Abstract
In this paper, we find all Pell and Pell–Lucas numbers expressible as sums of two base 10 repdigits.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Let \(g\ge 2\) be an integer. A natural number N is called a base g repdigit if all of its base g-digits are equal; that is, if
When \(g=10\), we omit the base and simply say that N is a repdigit. Diophantine equations involving repdigits were also considered in several papers which found all repdigits which are perfect powers, or Fibonacci numbers, or generalized Fibonacci numbers, and so on (see [1, 2, 4, 9, 11,12,13] for a sample of such results).
Let \(\{P_m\}_{m\ge 0}\) be the Pell sequence given by
for \(m\ge 0\), where \(P_0 = 0\) and \(P_1 = 1\). A few terms of this sequence are
We let \(\{Q_m\}_{m\ge 0}\) be the companion Lucas sequence of the Pell sequence given by \(Q_{m+2} = 2Q_{m+1} + Q_m\), for \(m\ge 0\), where \(Q_0 = 2\) and \(Q_1 = 2\). Its first few terms are
In [8], it was shown that there are no Pell or Pell–Lucas numbers larger than 10 with only one distinct digit.
Here, we extend this and prove the following results.
Theorem 1.1
The largest Pell number which is a sum of two repdigits is
Theorem 1.2
The largest Pell–Lucas number which is a sum of two repdigits is
We organize this paper as follows: In Sect. 2, we recall some elementary properties of Pell and Pell–Lucas numbers, a result due to Matveev concerning a lower bound for a linear form in logarithms of algebraic numbers, as well as a variant of a reduction result due to Baker and Davenport reduction. The proofs of Theorems 1.1 and 1.2 are achieved in Sects. 3, 4, respectively. We start with some elementary considerations.
2 Preliminaries
2.1 Some Properties of Pell and Pell–Lucas Numbers
In this subsection, we recall some properties of Pell sequence. Binet’s formula says that
for all \(m \ge 0\), where \(\alpha = 1 + \sqrt{2}\) and \(\beta = 1 - \sqrt{2}\) are the two roots of the characteristic equation \( x^2 - 2x - 1 = 0 \) of the Pell sequence.
The sequence of Pell–Lucas numbers \(\{Q_m\}_{m\ge 0}\) starts with \(Q_0 = 2, \; Q_1 = 2\) and obeys the same recurrence relation
for all \(m \ge 0\) as Pell sequence. Its Binet formula is
2.2 Linear Forms in Logarithms
We need some results from the theory of lower bounds for nonzero linear forms in logarithms of algebraic numbers. We start by recalling Theorem 9.4 of [3], which is a modified version of a result of Matveev [14]. Let \(\mathbb {L}\) be an algebraic number field of degree \(d_{\mathbb {L}}\). Let \(\eta _1, \eta _2, \ldots , \eta _l \in \mathbb {L}\) not 0 or 1 and \(d_1, \ldots , d_l\) be nonzero integers. We put
and
Let \(A_1, \ldots , A_l\) be positive integers such that
where for an algebraic number \(\eta \) of minimal polynomial
over the integers with positive \(a_0\), we write \(h(\eta )\) for its Weil height given by
The following consequence of Matveev’s theorem is Theorem 9.4 in [3].
Theorem 2.1
If \(\Gamma \ne 0\) and \(\mathbb {L}\subseteq \mathbb {R}\), then
2.3 The Baker–Davenport Lemma
Here, we present a variant of the reduction method of Baker and Davenport due to de Weger [6].
Let \(\vartheta _1,\vartheta _2,\beta \in \mathbb {R}\) be given, and let \(x_1,x_2\in \mathbb {Z}\) be unknowns. Let
Let c, \(\delta \) be positive constants. Set \(X=\max \{|x_1|,|x_2|\}\). Let \(X_0, Y\) be positive. Assume that
When \(\beta =0\) in (7), we get
Put \(\vartheta =-{\vartheta _1}/{\vartheta _2}\). We assume that \(x_1\) and \(x_2\) are coprime. Let the continued fraction expansion of \(\vartheta \) be given by
and let the kth convergent of \(\vartheta \) be \({p_k}/{q_k}\) for \(k=0,1,2,\ldots \). We may assume without loss of generality that \(|\vartheta _1|<|\vartheta _2|\) and that \(x_1>0\). We have the following results.
Lemma 2.1
(See Lemma 3.2 in [6]) Let
If (8) and (9) hold for \(x_1\), \(x_2\) and \(\beta =0\), then
When \(\beta \ne 0\) in (7), put \(\vartheta =-{\vartheta _1}/{\vartheta _2}\) and \(\psi ={\beta }/{\vartheta _2}\). Then we have
Let p / q be a convergent of \(\vartheta \) with \(q>X_0\). For a real number x we let \(\Vert x\Vert =\min \{|x-n|, n\in {\mathbb Z}\}\) be the distance from x to the nearest integer. We have the following result.
Lemma 2.2
(See Lemma 3.3 in [6]) Suppose that
Then, the solutions of (8) and (9) satisfy
3 The Proof of Theorem 1.1
3.1 An Elementary Estimate
We assume that
for some integers \(m_1 \le m_2\) and \(d_1, d_2 \in \{1, 2, \ldots , 9\}.\) A quick computation with Maple reveals no solutions in the interval \( n \in [7, 1000].\) So, from now on, we may assume that \(n > 1000.\) For this computation, we first note that \(P_{1000}\) has 383 digits. Thus, we generated the list of all repdigits with at most 383 digits; let us call it \(\mathcal {A}.\) Then, for every \(n \in [7, 1000],\) we computed \(M := \lfloor {\log P_n}/ {\log {10}} \rfloor + 1\) (the number of digits of \(P_n\)) and then checked whether \( P_n - d(10^m - 1)/9\) is a member of \(\mathcal {A}\), for some digit \(d \in \{1, \ldots , 9\}\) and some \(m \in \{M-1, M\}.\) This computation took a few minutes.
Lemma 3.1
All solutions of Eq. (11) satisfy
Proof
The proof follows easily from the fact that \(\alpha ^{n-2}< P_n < \alpha ^{n-1},\) for \(n\ge 2\). One can see that
Taking the logarithm of all sides, we get \( (n-2) \log \alpha < \log 2 + m_2 \log 10,\) which leads to
The lower bound follows similarly. \(\square \)
We first return to Eq. (11) and use the Binet formula (4) to get
i.e.,
We examine (12) in two different steps as follows.
Step 1 Equation (12) gives
which we rewrite as
Thus, dividing both sides by \(d_2 10^{m_2},\) we get
Put
We compare this upper bound with the lower bound on the quantity \(\Gamma \) given by Theorem 2.1. Observe first that \(\Gamma \) is not zero, for if it were, then \(\sqrt{2}=q\alpha ^n,\) with some \( q \in \mathbb {Q},\) and hence \(\alpha ^{2n}=2q^{-2} \in \mathbb {Q},\) which is false for any \(n > 0\). We take
where \( \eta _1 ,\; \eta _2 ,\; \eta _3 \in \mathbb {Q}(\sqrt{2})\) and \( d_1 ,\; d_2 ,\; d_3 \in \mathbb {Z}.\) The degree of \({\mathbb L}:=\mathbb {Q}(\sqrt{2})\) is \(d_{\mathbb {L}} = 2.\)
Since \(10^{m_2-1}< P_n < \alpha ^{n-1},\) we have that \(m_2 < n\). Therefore, we can take \( D = n.\) We note also that the conjugates of \(\eta _1, \; \eta _2,\) and \(\eta _3\) are \(\eta _1' = -\eta _1, \; \eta _2' = \beta , \; \eta _3' = \eta _3.\) Furthermore, \(\eta _1, \; \eta _2\) are algebraic integers, while the minimal polynomial of \(\eta _1\) over \(\mathbb {Q}\) is
Hence, we get
which implies that
Thus, we can take
Clearly,
We have
Theorem 2.1 tells us that
Comparing this last inequality with (13) leads to
giving
Thus, we obtain
Step 2 Equation (12) becomes
which we rewrite as
Thus, dividing both sides by \(\alpha ^n/{2 \sqrt{2}},\) we get
Put
Suppose that \(\Gamma ' = 0.\) We then have
Conjugating in \( \mathbb {Q}(\sqrt{2})\), we get
Consequently, we obtain
which leads to a contradiction as \(m_1 \ge 1.\) Thus, \(\Gamma '\ne 0\). With the notations of Theorem 2.1, we take
where \( \eta _1 ,\; \eta _2 ,\; \eta _3 \in \mathbb {Q}(\sqrt{2})\) and \( d_1 ,\; d_2 ,\; d_3 \in \mathbb {Z}.\) The degree of \({\mathbb L}:=\mathbb {Q}(\sqrt{2})\) is \(d_{\mathbb {L}} = 2\).
Since \(1 \le m_1 \le m_2\) and \(m_2 < n\), we can take \( D = n.\) We now discuss the \( A_j\)’s for \(j=1,2,3\). We have
i.e.,
Thus, we can take
Clearly, we get
We have
Theorem 2.1 implies that
Comparing this last inequality with (16) leads to
Hence, using inequality (15), we obtain
The above inequality gives us
Lemma 3.1 implies
We summarize what we have proved so far in the following lemma.
Lemma 3.2
All solutions of Eq. (11) satisfy
3.2 Bound Reduction
To lower this bound, we return to Eq. (11). We rewrite it into the form
Observe that the term in parentheses is always positive or zero and is zero only when \( d_1 = m_1 = 1\) and \( d_2 = 9\). In this last case, we get \(P_n = 10^{m_2},\) but such a relation is not possible for \(n > 1000,\) because by the primitive divisor theorem (see [5]), the Pell number \(P_n\) has a prime factor \(\ge n-1\), for all values of \(n > 12.\) Thus, the number appearing in parentheses is \(\ge 1/9.\) Hence,
Hence, from the above notations given by (13) and (14), we have
Put
We obtain that
so
which implies that
Thus, we see that
holds with \(Y:=m_2 - m_1< m_2 < 1.2 \cdot 10^{30}.\)
We also have
Thus, we take
The smallest value of \(q > X_0\) is \(q = q_{69}.\) We find that \(q_{72}\) satisfies the hypothesis of Lemma 2.2 for \(d_2 = 1, \ldots , 9.\) Applying it, we get \(m_2 - m_1 =Y\le 36. \) We now take \(0 \le m_2 - m_1 \le 36. \)
Put
From Eq. (12), we have that
Furthermore, we obtain
Thus, one can see that
Hence, \(\Lambda '> 0,\) and so from (23) we see that
which implies that
We consider \(X_0 = 3 \cdot 10^{30}\)
We get \(q = q_{78} > X_0\), and by Lemma 2.2, we obtain \(n \le 108.\) This contradicts the assumption that \(n > 1000.\) Therefore, the theorem is proved.
4 The Proof of Theorem 1.2
The proof is similar to that of Theorem 1.1. We may sometimes omit some details.
4.1 An Elementary Estimate
We assume that
for some integers \(m_1 \le m_2\) and \(d_1, d_2 \in \{1, 2, \ldots , 9\}.\) A quick computation with Maple reveals no solutions in the interval \( n \in [7, 1000].\) So, from now on, we may assume that \(n > 1000.\) For this computation, we first note that \(Q_{1000}\) has 383 digits. Thus, we generated the list of all repdigits with at most 383 digits; let us call it \(\mathcal {A}.\) Then, for every \(n \in [7, 1000],\) we computed \(M := \lfloor {\log Q_n}/ {\log {10}} \rfloor + 1\) (the number of digits of \(Q_n\)) and then checked whether \( Q_n - d\frac{10^m - 1}{9}\) is a member of \(\mathcal {A}\), for some digit \(d \in \{1, \ldots , 9\}\) and some \(m \in \{M-1, M\}.\) This computation took a few minutes.
Lemma 4.1
All solutions of Eq. (18) satisfy
Proof
The proof follows easily from the fact that \(\alpha ^{n-1}< Q_n < \alpha ^{n+1},\) for \(n\ge 1\). One can see that
Taking the logarithm of all sides, we get \( (n-1) \log \alpha < \log 2 + m_2 \log 10,\) which leads to
Similarly, we get the lower bound. \(\square \)
Next, we return to Eq. (18) and use Binet’s formula (6) to get
i.e.,
We examine (19) in two different steps as follows.
Step 1 Equation (19) gives
which we rewrite as
Thus, dividing both sides by \(d_2 10^{m_2},\) we get
Put
We apply Theorem 2.1 to \(\Gamma _1\). Observe first that \(\Gamma _1\) is not zero, for if it were, then \(\alpha ^n = \frac{d_2 10^{m_2}}{9}\). We get that \(\alpha ^n \in \mathbb {Q},\) which is false, for any \(n > 0.\) We take
where \( \eta _1,\; \eta _2 ,\; \eta _3 \in \mathbb {Q}(\alpha )\) and \( d_1 ,\; d_2 ,\; d_3 \in \mathbb {Z}.\) The degree of \({\mathbb L}:=\mathbb {Q}(\alpha )\) is \(d_{\mathbb {L}} = 2\).
Since \(10^{m_2-1}< Q_m < \alpha ^{n+1},\) we have that \(m_2 < n\). Therefore, we can take \( D = n.\) We note also that the conjugates of \(\eta _1, \eta _2,\) and \(\eta _3\) are \(\eta _1' = \eta _1, \eta _2' = \beta , \eta _3' = \eta _3.\) Furthermore, \(\eta _2\) is algebraic integer, while the minimal polynomial of \(\eta _1\) over \(\mathbb {Q}\) is
Hence, we have \(h(\eta _1) \le h(9) + h(d_2) \le h(9) + h(9) \le 2 h(9).\) This implies that \(2h(\eta _1) < 8.8.\) Thus, as in the previous section we can take
From Theorem 2.1, we obtain
Comparing this last inequality with (20) leads to
giving
Step 2 Equation (19) becomes
which we rewrite as
Thus, dividing both sides by \(\alpha ^n,\) we get
Put
Suppose that \(\Gamma '_1 = 0.\) Then, we have
Conjugating in \( \mathbb {Q}(\alpha )\), we get
Consequently, we obtain
which is impossible for any \(m_1 \ge 1.\) Thus, \(\Gamma '_1\ne 0\). To apply Theorem 2.1, we take
where \( \eta _1 ,\; \eta _2 ,\; \eta _3 \in \mathbb {Q}(\alpha )\) and \( d_1 ,\; d_2 ,\; d_3 \in \mathbb {Z}.\) The degree of \({\mathbb L}=\mathbb {Q}(\alpha )\) is \(d_{\mathbb {L}} = 2\).
As \(1 \le m_1 \le m_2\) and \(m_2 < n\), we can take \(D = n.\) Furthermore, we have
i.e.,
Thus, as before we take
Applying Theorem 2.1 to \(\Gamma '_1\), we get
Comparing this last inequality with (23) leads to
Hence, using inequality (15), we obtain
The above inequality gives us
Lemma 4.1 implies
We summarize what we have proved so far in the following lemma.
Lemma 4.2
All solutions of Eq. (18) satisfy
4.2 Bound Reduction
To lower the above bounds, we return to Eq. (18). We rewrite it into the form
Observe that the term in parentheses is always positive or zero and is zero only when \( d_1 = m_1 = 1\) and \( d_2 = 9\). In this last case, we get \(Q_n = 10^{m_2},\) but such a relation is not possible for \(n > 1000,\) because by the primitive divisor theorem (see [5]), the Pell–Lucas number \(Q_n\) has a prime factor \(\ge n-1\), for all values of \(n > 12.\) Thus, the number appearing in parentheses is \(\ge 1/9.\) Hence, one can see that
From \(\Gamma _1\) given by (21), we have
Let
We get
so
which implies that
Thus, one can see that
with \(Y:=m_2 - m_1< n < 2.3 \cdot 10^{30}.\) We also have that
Therefore, we take
The smallest value of \(q > X_0\) is \(q = q_{69}.\) We find that \(q_{71}\) satisfies the hypotheses of Lemma 2.2 for \(d_2 = 1, \ldots , 8\) (over all the values of \(d_2\ne 9\)). Applying Lemma 2.2, we get \(m_2 - m_1=Y \le 35 \) for \(d_2\ne 9\).
When \(d_2=9\), we get that \( \beta = 0.\) The largest partial quotient \( a_k\) for \(0 \le k \le 146\) is \( a_{120} = 561.\) Applying Lemma 2.1, \(m_2 - m_1=Y < m_2 \le X_0:=2.3\times 10^{30}\) that
We obtain \( m_2 - m_1 \le 34 \), so we get the same conclusion as before, namely that \( m_2 - m_1 \le 35\).
Taking \(1 \le d_1, d_2 \le 9\) and \( 0 \le m_2 - m_1 \le 35,\) we let
From Eq. (19), we have that
Furthermore, one can see that
Thus, we get
Hence, \(\Lambda '_1> 0\) and so from (23) we see that
which implies that
We keep the same values for \(X_0\) and only change \(\psi \) to
We get \(q = q_{75} > X_0. \) By Lemma 2.2, over all the possibilities for the digits \(d_1,d_2\in \{1,\ldots ,9\}\) and \(m_2-m_1\in \{0,\ldots ,33\}\) except for \(m_1=m_2\) and \(d_1+d_2=9\), we get \(n \le 101.\)
When \(m_2 = m_1\) and \(d_1 + d_2 = 9,\) we have \( \beta = 0.\) The largest partial quotient \( a_k\) for \(0 \le k \le 146\) is \( a_{119} = 561.\) Applying Lemma 2.1, we get
We obtain \( n <86\), so we get the same conclusion as before, namely that \( n \le 101.\) This contradicts the assumption that \(n > 1000.\) Hence, the theorem is proved.
References
Alvarado, S.D., Luca, F.: Fibonacci numbers which are sums of two repdigits. Proceedings of the XIVth International Conference on Fibonacci numbers and their applications, Sociedad Matematica Mexicana, Aportaciones Matemáticas, Investigación, vol. 20, pp. 97–108 (2011)
Bravo, J.J., Luca, F.: On a conjecture about repdigits in k-generalized Fibonacci sequences. Publ. Math. Debr. 82, 623–639 (2013)
Bugeaud, Y., Maurice, M., Siksek, S.: Classical and modular approaches to exponential Diophantine equations I. Fibonacci and Lucas perfect powers. Ann. Math. 163, 969–1018 (2006)
Bugeaud, Y., Mignotte, M.: On integers with identical digits. Mathematika 46, 411–417 (1999)
Carmichael, R.D.: On the numerical factors of the arithmetic forms \(\alpha ^n\pm \beta ^n\). Ann. Math. (2) 15, 30–70 (1913)
de Weger, B.M.M.: Algorithms for diophantine equations. Stichting Mathematisch Centrum, Amsterdam (1989)
Dossavi-Yovo, A., Luca, F., Togbé, A.: On the \(x\)-coordinates of Pell equations which are rep-digits. Publ. Math. Debr. 88, 3–4 (2016)
Faye, B., Luca, F.: Pell and Pell–Lucas numbers with only one distinct digit. Ann. Math. Inform. 45, 55–60 (2015)
Luca, F.: Fibonacci and Lucas numbers with only one distinct digit. Port. Math. 57, 243–254 (2000)
Luca, F.: Distinct digits in base b expansions of linear recurrence sequences. Quaest. Math. 23, 389–404 (2000)
Luca, F.: Repdigits which are sums of at most three Fibonacci number. Math. Comm. 17, 1–11 (2012)
Marques, D., Togbé, A.: On terms of linear recurrence sequences with only one distinct block of digits. Colloq. Math. 124, 145–155 (2011)
Marques, D., Togbé, A.: On repdigits as product of consecutive Fibonacci numbers. Rend. Istit. Mat. Univ. Trieste 44, 393–397 (2012)
Matveev, E.M.: An explicit lower bound for a homogeneous rational linear form in logarithms of algebraic numbers, II. Izv. Ross. Akad. Nauk Ser. Mat. 64, 125–180 (2000). English translation in Izv. Math. 64 (2000), 1217–1269
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by Emrah Kilic.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Adegbindin, C., Luca, F. & Togbé, A. Pell and Pell–Lucas Numbers as Sums of Two Repdigits. Bull. Malays. Math. Sci. Soc. 43, 1253–1271 (2020). https://doi.org/10.1007/s40840-019-00739-3
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40840-019-00739-3