content.

F. Faye, UFR SAT, Université Alioune Diop, Bambey, 30, Sénégal Email address: fallou [email protected]

B. Faye, UFR SAT, Université Alioune Diop, Bambey, 30, Sénégal Email address: [email protected]

References

  1. [2]H. Bensella, B. K. Patel, D. Behloul, On the exponential Diophantine equation (Fₘ₊₁⁽ᵏ⁾)ˣ − (Fₘ₋₁⁽ᵏ⁾)ˣ = Fₙ⁽ᵏ⁾, preprint (2022), arXiv:2205.13168.arxiv.org/abs/2205.13168
  2. [3]J. J. Bravo, J. L. Herrera, Repdigits in generalized Pell sequences, Arch. Math. (Brno) 56 (2020), 249–262.DOI
  3. [4]J. J. Bravo, J. L. Herrera, F. Luca, On a generalization of the Pell sequence, Math. Bohem. 146 (2021), no. 2, 199–213.DOI
  4. [5]J. J. Bravo, J. L. Herrera, F. Luca, Common values of generalized Fibonacci and Pell sequences, J. Number Theory 226 (2021), 51–71.DOI
  5. [6]A. Dujella, A. Pethő, A generalization of a theorem of Baker and Davenport, Quart. J. Math. Oxford Ser. (2) 49 (1998), no. 195, 291–306.DOI
  6. [7]F. Faye, B. Faye, On the exponential Diophantine equation (Pₙ⁽ᵏ⁾)ˣ + (Pₙ⁽ᵏ⁾)ˣ = Pₘ⁽ᵏ⁾, preprint (2026).
  7. [8]C. A. Gómez, J. C. Gómez, F. Luca, The complete solution of the Diophantine equation (Fₙ₊₁⁽ᵏ⁾)ˣ − (Fₙ₋₁⁽ᵏ⁾)ˣ = Fₘ⁽ᵏ⁾, Mediterr. J. Math. 21 (2024), Paper No. 13. https://doi.org/10.1007/s00009-023-02529-5
  8. [9]J. C. Gómez, F. Luca, C. A. Gómez, On the exponential Diophantine equation Fₙ₊₁ˣ − Fₙ₋₁ˣ = Fₘ, Taiwanese J. Math. 26 (2022), no. 4. https://doi.org/10.11650/tjm/220303
  9. [10]E. Kılıç, On the usual Fibonacci and generalized order-k Pell numbers, Ars Combin. 88 (2008), 33–45.
  10. [11]E. M. Matveev, An explicit lower bound for a homogeneous rational linear form in logarithms of algebraic numbers II, Izv. Ross. Akad. Nauk Ser. Mat. 64 (2000), no. 6, 125–180; English translation in Izv. Math. 64 (2000), no. 6, 1217–1269.
  11. [12]B. K. Patel, A. P. Chaves, On the exponential Diophantine equation Fₙ₊₁ˣ − Fₙ₋₁ˣ = Fₘ, Mediterr. J. Math. 18 (2021), Paper No. 187. https://doi.org/10.1007/s00009-021-01831-4

Paper details

vlnc:W_xkymxn72
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC0-1.0
Usage

views · 3

downloads · 0

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

  • Mathematics

Claude

View statement
Looking for related papers…
Contents