Abstract
Let \( (P_{n})_{n\ge 0} \) be the sequence of Padovan numbers defined by \( P_0=0 \), \( P_1 = P_2=1\), and \( P_{n+3}= P_{n+1} +P_n\) for all \( n\ge 0 \). In this paper, we find all positive square-free integers d such that the Pell equations \( x^2-dy^2 = N \) with \( N\in \{\pm 1, \pm 4\} \), have at least two positive integer solutions (x, y) and \((x^{\prime }, y^{\prime })\) such that both x and \(x^{\prime }\) are sums of two Padovan numbers.
Similar content being viewed by others
1 Introduction
Let \( (P_{n})_{n\ge 0} \) be the sequence of Padovan numbers defined by the linear recurrence
The Padovan sequence appears as sequence A000931 on the On-Line Encyclopedia of Integer Sequences (OEIS) [20]. The first few terms of this sequence are
Let \(d\ge 2\) be a positive square-free integer. It is well known that the Pell equations
and
have infinitely many positive integer solutions (x, y) and (X, Y) , respectively. By putting \((x_1, y_1)\) and \( (X_1, Y_1) \) for the smallest positive solutions to (1) and (2), respectively, all the solutions \( (x_k, y_k) \) and \( (X_k, Y_k) \) have the form
and
Furthermore, \( (x_k)_{k\ge 1} \) and \( (X_k)_{k\ge 1} \) are binary recurrent sequences. More exactly, the following formulae
and
hold for all positive integers k.
In the recent years, Luca et al. [17] considered the Diophantine equation
where \( x_k \) is given by (3) and \((T_n)_{n \ge 0}\) is the Tribonacci sequence defined by \(T_0 = 0\), \(T_1 =T_2= 1\), and \(T_{n+3} = T_{n+2} + T_{n+1}+T_{n}\) for all \(n \ge 0\). The Tribonacci sequence appears as sequence A000073 on the OEIS [20]. The authors in [17] proved that Eq. (5) has at most one solution (k, n) in positive integers for all d except for \(d=2\) when Eq. (5) has the three solutions \((k,n)=\{(1,1), (1,2), (3,5)\}\) and when \(d=3\) case in which Eq. (5) has the two solutions \((k,n)=\{(1,3), (2,5)\}\).
Inspired by the main result of Luca et al. [17], E. F. Bravo et al. [3, 4] studied the Diophantine equation
They proved that for each square-free integer \( d\ge 2 \), there is at most one positive integer k such that \( x_k \) admits the representation (6) for some nonnegative integers \( 0\le m\le n \), except for \( d\in \{2,3,5,15,26\} \). Furthermore, they explicitly stated all the solutions for these exceptional cases.
In the same spirit of the main result of Luca et al. [17], Rihane et al. [21] studied the Diophantine equations
where \( x_k \) and \( X_k \) are given by (3) and (4), respectively. They proved that for each square-free integer \( d\ge 2 \), there is at most one positive integer x participating in the Pell equation (1) and at most one positive integer X participating in the Pell equation (2) that is a Padovan number with a few exceptions of d that they effectively computed. Furthermore, the exceptional cases were \( d\in \{2,3,5,6\} \) and \( d=5 \) for the the first and second equations in (7), respectively. Several other related problems have been studied where \( x_k \) belongs to some interesting positive integer sequences. For example, see [8, 9, 11, 12, 14,15,16, 18].
2 Main results
In this paper, we study the same problem considered by E. F. Bravo et al.[3, 4] but with Padovan numbers instead of Tribonacci numbers. We also extend the results from the Pell equation (1) to the Pell equation (2). In both cases we find that there are only finitely many solutions that we effectively compute. Since \( P_1=P_2=P_3 = 1\), we discard the situations when \( n=1 \) and \( n=2 \) and just count the solutions for \( n=3 \). Similarly, \( P_4=P_5=2 \), so we just count the solutions for \( n=5 \).
The main aim of this paper is to prove the following results.
Theorem 1
For each square-free integer \( d\ge 2 \), there is at most one positive integer k such that
except when \( d\in \{2,3,6,15,110,483\} \) in the \( +1 \) case and \( d\in \{2,5,10,17\} \) in the \( -1 \) case.
Theorem 2
For each square-free integer \( d\ge 2 \), there is at most one positive integer k such that
except when \( d\in \{3,5,21\}\) in the \( +4 \) case and \( d\in \{2, 5\} \) in the \( -4 \) case.
For the exceptional values of d listed in Theorem 1 and Theorem 2, all solutions (k, n, m) are listed at the end of the proof of each result. The main tools used in this paper are lower bounds for nonzero linear forms in logarithms of algebraic numbers “á la Baker” and the Baker-Davenport reduction procedure, as well as the elementary properties of the Padovan sequence and solutions to Pell equations. Computations are done with the help of a computer program in Mathematica.
3 Preliminary results
3.1 The Padovan sequence
Here, we recall some important properties of the Padovan sequence \( (P_n)_{n\ge 0} \). The characteristic equation
has roots \( \alpha , \beta , \gamma = {\bar{\beta }} \), where
with
Furthermore, a Binet-like formula for Padovan numbers is given by
where
Numerically, the following estimates hold:
From (10), (11), and (14), it is easy to see that the contribution of the complex conjugate roots \( \beta \) and \( \gamma \), to the right-hand side of (12), is very small. More exactly, setting \( e(n):=P_n-a\alpha ^{n} \) and taking into account the facts that \( |\beta |=|\gamma |=\alpha ^{-\frac{1}{2}} \) and \( |b|=|c|< 0.29 \) (by (14)), it follows that, for any \( n\ge 1 \),
Finally, one can prove by induction that
3.2 Linear forms in logarithms
Let \( \eta \) be an algebraic number of degree d with minimal primitive polynomial over the integers
where the leading coefficient \( a_{0} \) is positive and the \( \eta ^{(i)} \)’s are the conjugates of \( \eta \). Then, the logarithmic height of \( \eta \) is given by
In particular, if \( \eta = p/q \) is a rational number with \( \gcd (p,q) = 1 \) and \( q>0 \), then \( h(\eta ) = \log \max \{|p|, q\} \). The following are some of the properties of the logarithmic height function \( h(\cdot ) \), which will be used in the next sections of this paper without reference:
We recall the result of Bugeaud et al. (see [6], Theorem 9.4), which is a modified version of the result of Matveev [19], which is one of our main tools in this paper.
Theorem 3
(Matveev according to Bugeaud et al., [6, 19]) Let \(\eta _1,\ldots ,\eta _t\) be nonzero elements of an algebraic number field \({\mathbb {K}} \subset {\mathbb {R}}\) of degree \(D_{\mathbb {K}}\) over \( {\mathbb {Q}} \), \(b_1,\ldots ,b_t\) be nonzero integers, and assume that
is nonzero. Then
where
and
3.3 Reduction procedure
During the calculations, we get upper bounds on our variables which are too large, thus we need to reduce them. To do so, we use some results from the theory of continued fractions.
For the treatment of linear forms homogeneous in two integer variables, we use the following well-known classical result in the theory of Diophantine approximation. For further details, we refer the reader to the books of Baker and Wüstholz [2] and Cohen [7].
Lemma 1
(Legendre, [2, 7]). Let \(\tau \) be an irrational number, \( {p_0}/{q_0}, {p_1}/{q_1}, {p_2}/{q_2}, \ldots \) be all the convergents of the continued fraction expansion of \( \tau \) and M be a positive integer. Let N be a nonnegative integer such that \( q_N> M \). Then putting \( a(M):=\max \{a_{i}: i=0, 1, 2, \ldots , N\} \), the inequality
holds for all pairs (r, s) of positive integers with \( 0<s<M \).
For a nonhomogeneous linear form in two integer variables, we use a slight variation of a result due to Dujella and Pethő (see [10], Lemma 5a) and itself is a generalization of a result of Baker and Davenport [1]. In this paper, we use an immediate variation of the result of Dujella and Pethő [10] due to J.J. Bravo et al. (see [5], Lemma 1). For a real number X, we write \(\Vert X \Vert := \min \{|X-n|: n\in {\mathbb {Z}}\}\) for the distance from X to the nearest integer.
Lemma 2
(Dujella, Pethő according to J. J. Bravo et al. [5, 10]) Let M be a positive integer, p/q be a convergent of the continued fraction expansion of the irrational number \(\tau \) such that \(q>6M\), and \(A,B,\mu \) be some real numbers with \(A>0\) and \(B>1\). Furthermore, let \(\varepsilon : = \Vert \mu q\Vert -M\Vert \tau q\Vert \). If \( \varepsilon > 0 \), then there is no solution to the inequality
in positive integers u, v, and w with
At various occasions, we need to find a lower bound for linear forms in logarithms with bounded integer coefficients in three and four variables. In this case, we use the LLL algorithm that we describe below. Let \( \tau _1, \tau _2, \ldots \tau _t \in {\mathbb {R}}\) and the linear form
We put \( X:=\max \{X_i\} \), \( C> (tX)^{t} \) and consider the integer lattice \( \Omega \) generated by
where C is a sufficiently large positive constant.
Lemma 3
(LLL-algorithm, [7]) Let \( X_1, X_2, \ldots , X_t \) be positive integers such that \( X:=\max \{X_i\} \) and \( C> (tX)^{t} \) be a fixed sufficiently large constant. With the above notation on the lattice \( \Omega \), we consider a reduced base \( \{\mathbf{b }_i \}\) to \( \Omega \) and its associated Gram-Schmidt orthogonalization base \( \{\mathbf{b }_i^*\}\). We set
If the integers \( x_i \) are such that \( |x_i|\le X_i \), for \( 1\le i \le t \) and \( \theta ^2\ge Q+R^2 \), then we have
For the proof and further details, we refer the reader to the book of Cohen (see [7], Proposition 2.3.20).
Finally, the following Lemma is also useful. It is Lemma 7 in [13].
Lemma 4
(Gúzman Sánchez, Luca, [13]) Let \( r, \, H,\) and L be positive real numbers. If \(r\ge 1\), \(H>(4r^2)^r\), and \(H>L/(\log L)^r\), then
4 Proof of Theorem 1
Let \( (x_1, y_1) \) be the smallest positive integer solution to the Pell equation (1). We put
From which we get that
Then,
Since \( \delta \ge 1+\sqrt{2} \), it follows that the estimate
We assume that \( (k_1, n_1, m_1) \) and \( (k_2, n_2, m_2) \) are triples of integers such that
We assume that \( 1\le k_1 < k_2 \). We also assume that \( 3\le m_i< n_i \) for \( i=1,2 \). We set \( (k,n,m):=(k_i, n_i, m_i) \), for \( i=1,2 \). Using the inequalities (16) and (20), we get from (21) that
The above inequalities give
Dividing through by \( \log \alpha \) and setting \( c_2:=1/\log \alpha \), we get that
and since \( \alpha ^3>2 \), we get
Furthermore, \( k<n \), for if not, we would then get that
which is false since \( \delta \ge 1+\sqrt{2} \), \(\alpha \in (1.32, 1.33) \) (by (14)), and \( n\ge 4 \).
Besides, given that \( k_1<k_2 \), we have by (16) and (21) that
Thus, we get that
4.1 An inequality for n and k
Using the Eqs. (8), (12), (19), and (21), we have
Therefore,
and by (15), we have
Thus, we have
Put
and
Since \(|\Lambda _1|=|e^{\Gamma _1}-1|<1/2\) for \( n\ge 4 \) (because \(1.5/\alpha ^{4}<1/2\)), and the inequality \(|y|<2|e^{y}-1|\) holds for all \( y\in \left( -1/2, 1/2\right) \), it follows that \( e^{|\Gamma _1|}<2 \) and so
Thus, we get that
We apply Theorem 3 on the left-hand side of (24) with the data:
Furthermore, we take the number field \( {\mathbb {K}}:={\mathbb {Q}}(\sqrt{d}, \alpha ) \) which has degree \( D_{{\mathbb {K}}}:=6 \). Since \( \max \{1,k,n\}\le n \), we take \(B:=n\). First, we note that the left-hand side of (24) is nonzero, since otherwise,
The left-hand side belongs to the quadratic field \( {\mathbb {Q}}(\sqrt{d}) \) while the right-hand side belongs to the cubic field \( {\mathbb {Q}}(\alpha ) \). These fields only intersect when both sides are rational numbers. Since \( \delta ^{k} \) is a positive algebraic integer and a unit, we get that to \( \delta ^{k} =1 \). Hence, \( k=0 \), which is a contradiction. Thus, \( \Lambda _1\ne 0 \) and we can apply Theorem 3.
We have \( h(\eta _1)=h(\delta )=(\log \delta )/2 \) and \( h(\eta _3)=h(\alpha )=(\log \alpha )/3 \). Furthermore,
The minimal polynomial of 2a is \( 23x^3-20x-8 \) and has roots 2a, 2b, 2c. Since \( 2|b|=2|c|<1 \) (by (14)), then
On the other hand,
Thus, we can take \( A_1:=3\log \delta \),
Now, Theorem 3 tells us that
Comparing the above inequality with (24), we get
Hence, we get that
We now return to the Diophantine equation (8) and rewrite it as
we obtain
Put
We assume for technical reasons that \( n-m\ge 10 \). Therefore, \( |e^{\Lambda _2}-1|<1/2 \). It follows that
Furthermore, \( \Lambda _2\ne 0 \) (so \(\Gamma _2 \ne 0\)), since \( \delta ^{k}\not \in {\mathbb {Q}}(\alpha ) \) by a previous argument.
We now apply Theorem 3 to the left-hand side of (27) with the data
Thus, we have the same \( A_1, \,A_2, \,A_3 \), as before. Then, by Theorem 3, we conclude that
By comparing with (27), we get
This was obtained under the assumption that \( n-m\ge 10 \), but if \( n-m<10\), then the above inequality also holds as well. We replace \( n-m \) in (26) by its upper bound that we obtained in (29) and use the fact that \( \delta ^{k}\le 2\alpha ^{n+3} \), to obtain bounds on n and k in terms of \( \log n \) and \(\log \delta \). We now record what we have proved so far.
Lemma 5
Let (k, n, m) be a solution to the Diophantine equation (8) with \( 3\le m<n \), then
4.2 Absolute bounds
We recall that \( (k,n,m)=(k_i,n_i, m_i) \), where \( 3\le m_i<n_i \), for \( i=1,2 \) and \( 1\le k_1<k_2 \). Furthermore, \( n_i\ge 4 \) for \( i=1,2 \). We return to (28) and write
We do a suitable cross product between \( \Gamma _2^{(1)}, \, \Gamma _2^{(2)} \) and \( k_1, k_2 \) to eliminate the term involving \( \log \delta \) in the above linear forms in logarithms:
where \(\displaystyle {\lambda :=\min _{1\le i\le 2} \{n_i-m_i\}.}\)
We need to find an upper bound for \( \lambda \). If \( 10n_2/\alpha ^{\lambda } > 1/2 \), we then get
Otherwise, \( |\Gamma _3|<1/2 \), so
We apply Theorem 3 with the data: \( t:=2 \), \( \eta _1 := 2a\), \( \eta _2:= \alpha \), \( b_1:=k_1-k_2 \), \( b_2:=k_1n_2-k_2n_1 \). We take the number field \( {\mathbb {K}}:={\mathbb {Q}}(\alpha ) \) and \( D_{{\mathbb {K}}}:=3 \). We begin by checking that \( e^{\Gamma _3}-1\ne 0 \) (so \( \Gamma _3\ne 0 \)). This is true, because \( \alpha \) and 2a are multiplicatively independent, since \( \alpha \) is a unit in the ring of integers \( {\mathbb {Q}}(\alpha ) \) while the norm of 2a is 8/23.
We note that \( |k_1-k_2|<k_2<n_2 \). Furthermore, from (31), we have
given that \( \lambda \ge 1 \). Therefore, we can take \( B:=11n_2 \). By Theorem 3, with the same \( A_1:=\log 23 \) and \( A_2:=\log \alpha \), we have that
By comparing this with (33), we get
Note that (34) is a better bound than (32), so (34) always holds. Without loss of generality, we can assume that \( \lambda = n_i-m_i \), for \( i=1,2 \) fixed.
We set \( \{i,j\}=\{1,2\} \) and return to (25) to replace \( (k,n,m)=(k_i,n_i, m_i) \):
and also return to (28), with \( (k, n,m)=(k_j, n_j, m_j) \):
We perform a cross product on (35) and (36) to eliminate the terms on \( \log \delta \):
with \( \nu :=\min \{n_i, n_j-m_j\} \). As before, we need to find an upper bound on \( \nu \). If \( 8n_2/\alpha ^{\nu }>1/2 \), then we get
Otherwise, \( |\Gamma _4|<1/2 \), so we have
To apply Theorem 3, first if \( e^{\Gamma _4}=1 \), we obtain
Since \( \alpha \) is a unit, the right-hand side in above is an algebraic integer. This is a contradiction, because \( k_1<k_2 \) so \( k_i-k_j\ne 0 \), and neither (2a) nor \( (2a)^{-1} \) are algebraic integers. Hence, \( e^{\Gamma _4}\ne 1 \). By assuming that \( \nu \ge 100 \), we apply Theorem 3 with the data:
and the inequalities (34) and (39). We get
The above inequality also holds when \( \nu <100 \). Furthermore, it also holds when the inequality (38) holds. Therefore, the above inequality holds in all cases. Note that the case \( (i, j) =(2, 1) \) leads to \( n_1-m_1\le n_1\le n_2+4 \) whereas \( (i, j) = (1,2) \) leads to \( \nu = \min \{n_1, n_2-m_2\} \). Hence, either the minimum is \( n_1 \), so
or the minimum is \( n_j-m_j \) and from the inequality (34), we get that
Next, we assume that we are in the case (41). We evaluate (35), for \( i=1,2 \) and make a suitable cross-product to eliminate the terms involving \( \log \delta \):
In the above inequality, we used the inequality (23) to conclude that \( \min \{n_1, n_2\}\ge n_1-4 \) as well as the fact that \( n_i\ge 4 \) for \( i=1.2 \). Next, we apply a linear form in four logarithms to obtain an upper bound to \( n_1 \). As in the previous calculations, we pass from (42) to
which is implied by (42) except if \( n_1 \) is very small, say
Thus, we assume that (44) does not hold, therefore (43) holds. Then, to apply Theorem 3, we first justify that \( e^{\Gamma _5}\ne 1 \). Otherwise,
By the fact that \( k_1<k_2 \), the norm \( \mathbf{N }_{{\mathbb {Q}}(\alpha )/{\mathbb {Q}}} (2a) ={8}/{23} \) and that \( \alpha \) is a unit, we have that 23 divides the norm \( \mathbf{N }_{{\mathbb {K}}/{\mathbb {Q}}} (1+\alpha ^{n_1-m_1}) \). The factorization of the ideal generated by 23 in \( {\mathcal {O}}_{{\mathbb {Q}}(\alpha )} \) is \( (23)={\mathfrak {p}}_1^2{\mathfrak {p}}_2 \), where \( {\mathfrak {p}}_1=(23, \,\alpha +13) \) and \( {\mathfrak {p}}_2=(23,\, \alpha +20) \). Hence, \( {\mathfrak {p}}_2 \) divides \( \alpha ^{n_1-m_1} + 1 \). Given that \( \alpha \equiv -20\, (\text {mod}\, {\mathfrak {p}}_2) \), then \( (-20)^{n_1-m_1}\equiv -1(\text {mod}\,{\mathfrak {p}}_2) \). Taking the norm \( \mathbf{N }_{{\mathbb {Q}}(\alpha )/{\mathbb {Q}}} \), we obtain that \( (-20)^{n_1-m_1} \equiv -1\,(\text {mod}\, 23) \). If \( n_1-m_1 \) is even, \( -1 \) is a quadratic residue modulo 23 and if \( n_1-m_1 \) is odd then 20 is a quadratic residue modulo 23. But, neither \( -1 \) nor 20 are quadratic residues modulo 23. Thus, \( e^{\Gamma _{5}}\ne 1 \).
Then, we apply Theorem 3 on the left-hand side of the inequality (43) with the data
Combining the right-hand side of (43) with the inequalities (34) and (41), Theorem 3 gives
In the above, we used the facts that
This was obtained under the assumption that the inequality (44) does not hold. If (44) holds, then so does (45). Thus, we have that inequality (45) holds provided that inequality (41) holds. Otherwise, inequality (40) holds which is a better bound than (45). Hence, we conclude that (45) holds in all possible cases.
By the inequality (22),
By substituting this into (30) we get \(n_2<4.64\times 10^{137}(\log n_2)^{10}\), and then, by Lemma 4, with the data \( r:=10, \,H:=4.64\times 10^{137}\) and \(L:=n_2 \), we get that \( n_2< 4.87\times 10^{165}\). This immediately gives that \( n_1<1.76\times 10^{63} \).
We record what we have proved.
Lemma 6
Let \((k_i, n_i, m_i)\) be a solution to the Diophantine equation (8), with \( 3\le m_i<n_i \) for \( i\in \{1,2\} \) and \( 1\le k_1<k_2 \), then
5 Reducing the bounds for \( n_1 \) and \( n_2 \)
In this section, we reduce the bounds for \( n_1 \) and \( n_2 \) given in Lemma 6 to cases that can be computationally treated. For this, we return to the inequalities for \( \Gamma _3 \), \( \Gamma _4 \), and \( \Gamma _5 \).
5.1 The first reduction
We divide both sides of the inequality (31) by \((k_2-k_1)\log \alpha \). We get that
We assume that \( \lambda \ge 10 \). Below we apply Lemma 1. We put \( \tau := \log (2a)/\log \alpha \), which is irrational and compute its continued fraction
and its convergents
Furthermore, we note that taking \( M:=4.87\times 10^{165} \) (by Lemma 6), it follows that
Thus, by Lemma 1, we have that
Hence, combining the inequalities (46) and (47), we obtain
so \( \lambda \le 2714 \). This was obtained under the assumption that \( \lambda \ge 10 \). Otherwise, \( \lambda<10<2714 \) holds as well.
Now, for each \( n_i-m_i = \lambda \in [1, 2714] \), we estimate a lower bound for \( |\Gamma _4| \), with
given in the inequality (37), via the procedure described in Sect. 3.3 (LLL-algorithm). We recall that \( \Gamma _4\ne 0 \). We apply Lemma 3 with the data:
We set \( X:= 5.4\times 10^{166} \) as an upper bound to \( |x_i|<11n_2 \) for all \( i=1, 2, 3 \), and \( C:=(20X)^{5} \). A computer search using Mathematica allows us to conclude, together with the inequality (37), that
which leads to \( \nu \le 6760 \). As we have noted before, \( \nu = n_1 \) (so \(n_1\le 6760\)) or \( \nu = n_j-m_j \).
Next, we suppose that \( n_j-m_j = \nu \le 6760 \). Since \( \lambda \le 2714 \), we have
Now, returning to the inequality (42) which involves
We use again the LLL algorithm to estimate the lower bound for \( |\Gamma _5| \) and thus, find a bound for \( n_1 \) that is better than the one given in Lemma 6.
We distinguish the cases \( \lambda < \chi \) and \( \lambda = \chi \).
5.2 The case \( \lambda < \chi \)
We take \( \lambda \in [1, 2714] \) and \( \chi \in [\lambda +1, 6760] \) and apply Lemma 3 with the data: \( t:=4 \),
We also put \( X:=5.4\times 10^{166} \) and \( C:=(20X)^{9} \). After a computer search in Mathematica together with the inequality (42), we can confirm that
This leads to the inequality
Substituting for the bound \( n_2 \) given in Lemma 6, we get that \( n_1\le 12172 \).
5.3 The case \( \lambda =\chi \)
In this case, we have
We divide through the inequality (42) by \((k_2-k_1)\log \alpha \) to obtain
We now put
and compute its continued fractions \( [a_0^{(\lambda )}, a_1^{(\lambda )} , a_2^{(\lambda )},\ldots ] \), and its convergents \([p_0^{(\lambda )}/q_0^{(\lambda )}, p_1^{(\lambda )}/q_1^{(\lambda )}, p_2^{(\lambda )}/q_2^{(\lambda )},\ldots ]\), for each \( \lambda \in [1, 2714] \). Furthermore, for each case we find an integer \( t_{\lambda } \) such that \( q_{t_{\lambda }}^{(\lambda )}>M:=4.87\times 10^{165}>n_2>k_2-k_1 \) and calculate
A computer search in Mathematica reveals that for \( \lambda = 321 \), \( t_{\lambda } = 330 \) and \( i=263 \), we have that \( a(M) = a_{321}^{(330)}=306269 \). Hence, combining the conclusion of Lemma 1 and the inequality (50), we get
so \( n_1\le 2730 \). Hence, we obtain that \( n_1\le 12172 \) holds in all cases (\(\nu =n_1\), \(\lambda < \chi \) or \(\lambda = \chi \)). By the inequality (22), we have that
By considering the second inequality in (30), we can conclude that \( n_2\le 9.9\times 10^{39}(\log n_2)^2 \), which immediately yields \( n_2<3.36\times 10^{44} \), by a simple application of Lemma 4. We summarise the first cycle of our reduction process as follows:
From the above inequalities, we note that the upper bound on \( n_2 \) represents a very good reduction of the bound given in Lemma 6. Hence, we expect that if we restart our reduction cycle with the new bound on \( n_2 \), then we get a better bound on \( n_1 \). Thus, we return to the inequality (46) and take \( M:=3.36\times 10^{44} \). A computer search in Mathematica reveals that
from which it follows that \( \lambda \le 752 \). We now return to (48) and we put \( X:=3.36\times 10^{44} \) and \( C:=(10X)^{5} \) and then apply the LLL algorithm in Lemma 3 to \( \lambda \in [1, 752] \). After a computer search, we get
then \( \nu \le 1846 \). By continuing under the assumption that \( n_j-m_j=\nu \le 1846 \), we return to (49) and put \( X:=3.36\times 10^{44} \), \( C:=(10X)^9 \) and \( M:=3.36\times 10^{44} \) for the case \( \lambda <\chi \) and \( \lambda = \chi \). After a computer search, we confirm that
gives \( n_1\le 3318 \), and \( a(M)=a_{175}^{(205)}=206961 \), which leads to \( n_1\le 772 \). Hence, in both cases \( n_1\le 3318 \) holds. This gives \( n_2\le 5\times 10^{42} \) by a similar procedure as before, and \( k_1\le 3125 \).
We record what we have proved.
Lemma 7
Let \( (k_i, n_i, m_i) \) be a solution to the Diophantine equation (8), with \( 3\le m_i<n_i \) for \( i=1,2 \) and \( 1\le k_1<k_2 \), then
5.4 The final reduction
Returning to (17) and (19) and using the fact that \( (x_1, y_1) \) is the smallest positive solution to the Pell equation (1), we obtain
Thus, we return to the Diophantine equation \( x_{k_1}=P_{n_1}+P_{m_1} \) and consider the equations
with \( k_1\in [1, 3125] \), \( m_1\in [3,3318 ] \) and \(n_1\in [m_1+1, 3318 ] \).
Besides the trivial case \( k_1=1 \), with the help of a computer search in Mathematica on the above equations in (51), we list the only nontrivial solutions in Table 1. We also note that \( 3+2\sqrt{2}=(1+\sqrt{2})^2 \), so these solutions come from the same Pell equation when \( d=2 \).
From Table 1, we set each \( \delta :=\delta _{t} \) for \( t=1, 2, \ldots , 17 \). We then work on the linear forms in logarithms \( \Gamma _1 \) and \( \Gamma _2 \), to reduce the bound on \( n_2 \) given in Lemma 7. From the inequality (28), for \( (k,n,m):=(k_2, n_2, m_2) \), we write
We put
We note that \( \tau _t \) is transcendental by Gelfond-Schneider Theorem (see [2], Theorem 2.1). Thus, \( \tau _t \) is irrational. We can rewrite the inequality (52) as
We take \( M:= 5\times 10^{42} \) which is the upper bound on \( n_2 \) according to Lemma 7 and apply Lemma 2 to the inequality (53). As before, for each \( \tau _t \) with \( t=1, 2, \ldots , 17 \), we compute its continued fraction \( [a_0^{(t)}, a_1^{(t)}, a_2^{(t)}, \ldots ] \) and its convergents \( p_0^{(t)}/q_0^{(t)}, p_1^{(t)}/q_1^{(t)}, p_2^{(t)}/q_2^{(t)}, \ldots \). For each case, by means of a computer search in Mathematica, we find an integer \(s_{t}\) such that
We finally compute all the values of \( b_t:=\lfloor \log (A_t q^{(t)}_{s_t}/\epsilon _t)/\log B_t \rfloor \). The values of \( b_t \) correspond to the upper bounds on \( n_2-m_2 \), for each \( t=1, 2, \ldots , 17 \), according to Lemma 2. The results of the computation for each t are recorded in Table 2.
By replacing \( (k, n, m):=(k_2, n_2, m_2) \) in the inequality (25), we can write
We now put
With the above notations, we can rewrite the inequality (54) as
We again apply Lemma 2 to the inequality (55), for
We take
and
With the help of Mathematica, we obtain the results in Table 3.
Thus, \(\max \{b_{t, n_2-m_2}: t=1, 2, \ldots , 17\) and \(n_2-m_2 = 1, 2, \ldots b_t\} \le 408\). So, by Lemma 2, we have that \( n_2\le 408 \), for all \( t=1,2, \ldots , 17 \), and by the inequality (23) we have that \( n_1\le n_2+4 \). From the fact that \( \delta ^{k}\le 2\alpha ^{n+3} \), we can conclude that \( k_1< k_2\le 133 \). Collecting everything together, our problem is reduced to search for the solutions for (21) in the following range
After a computer search for the solutions to the Diophantine equations in (21) on the range above, we obtained the following solutions, which are the only solutions for the exceptional d cases we have stated in Theorem 1:
For the \( +1 \) case:
For the \( -1 \) case:
This completes the proof of Theorem 1.
6 Proof of Theorem 2
The proof of Theorem 2 follows from similar steps, techniques, and arguments as given in the proof of Theorem 1. So, we do not give the details here. Below, we give the solutions to the Diophantine equation (9) for the exceptional d cases stated in Theorem 2.
For the \( +4 \) case:
For the \( -4 \) case:
References
Baker, A., Davenport, H.: The equations \(3x^2 - 2= y^2\) and \(8x^2 - 7= z^2\). Q. J. Math. Oxford Ser. (2) 20(1), 129–137 (1969)
Baker, A., Wüstholz, G.: Logarithmic Forms and Diophantine Geometry. Cambridge University Press, Cambridge (2008)
Bravo, E.F., Gómez, C.A., Luca, F.: \(X\)-coordinates of Pell equations as sums of two tribonacci numbers. Period. Math. Hungar. 77(2), 175–190 (2018)
Bravo, E.F., Gómez, C.A., Luca, F.: Correction to: \(X\)-coordinates of Pell equations as sums of two tribonacci numbers. Period. Math. Hungar. 80(1), 145–146 (2020)
Bravo, J.J., Gómez, C.A., Luca, F.: Powers of two as sums of two \(k\)-Fibonacci numbers. Miskolc Math. Notes 17(1), 85–100 (2016)
Bugeaud, Y., Mignotte, M., Siksek, S.: Classical Classical and modular approaches to exponential Diophantine equations I. Fibonacci and Lucas perfect powers. Ann. Math. (2) 163(2), 969–1018 (2006)
Cohen, H.: Number Theory. Volume I: Tools and Diophantine Equations, Graduate Texts in Mathematics 239. Springer, Berlin (2007)
Ddamulira, M., Luca, F.: On the \( x \)-coordinates of Pell equations which are \(k\)-generalized Fibonacci numbers. J. Numb. Theory 207, 156–195 (2020)
Dossavi-Yovo, A., Luca, F., Togbé, A.: On the \(X\)-coordinates of Pell equations which are rep-digits. Publ. Math. Debrecen 88, 381–399 (2016)
Dujella, A., Pethő, A.: A generalization of a theorem of Baker and Davenport. Q. J. Math. Oxford Ser. (2) 49(195), 291–306 (1998)
Faye, B., Luca, F.: On the \(X\)-coordinates of Pell equations which are repdigits. Fibonacci Q. 56(1), 52–62 (2018)
Gómez, C.A., Luca, F.: Zeckendorf representations with at most two terms to \(x\)-coordinates of Pell equations. Sci. China Math. 64(4), 627–642 (2020)
Gúzman Sánchez, S., Luca, F.: Linear combinations of factorials and \(s\)-units in a binary recurrence sequence. Ann. Math. Qué. 38(2), 169–188 (2014)
Kafle, B., Luca, F., Togbé, A.: On the \(x\)-coordinates of Pell equations which are Fibonacci numbers II. Colloq. Math. 149(1), 75–85 (2017)
Kafle, B., Luca, F., Togbé, A.: \(x\)-Coordinates of Pell equations which are Tribonacci numbers II. Period. Math. Hungar. 79(2), 157–167 (2019)
Kafle, B., Luca, F., Togbé, A.: \(X\)-coordinates of Pell equations which are Lucas numbers. Bol. Soc. Mat. Mex. (3) 25(3), 481–493 (2019)
Luca, F., Montejano, A., Szalay, L., Togbé, A.: On the \( x \)-coordinates of Pell equations which are Tribonacci numbers. Acta Arith. 179(1), 25–35 (2017)
Luca, F., Togbé, A.: On the \( x \)-coordinates of Pell equations which are Fibonacci numbers. Math. Scand. 122(1), 18–30 (2018)
Matveev, E.M.: An explicit lower bound for a homogeneous rational linear form in the logarithms of algebraic numbers II, Izv. Ross. Akad. Nauk Ser. Mat. 64 (6) (2000), 125–180, In Russian. English translation in Izv. Math. 64 (6) (2000), 1217–1269
OEIS Foundation Inc. The On-Line Encyclopedia of Integer Sequences (2020). https://oeis.org/
Rihane, S.S., Hernane, M.O., Togbé, A.: The \( x \)-coordinates of Pell equations and Padovan numbers. Turk. J. Math. 43(1), 207–223 (2019)
Acknowledgements
The author thanks the anonymous referees and the editor for the careful reading of the manuscript and the useful comments and suggestions that greatly improved the quality of presentation of this paper. Part of the work in this paper was done when the author visited the Institut de Mathématiques de Bordeaux, Université de Bordeaux, in May 2019. He thanks this institution for its hospitality and the fruitful working environment.
Funding
Open Access funding provided by Austrian Science Fund (FWF).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they do not have conflict of interests.
Ethical standard
This research complies with ethical standards.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supported by the Austrian Science Fund (FWF) projects: F5510-N26—Part of the special research program (SFB), “Quasi-Monte Carlo Methods: Theory and Applications” and W1230—“Doctoral Program Discrete Mathematics”.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Ddamulira, M. On the x-coordinates of Pell equations that are sums of two Padovan numbers. Bol. Soc. Mat. Mex. 27, 4 (2021). https://doi.org/10.1007/s40590-021-00312-8
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s40590-021-00312-8