Sums of three Fibonacci numbers as concatenations of three repdigits in base b b Passimzouwé Dagou, Pagdame Tiebekabe, Kouèssi Norbert Adédji, Kokou Tchariè Abstract Abstract . In this paper, we investigate sums of three Fibonacci numbers that can be expressed as concatenations of three repdigits in base b , where b\geq 2 is an integer. We prove that for bases 2\leq b\leq 10 , only finitely many such sums exist, and we determine all of them explicitly. Among these solutions, the largest occurs for b=4 and is given by F_{42}+F_{29}+F_{20}=268435290=\overline{33333333331122}_{4}. AMS Subject Classifications: 11B39, 11J86, 11D61, 11Y50 11A63, 11B37. Keywords: Diophantine equations; Fibonacci numbers; b -repdigits; linear forms in logarithms; reduction method; concatenations. 1. Introduction The interaction between recurrence sequences and digital representations of integers has attracted considerable attention in recent years. In particular, problems concerning the occurrence of special digital patterns within classical sequences such as the Fibonacci sequence have proved to be a rich source of Diophantine questions combining number theory, recurrence relations, and combinatorial properties of numeration systems. Among these patterns, repdigits —integers whose digits are all identical in a given base—and their concatenations play a central role. In a seminal paper published in 2019, Alahmadi, Altassan, Luca, and Shoaib (see [ 4 ] ) initiated a systematic study of Fibonacci numbers that can be expressed as concatenations of two repdigits. Their work opened a new line of research, leading to several subsequent contributions that extended the notion of concatenation and repdigits to other linear recurrence sequences. Notably, analogous problems have been investigated for sequences such as Pell, Pell–Lucas, Padovan, Perrin, and balancing numbers, highlighting the broad relevance and flexibility of this framework (see [ 1 ] , [ 2 ] , [ 3 ] , [ 10 ] , [ 14 ] , [ 15 ] , [ 16 ] ). Motivated by these developments, and after a careful review of the existing literature, we address a natural and yet unexplored question: can sums of Fibonacci numbers admit representations as concatenations of repdigits? More precisely, in this paper we study the sum of three Fibonacci numbers that can be written as the concatenation of three repdigits in base b . The main objective of this work is to characterize such representations, establish finiteness results where applicable, and determine all possible solutions under suitable constraints. Our results extend and complement previous studies on digital properties of Fibonacci numbers, and contribute further insight into the interplay between recurrence sequences and digital concatenation phenomena. Let \{F_{n}\}_{n\geq 0} be the Fibonacci sequence given by F_{n+2}=F_{n+1}+F_{n}, with initial values F_{0}=0 and F_{1}=1. If (\alpha,\beta)=\left(\dfrac{1+\sqrt{5}}{2},\dfrac{1-\sqrt{5}}{2}\right) is the pair of roots of the characteristic equation x^{2}-x-1=0 of the Fibonacci numbers, then the Binet’s formula for its general term is (1) \displaystyle F_{n}=\dfrac{\alpha^{n}-\beta^{n}}{\alpha-\beta}\quad n\geq 0. It can be seen that 1<\alpha<2,\;-1<\beta<0 and \alpha\beta=-1. The following relations between n -th Fibonacci number F_{n} and \alpha is well known (2) \displaystyle\alpha^{n-2}0. Then we have the explicit upper bound n_{1}<3.09\times 10^{82}\,\log^{11}b. Theorem 2 . For each integer base b with 2\leq b\leq 10 , there exist a finite number of sums of three Fibonacci numbers that can be written as concatenations of three repdigits in base b . In total, there are 2665 such distinct sums across all bases. Moreover, the following table presents, for each base b , • the number of distinct sums, • the largest sum F_{n_{1}}+F_{n_{2}}+F_{n_{3}} , • and its representation in base b . Table 1 . Largest sums of three Fibonacci numbers represented as concatenations of three repdigits in bases 2 to 10 b 2 3 4 Number of distinct sums 113 138 217 Largest sum F_{n_{1}}+F_{n_{2}}+F_{n_{3}} F_{23}+F_{8}+F_{2}=28679 F_{27}+F_{14}+F_{9}=196829 F_{42}+F_{29}+F_{20}=268435290 Representation 111000000000111 100222222222 33333333331122 b 5 6 7 Number of distinct sums 250 334 348 Largest sum F_{n_{1}}+F_{n_{2}}+F_{n_{3}} F_{27}+F_{14}+F_{10}=196850 F_{42}+F_{29}+F_{20}=24186563 F_{34}+F_{24}+F_{17}=5750852 Representation 22244400 2222222455 66611222 b 8 9 10 Number of distinct sums 387 413 465 Largest sum F_{n_{1}}+F_{n_{2}}+F_{n_{3}} F_{29}+F_{25}+F_{23}=617911 F_{33}+F_{21}+F_{9}=3535558 F_{35}+F_{28}+F_{26}=3535558 Representation 2266667 6577777 9666669 As immediate consequences of the main theorems, we obtain the following corollaries. Corollaire 1.1 . In equation ( 3 ), setting n_{3}=0 yields all distinct sums of two Fibonacci numbers that can be expressed as concatenations of three repdigits in base b . In particular we have the following results listened in the table below. Table 2 . Largest sums of two Fibonacci numbers represented as concatenations of three repdigits in bases 2 to 10 b 2 3 4 Number of distinct sums 34 42 49 Largest sum F_{n_{1}}+F_{n_{2}} F_{21}+F_{17}=12543 F_{23}+F_{16}=29644 F_{21}+F_{17}=12543 Representation 11000011111111 1111122221 3003333 b 5 6 7 Number of distinct sums 57 71 65 Largest sum F_{n_{1}}+F_{n_{2}} F_{27}+F_{12}=196562 F_{37}+F_{23}=24186474 F_{21}+F_{13}=11179 Representation 22242222 2222222230 44410 b 8 9 10 Number of distinct sums 81 78 80 Largest sum F_{n_{1}}+F_{n_{2}} F_{27}+F_{12}=196562 F_{21}+F_{19}=15127 F_{28}+F_{22}=335522 Representation 577722 22667 335522 Corollaire 1.2 . In equation ( 3 ), setting n_{2}=n_{3}=0 yields all Fibonacci numbers that can be expressed as concatenations of three repdigits in base b . In particular we get the following results listened in the table bellow. Table 3 . Largest Fibonacci numbers represented as concatenations of three repdigits in bases 2 to 10 b 2 3 4 Number of distinct sums 4 6 7 Largest F_{n} F_{10}=55 F_{14}=377 F_{19}=4181 Representation 110111 111222 1001111 b 5 6 7 Number of distinct sums 5 6 5 Largest F_{n} F_{15}=610 F_{24}=46386 F_{17}=1597 Representation 4420 554400 4441 b 8 9 10 Number of distinct sums 6 6 6 Largest F_{n} F_{16}=987 F_{19}=4181 F_{22}=17711 Representation 1733 5655 17711 The proofs of our results rely crucially on lower bounds for linear forms in logarithms, combined with Baker’s reduction method. Furthermore, our computations are carried out using the approach developed by Chim and Ziegler in [ 9 ] . 2. Preliminaries 2.1. Linear form in logarithms Denote by \eta_{1},\ldots,\eta_{s} algebraic numbers, none of which is 0 or 1 , and let \log\eta_{1},\ldots,\log\eta_{s} be a fixed determination of their logarithms. Let \mathbb{K}=\mathbb{Q}(\eta_{1},\ldots,\eta_{s}) and denote by d_{\mathbb{K}}=[\mathbb{K}:\mathbb{Q}] the degree of \mathbb{K} over \mathbb{Q} . For any \eta\in\mathbb{K} , let its minimal polynomial over \mathbb{Z} be f(X)=a_{0}X^{\delta}+a_{1}X^{\delta-1}+\cdots+a_{\delta}=a_{0}\prod_{j=1}^{\delta}(X-\eta^{(j)}), where \eta^{(j)} , j=1,\ldots,\delta , are the roots of f(X) . The absolute logarithmic Weil height of \eta is defined by h_{0}(\eta)=\frac{1}{\delta}\left(\log|a_{0}|+\sum_{j=1}^{\delta}\log\big(\max\{|\eta^{(j)}|,1\}\big)\right). In particular, if \eta=p/q is a rational number in lowest terms, i.e., \gcd(p,q)=1 and q>0 , then h_{0}(\eta)=\log\max(|p|,q) . The modified height h^{\prime}(\eta) is defined by h^{\prime}(\eta)=\frac{1}{d_{\mathbb{K}}}\max\{h(\eta),|\log\eta|,1\}, where h(\eta)=d_{\mathbb{K}}\,h_{0}(\eta) is the standard logarithmic Weil height of \eta . Consider the linear form L(z_{1},\ldots,z_{s})=b_{1}z_{1}+\cdots+b_{s}z_{s}, where b_{1},\ldots,b_{s} are integers, not all zero. Its modified height is h^{\prime}(L)=\frac{1}{d_{\mathbb{K}}}\max\{h(L),1\}, where h(L)=d_{\mathbb{K}}\log\left(\max_{1\leq j\leq s}\left\{\frac{|b_{j}|}{b}\right\}\right) is the logarithmic Weil height of L , with b=\gcd(b_{1},\ldots,b_{s}) . If we define B=\max\{|b_{1}|,\ldots,|b_{s}|,e\} , then we have the simple bound h^{\prime}(L)\leq\log B. With these notations, we are now able to state the following fundamental result due to Baker and Wüstholz [ 5 ] . Theorem 3 . If \Gamma=L(\log\eta_{1},\ldots,\log\eta_{s})\neq 0 , then \log|\Gamma|\geq-C(s,d_{\mathbb{K}})\,h^{\prime}(\eta_{1})\cdots h^{\prime}(\eta_{s})\,h^{\prime}(L), where C(s,d_{\mathbb{K}})=18(s+1)!\,s^{s+1}(32d_{\mathbb{K}})^{s+2}\log(2sd_{\mathbb{K}}). With |\Lambda|\leq\tfrac{1}{2} , where \Lambda=e^{\Gamma}-1=\eta_{1}^{b_{1}}\cdots\eta_{s}^{b_{s}}-1, we have \tfrac{1}{2}|\Gamma|\leq|\Lambda|\leq 2|\Gamma| so that \log\left|\eta_{1}^{b_{1}}\cdots\eta_{s}^{b_{s}}-1\right|\geq\log|\Gamma|-\log 2. We apply Theorem 3 mainly in the situation where \mathbb{K}=\mathbb{Q}(\sqrt{5}) , s=3 and d_{\mathbb{K}}=2 . In this case we obtain C(3,2)=18\cdot 4!\cdot 3^{4}\cdot 64^{5}\log 12<9.34\cdot 10^{13}. This value will be used throughout the paper without further mention. We recall below some well-known properties of the absolute logarithmic height. h_{0}(\eta+\gamma)\leq h_{0}(\eta)+h_{0}(\gamma)+\log 2, h_{0}(\eta\gamma^{\pm 1})\leq h_{0}(\eta)+h_{0}(\gamma), h_{0}(\eta^{\ell})=|\ell|\,h_{0}(\eta), where \eta,\gamma are algebraic numbers and \ell\in\mathbb{Z} . We also need the following result from Guzmán and Luca [ 13 ] . Lemma 1 . (Lemma 7 of [ 13 ] ) If l\geq 1,H>(4l^{2})^{l} and H>L/(\log L)^{l} , then L<2^{l}H(\log H)^{l} . 2.2. Reduction method The bounds for the variables derived from Baker’s theory are far too large to allow for effective computational verification. To reduce the bounds we use the reduction method due to Bravo, Gómez and Luca [ 7 ] which is a modified version due to Dujella and Pethő (see [ 11 ] ). Lemma 2 . Let M be a positive integer, p/q be a convergent of the continued fraction expansion of irrational number \tau such as q>6M , and A,B,\mu be some real numbers with A>0 and B>1. Furthermore, let \varepsilon:=||\mu q||-M.||\tau q||. If \varepsilon>0 , the there is no solution to the inequality 0<|u\tau-v+\mu|300 , since the remaining cases can be checked directly by computation. The proof of Theorem 1 is divided into several steps. We derive initial bounds for the variables using estimates coming from linear forms in logarithms. Step 3.1 . We prove the following Lemma. Lemma 3 . All the solution of equation ( 3 ) satisfy \min\left\{\ell_{1}\log b,(n_{1}-n_{2})\log\alpha\right\}<1.5\times 10^{14}\log^{2}b\log(n_{1}+2). Proof. As a first step, we analyze the Diophantine equation ( 3 ) and transform it into the equivalent form \displaystyle F_{n_{1}}+F_{n_{2}}+F_{n_{3}} \displaystyle=\overline{\underbrace{d_{1}\,\ldots\,d_{1}}_{\ell_{1}\ \text{times}}\;\underbrace{d_{2}\,\ldots\,d_{3}}_{\ell_{2}\ \text{times}}\;\underbrace{d_{3}\,\ldots\,d_{3}}_{\ell_{3}\ \text{times}}} \displaystyle=\underbrace{d_{1}\,\ldots\,d_{1}}_{\ell_{1}\ \text{times}}.b^{\ell_{2}+\ell_{3}}+\underbrace{d_{2}\,\ldots\,d_{2}}_{\ell_{2}\ \text{times}}.b^{\ell_{3}}+\underbrace{d_{3}\,\ldots\,d_{3}}_{\ell_{3}\ \text{times}} \displaystyle F_{n_{1}}+F_{n_{2}}+F_{n_{3}} \displaystyle=d_{1}\left(\frac{b^{\ell_{1}}-1}{b-1}\right).b^{\ell_{2}+\ell_{3}}+d_{2}\left(\frac{b^{\ell_{2}}-1}{b-1}\right).b^{\ell_{3}}+d_{3}\left(\frac{b^{\ell_{3}}-1}{b-1}\right) and therefore we get (4) F_{n_{1}}+F_{n_{2}}+F_{n_{3}}=\frac{1}{b-1}\left(d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}-(d_{1}-d_{2})b^{\ell_{2}+\ell_{3}}-(d_{2}-d_{3})b^{\ell_{3}}-d_{3}\right). We first observe that combining ( 2 ) and ( 3 ) yields b^{\ell_{1}+\ell_{2}+\ell_{3}}>F_{n_{1}}+F_{n_{2}}+F_{n_{3}}>F_{n_{1}}>\alpha^{n_{1}-2} and (5) 3\alpha^{n_{1}-1}>3F_{n_{1}}>F_{n_{1}}+F_{n_{2}}+F_{n_{3}}\geq b^{\ell_{1}+\ell_{2}+\ell_{3}-1}. By taking logarithm on the both sides of ( 5 ), we obtain (\ell_{1}+\ell_{2}+\ell_{3}-1)\log b<(n_{1}-1)\log\alpha+\log 3 which leads to \ell_{1}+\ell_{2}+\ell_{3}300 . Dividing both sides of ( 6 ) by \sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}} , we obtain \displaystyle\left|\left(\frac{b-1}{\sqrt{5}d_{1}}\right).\alpha^{n_{1}}.b^{-(\ell_{1}+\ell_{2}+\ell_{3})}-1\right| \displaystyle<\max\left\{\frac{6.7b^{\ell_{2}+\ell_{3}+1}}{\sqrt{5}b^{\ell_{1}+\ell_{2}+\ell_{3}}},\frac{6.7b\alpha^{n_{2}}}{\sqrt{5}b^{\ell_{1}+\ell_{2}+\ell_{3}}}\right\} \displaystyle<\max\left\{\frac{6.7b^{-\ell_{1}+1}}{\sqrt{5}},\frac{6.7b\alpha^{n_{2}}}{\sqrt{5}\alpha^{n_{1}-2}}\right\} (7) \displaystyle<7.85b\max\left\{b^{-\ell_{1}},\alpha^{-(n_{1}-n_{2})}\right\}. We now define the linear form \Gamma_{1}=\log\left(\frac{b-1}{d_{1}\sqrt{5}}\right)+n_{1}\log\alpha-(\ell_{1}+\ell_{2}+\ell_{3})\log b and put (8) \Lambda_{1}:=\mathrm{e}^{\Gamma_{1}}-1=\left(\frac{b-1}{\sqrt{5}d_{1}}\right).\alpha^{n_{1}}.b^{-(\ell_{1}+\ell_{2}+\ell_{3})}-1. We assume that \left|\Lambda_{1}\right|\leq 0.5 and use the theorem of Baker and Wüstholz (Theorem 3 ) with data \eta_{1}:=\frac{b-1}{d_{1}\sqrt{5}},\eta_{2}:=\alpha,\eta_{3}:=b,b_{1}=1,b_{2}:=n_{1},b_{3}:=-(\ell_{1}+\ell_{2}+\ell_{3}),s:=3. Since \ell_{1}+\ell_{2}+\ell_{3}0 . Using the properties of the logarithmic height, we have \displaystyle h_{0}(\eta_{1})=h_{0}\left(\frac{b-1}{\sqrt{5}d_{1}}\right) \displaystyle\leq h_{0}\left(\frac{b-1}{d_{1}}\right)+h_{0}\left(\sqrt{5}\right) \displaystyle\leq h_{0}(b-1)+h_{0}(d_{1})+\frac{\log 5}{2} \displaystyle\leq 2\log b+\frac{\log 5}{2}\leq 3.2\log b\text{ since }b\geq 2. Then, we have h^{\prime}(\eta_{1})<3.2\log b. Similarly, we have h^{\prime}\left(\eta_{2}\right)=h(\alpha)=\frac{1}{2},\;h\left(\eta_{3}\right)=h(b)=\log b . Theorem 3 tells us that \displaystyle\log\left|\Lambda_{1}\right| \displaystyle>-C(3,2)\left(3.2\log b\right)\left(\frac{1}{2}\right)\left(\log b\right)\left(\log(n_{1}+2)\right)-\log 2. Comparing this last inequality with ( 7 ) leads to \min\left\{\ell_{1}\log b,(n_{1}-n_{2})\log\alpha\right\}<1.5\times 10^{14}\log^{2}b\log(n_{1}+2). Note that in the case that \left|\Lambda_{1}\right|>0.5 , inequality ( 7 ) is possible only if either \ell_{1}\leq 4 and n_{1}-n_{2}\leq\left\lfloor\frac{\log(15.7b)}{\log\alpha}\right\rfloor , which are covered by the bound provided by Lemma 3 ∎ We now distinguish between the following two cases: Case 1: \min\left\{\ell_{1}\log b,(n_{1}-n_{2})\log\alpha\right\}=(n_{1}-n_{2})\log\alpha, Case 2: \min\left\{\ell_{1}\log b,(n_{1}-n_{2})\log\alpha\right\}=\ell_{1}\log b. We will treat these two cases separately in the following steps. Step 3.2 . We consider Case 1 and show that under the assumption that (n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\log^{2}b\log(n_{1}+2) we obtain that \min\left\{\ell_{1}\log b,(n_{1}-n_{3})\log\alpha\right\}<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2). Since we consider Case 1, we assume that \min\left\{\ell_{1}\log b,(n_{1}-n_{2})\log\alpha\right\}=(n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\log^{2}b\log(n_{1}+2). Using ( 1 ) and ( 4 ), we get \frac{\alpha^{n_{1}}-\beta^{n_{1}}}{\sqrt{5}}+\frac{\alpha^{n_{2}}-\beta^{n_{2}}}{\sqrt{5}}+F_{n_{3}}=\frac{1}{b-1}\left(d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}-(d_{1}-d_{2})b^{\ell_{2}+\ell_{3}}-(d_{2}-d_{3})b^{\ell_{3}}-d_{3}\right) which implies that \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}}-\sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}} \displaystyle=-\sqrt{5}\left((d_{1}-d_{2})b^{\ell_{2}+\ell_{3}}+(d_{2}-d_{3})b^{\ell_{3}}+d_{3}\right) \displaystyle-\sqrt{5}(b-1)F_{n_{3}}+(b-1)\left(\beta^{n_{1}}+\beta^{n_{2}}\right). Thus \displaystyle\left|(b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}\right)-\sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}\right| \displaystyle<\sqrt{5}\left((b-1)b^{\ell_{2}+\ell_{3}}+(b-1)b^{\ell_{3}}+(b-1)\right) \displaystyle+\sqrt{5}(b-1)\alpha^{n_{3}}+2(b-1)\alpha^{-n_{2}} \displaystyle<1.75\sqrt{5}b^{\ell_{2}+\ell_{3}+1}+\sqrt{5}b\alpha^{n_{3}-1}+2b\alpha^{-n_{2}} and therefore (9) \left|(b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}\right)-\sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}\right|<5.3b\max\{b^{\ell_{2}+\ell_{3}},\alpha^{n_{3}}\}. Dividing both sides of ( 9 ) by \sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}} , we obtain \displaystyle\left|\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}\right)}{\sqrt{5}d_{1}}\right).\alpha^{n_{1}}.b^{-(\ell_{1}+\ell_{2}+\ell_{3})}-1\right| \displaystyle<\max\left\{\frac{5.3b^{\ell_{2}+\ell_{3}+1}}{\sqrt{5}b^{\ell_{1}+\ell_{2}+\ell_{3}}},\frac{5.3b\alpha^{n_{3}}}{\sqrt{5}b^{\ell_{1}+\ell_{2}+\ell_{3}}}\right\} \displaystyle<\max\left\{\frac{5.3b^{-\ell_{1}+1}}{\sqrt{5}},\frac{5.3b\alpha^{n_{3}}}{\sqrt{5}\alpha^{n_{1}-2}}\right\} (10) \displaystyle<9.93b\max\left\{b^{-\ell_{1}},\alpha^{-(n_{1}-n_{3})}\right\}. Let set the linear form \Gamma_{2}=\log\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}\right)}{\sqrt{5}d_{1}}\right)+n_{1}\log\alpha-(\ell_{1}+\ell_{2}+\ell_{3})\log b and put (11) \Lambda_{2}:=\mathrm{e}^{\Gamma_{2}}-1=\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}\right)}{\sqrt{5}d_{1}}\right).\alpha^{n_{1}}.b^{-(\ell_{1}+\ell_{2}+\ell_{3})}-1. We assume that \left|\Lambda_{2}\right|\leq 0.5 and use the theorem of Baker and Wüstholz (Theorem 3 ) with data \eta_{1}:=\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}\right)}{\sqrt{5}d_{1}}\right),\eta_{2}:=\alpha,\eta_{3}:=b,b_{1}=1,b_{2}:=n_{1},b_{3}:=-(\ell_{1}+\ell_{2}+\ell_{3}),s:=3. Since \ell_{1}+\ell_{2}+\ell_{3}300\text{ and }n_{2}>0, which is false. Therefore, \Lambda_{2}\neq 0 . Using the properties of the logarithmic height, we have \displaystyle h_{0}(\eta_{1}) \displaystyle=h_{0}\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}\right)}{\sqrt{5}d_{1}}\right) \displaystyle\leq h_{0}\left(b-1\right)+(n_{1}-n_{2})h_{0}(\alpha)+h_{0}(d_{1})+\log\sqrt{5}+\log 2 \displaystyle\leq 2\log b+(n_{1}-n_{2})\log\alpha+\log\sqrt{5}+\log 2 \displaystyle\leq 1.5\times 10^{14}\log^{2}b\log(n_{1}+2)+2\log b+\log\sqrt{5}+\log 2 \displaystyle<1.6\times 10^{14}\log^{2}b\log(n_{1}+2). Then we have h^{\prime}(\eta_{1})<1.6\times 10^{14}\log^{2}b\log(n_{1}+2). Similarly, we have h^{\prime}\left(\eta_{2}\right)=h(\alpha)=\frac{1}{2},\;h\left(\eta_{3}\right)=h(b)=\log b . Theorem 3 tells us that \displaystyle\log\left|\Lambda_{2}\right| \displaystyle>-C(3,2)\left(1.6\times 10^{14}\log^{2}b\log(n_{1}+2)\right)\left(\frac{1}{2}\right)\left(\log b\right)\left(\log(n+2)\right)-\log 2 \displaystyle>-7.472\times 10^{27}\log^{3}b\log^{2}(n_{1}+2). By comparing this inequality with ( 10 ), we obtain \min\left\{\ell_{1}\log b,(n_{1}-n_{3})\log\alpha\right\}<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2). Note that in the case where |\Lambda_{2}|>0.5 , inequality ( 10 ) can only hold if either \ell_{1}\leq 5 or n_{1}-n_{3}\leq\left\lfloor\frac{\log(19.86b)}{\log\alpha}\right\rfloor, both of which are already covered by the bounds provided in Step 2. At this stage, we consider the following two sub-cases for Case 1: Case 1A: \min\left\{\ell_{1}\log b,(n_{1}-n_{3})\log\alpha\right\}=(n_{1}-n_{3})\log\alpha, Case 1B: \min\left\{\ell_{1}\log b,(n_{1}-n_{3})\log\alpha\right\}=\ell_{1}\log b. We will treat these sub-cases separately in the following steps. Step 3.3 . We now consider Case 1A and show that, under the assumptions (n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\,\log^{2}b\,\log(n_{1}+2)\quad\text{and}\quad(n_{1}-n_{3})\log\alpha<7.48\times 10^{27}\,\log^{3}b\,\log^{2}(n_{1}+2), it follows that \ell_{1}\log b<1.76\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2). Combining ( 1 ) and ( 4 ), we deduce \frac{\alpha^{n_{1}}-\beta^{n_{1}}}{\sqrt{5}}+\frac{\alpha^{n_{2}}-\beta^{n_{2}}}{\sqrt{5}}+\frac{\alpha^{n_{3}}-\beta^{n_{3}}}{\sqrt{5}}=\frac{1}{b-1}\left(d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}-(d_{1}-d_{2})b^{\ell_{2}+\ell_{3}}-(d_{2}-d_{3})b^{\ell_{3}}-d_{3}\right) which implies \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}}+(b-1)\alpha^{n_{3}}-\sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}} \displaystyle=-\sqrt{5}\left((d_{1}-d_{2})b^{\ell_{2}+\ell_{3}}+(d_{2}-d_{3})b^{\ell_{3}}+d_{3}\right) \displaystyle+(b-1)\left(\beta^{n_{1}}+\beta^{n_{2}}+\beta^{n_{3}}\right) and \displaystyle\left|(b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)-\sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}\right| \displaystyle<\sqrt{5}\left((b-1)b^{\ell_{2}+\ell_{3}}+(b-1)b^{\ell_{3}}+(b-1)\right) \displaystyle+3(b-1)\alpha^{-n_{3}}<4b^{\ell_{2}+\ell_{3}+1}. We then deduce (12) \left|(b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)-\sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}}\right|<4b^{\ell_{2}+\ell_{3}+1}. Dividing both sides of ( 12 ) by \sqrt{5}d_{1}b^{\ell_{1}+\ell_{2}+\ell_{3}} , we obtain (13) \displaystyle\left|\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)}{\sqrt{5}d_{1}}\right).\alpha^{n_{1}}.b^{-(\ell_{1}+\ell_{2}+\ell_{3})}-1\right| \displaystyle<1.8b^{-\ell_{1}+1}. Let set the linear form \Gamma_{3}=\log\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)}{\sqrt{5}d_{1}}\right)+n_{1}\log\alpha-(\ell_{1}+\ell_{2}+\ell_{3})\log b and put (14) \Lambda_{3}:=\mathrm{e}^{\Gamma_{3}}-1=\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)}{\sqrt{5}d_{1}}\right).\alpha^{n_{1}}.b^{-(\ell_{1}+\ell_{2}+\ell_{3})}-1. We apply Theorem 3 to inequality ( 14 ) by taking B=n_{1}+2 , \eta_{1}:=\left(\frac{(b-1)\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)}{\sqrt{5}d_{1}}\right),\eta_{2}:=\alpha,\eta_{3}:=b,b_{1}=1,b_{2}:=n_{1},b_{3}:=-(\ell_{1}+\ell_{2}+\ell_{3}),s:=3. With this choice we have h^{\prime}(\eta_{1})<3.75\times 10^{27}\log^{3}b\log^{2}(n_{1}+2) and \Lambda_{3}\neq 0 . We obtain \ell_{1}\log b<1.76\times 10^{41}\log^{4}b\log^{3}(n_{1}+2). Step 3.4 . We consider Case 1A and show tat under the assumption that (n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\log^{2}b\log(n_{1}+2),\,\,(n_{1}-n_{3})\log\alpha<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2)\text{ and } \ell_{1}\log b<1.76\times 10^{41}\log^{4}b\log^{3}(n_{1}+2), we obtain that \ell_{2}\log b<8.27\times 10^{54}\log^{5}b\log^{4}(n_{1}+2). Using ( 1 ) and ( 4 ), we obtain \displaystyle(b-1)\alpha^{n_{1}} \displaystyle+(b-1)\alpha^{n_{2}}+(b-1)\alpha^{n_{3}}-\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)b^{\ell_{2}+\ell_{3}} \displaystyle=-\sqrt{5}\left((d_{2}-d_{3})b^{\ell_{3}}+d_{3}\right)+(b-1)\left(\beta^{n_{1}}+\beta^{n_{2}}+\beta^{n_{3}}\right) and the following estimates \displaystyle\left|(b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)-\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)b^{\ell_{2}+\ell_{3}}\right| \displaystyle<\sqrt{5}(b-1)\left(b^{\ell_{3}}+1\right)+3(b-1)\alpha^{-n_{3}} \displaystyle<3.36(b-1)b^{\ell_{3}}. Dividing both sides of the above inequality by (b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right) , we get \displaystyle\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{(b-1)(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}})}\alpha^{-n_{1}}b^{\ell_{2}+\ell_{3}}-1\right| \displaystyle<\frac{3.36b^{\ell_{3}}}{\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)} \displaystyle<\frac{3.36b^{\ell_{3}}}{\alpha^{n_{1}}}. Using ( 5 ) we have \frac{1}{\alpha^{n_{1}}}<\frac{3}{\alpha b^{\ell_{1}+\ell_{2}+\ell_{3}-1}}<\frac{1.86}{b^{\ell_{1}+\ell_{2}+\ell_{3}-1}}. Therefore the above inequality implies \displaystyle\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{(b-1)(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}})}\alpha^{-n_{1}}b^{\ell_{2}+\ell_{3}}-1\right| \displaystyle<\frac{6.25b^{-(\ell_{2}-1)}}{b^{\ell_{1}}}, which leads to (15) \displaystyle|\Lambda_{4}|=\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{(b-1)(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}})}\alpha^{-n_{1}}b^{\ell_{2}+\ell_{3}}-1\right|<3.13b^{-(\ell_{2}-1)}\text{ since }b\geq 2. We apply Theorem 3 to inequality ( 15 ) by taking B=n_{1}+2 , \eta_{1}:=\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{(b-1)(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}})},\eta_{2}:=\alpha,\eta_{3}:=b,b_{1}=1,b_{2}:=-n_{1},b_{3}:=\ell_{2}+\ell_{3},s:=3. Note that h^{\prime}(\eta_{1})<1.77\times 10^{41}\log^{4}b\log^{3}(n_{1}+2) and |\Lambda_{4}|\neq 0 . Therefore, we get \ell_{2}\log b<8.27\times 10^{54}\log^{5}b\log^{4}(n_{1}+2). Step 3.5 . We now consider Case 1B and show that, under the assumptions (n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\,\log^{2}b\,\log(n_{1}+2),\quad\ell_{1}\log b<7.48\times 10^{27}\,\log^{3}b\,\log^{2}(n_{1}+2), it follows that \min\left\{\ell_{2}\log b,\,(n_{1}-n_{3})\log\alpha\right\}<3.6\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2). From ( 1 ) and ( 4 ), we deduce \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}}-\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)b^{\ell_{2}+\ell_{3}} \displaystyle\quad=-\sqrt{5}\bigl((d_{2}-d_{3})b^{\ell_{3}}+d_{3}\bigr)-\sqrt{5}(b-1)F_{n_{3}}+(b-1)(\beta^{n_{1}}+\beta^{n_{2}}), which implies \displaystyle\Bigl|(b-1)\alpha^{n_{1}}\bigl(1+\alpha^{n_{2}-n_{1}}\bigr)-\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)b^{\ell_{2}+\ell_{3}}\Bigr| \displaystyle\quad<\sqrt{5}(b-1)(b^{\ell_{3}}+1)+\sqrt{5}(b-1)\alpha^{n_{3}-1}+2(b-1)\alpha^{-n_{3}} \displaystyle\quad<1.5\sqrt{5}(b-1)b^{\ell_{3}}+1.39(b-1)\alpha^{n_{3}} \displaystyle\quad<4.75(b-1)\max\{b^{\ell_{3}},\alpha^{n_{3}}\}. Dividing both sides by (b-1)\alpha^{n_{1}}(1+\alpha^{n_{2}-n_{1}}) , we obtain \displaystyle\Biggl|\frac{\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)}{(b-1)(1+\alpha^{n_{2}-n_{1}})}\alpha^{-n_{1}}b^{\ell_{2}+\ell_{3}}-1\Biggr| \displaystyle<\max\Bigl\{\frac{4.75b^{\ell_{3}}}{\alpha^{n_{1}}},\,\alpha^{-(n_{1}-n_{3})}\Bigr\}. Applying the bound \frac{1}{\alpha^{n_{1}}}<\frac{1.86}{b^{\ell_{1}+\ell_{2}+\ell_{3}-1}}, we then deduce (16) \displaystyle|\Lambda_{5}|=\Biggl|\frac{\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)}{(b-1)(1+\alpha^{n_{2}-n_{1}})}\alpha^{-n_{1}}b^{\ell_{2}+\ell_{3}}-1\Biggr|<4.42\,\max\bigl\{b^{-(\ell_{2}-1)},\,\alpha^{-(n_{1}-n_{3})}\bigr\}. We apply Theorem 3 to inequality ( 16 ) with parameters B=n_{1}+2,\quad\eta_{1}:=\frac{\sqrt{5}(d_{1}b^{\ell_{1}}-(d_{1}-d_{2}))}{(b-1)(1+\alpha^{n_{2}-n_{1}})},\quad\eta_{2}:=\alpha,\quad\eta_{3}:=b,\quad b_{1}=1,\quad b_{2}:=-n_{1}, b_{3}:=\ell_{2}+\ell_{3}, and s:=3. In this case, we have h^{\prime}(\eta_{1})<7.5\times 10^{27}\,\log^{3}b\,\log^{2}(n_{1}+2) and |\Lambda_{5}|\neq 0 . Hence, we conclude that \min\bigl\{\ell_{2}\log b,\,(n_{1}-n_{3})\log\alpha\bigr\}<3.6\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2). We now consider the following sub-sub-cases for Case 1B: Case 1B-A: \min\left\{\ell_{2}\log b,\,(n_{1}-n_{3})\log\alpha\right\}=(n_{1}-n_{3})\log\alpha, Case 1B-B: \min\left\{\ell_{2}\log b,\,(n_{1}-n_{3})\log\alpha\right\}=\ell_{2}\log b. We will treat these sub-sub-cases separately in the following steps. Step 3.6 . We consider Case 1B-A and show that, under the assumptions (n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\,\log^{2}b\,\log(n_{1}+2),\quad\ell_{1}\log b<7.48\times 10^{27}\,\log^{3}b\,\log^{2}(n_{1}+2), (n_{1}-n_{3})\log\alpha<3.6\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2) it follows that \ell_{2}\log b<8.7\times 10^{54}\,\log^{5}b\,\log^{4}(n_{1}+2). From ( 1 ) and ( 4 ), we deduce \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}}+(b-1)\alpha^{n_{3}}-\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)b^{\ell_{2}+\ell_{3}} \displaystyle\quad=-\sqrt{5}\bigl((d_{2}-d_{3})b^{\ell_{3}}+d_{3}\bigr)+(b-1)(\beta^{n_{1}}+\beta^{n_{2}}+\beta^{n_{3}}), which implies \displaystyle\Bigl|(b-1)\alpha^{n_{1}}\bigl(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\bigr)-\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)b^{\ell_{2}+\ell_{3}}\Bigr| \displaystyle\quad<\sqrt{5}(b-1)(b^{\ell_{3}}+1)+3(b-1)\alpha^{-n_{3}} \displaystyle\quad<3.36\,b^{\ell_{3}+1}. Dividing both sides by \sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)b^{\ell_{2}+\ell_{3}} , we obtain \displaystyle\Biggl|\left(\frac{(b-1)\bigl(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\bigr)}{\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)}\right)\alpha^{n_{1}}b^{-(\ell_{2}+\ell_{3})}-1\Biggr|<\frac{3.36\,b^{\ell_{3}+1}}{\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)b^{\ell_{2}+\ell_{3}}}. Hence, we obtain (17) \displaystyle|\Lambda_{6}|=\Biggl|\left(\frac{(b-1)\bigl(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\bigr)}{\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)}\right)\alpha^{n_{1}}b^{-(\ell_{2}+\ell_{3})}-1\Biggr|<1.6\,b^{-\ell_{2}+1}. Applying Theorem 3 to inequality ( 17 ) with parameters B=n_{1}+2,\quad\eta_{1}:=\frac{(b-1)(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}})}{\sqrt{5}(d_{1}b^{\ell_{1}}-(d_{1}-d_{2}))},\quad\eta_{2}:=\alpha,\quad\eta_{3}:=b,\quad b_{1}=1,\quad b_{2}:=n_{1}, b_{3}:=-(\ell_{2}+\ell_{3}),\quad s:=3, and noting that h^{\prime}(\eta_{1})<1.85\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2) and |\Lambda_{6}|\neq 0 , we conclude that \ell_{2}\log b<8.7\times 10^{54}\,\log^{5}b\,\log^{4}(n_{1}+2). Step 3.7 . We consider Case 1B-B and show that, under the assumptions (n_{1}-n_{2})\log\alpha<1.5\times 10^{14}\,\log^{2}b\,\log(n_{1}+3),\quad\ell_{1}\log b<7.48\times 10^{27}\,\log^{3}b\,\log^{2}(n_{1}+2), \ell_{2}\log b<3.6\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2), it follows that (n_{1}-n_{3})\log\alpha<8.7\times 10^{54}\,\log^{5}b\,\log^{4}(n_{1}+2). From ( 1 ) and ( 4 ), we deduce \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}}-\sqrt{5}\bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\bigr)b^{\ell_{3}} \displaystyle\quad=-\sqrt{5}d_{3}-\sqrt{5}(b-1)F_{n_{3}}+(b-1)(\beta^{n_{1}}+\beta^{n_{2}}), which implies \displaystyle\Bigl|(b-1)\alpha^{n_{1}}(1+\alpha^{n_{2}-n_{1}})-\sqrt{5}\bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\bigr)b^{\ell_{3}}\Bigr| \displaystyle\quad<\sqrt{5}(b-1)+\sqrt{5}(b-1)\alpha^{n_{3}-1}+2(b-1)\alpha^{-n_{3}} \displaystyle\quad<1.4(b-1)\alpha^{n_{3}}. Dividing both sides by (b-1)\alpha^{n_{1}}(1+\alpha^{n_{2}-n_{1}}) , we obtain (18) \displaystyle|\Lambda_{7}|=\Biggl|\frac{\sqrt{5}\bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{1}-d_{3})\bigr)}{(b-1)(1+\alpha^{n_{2}-n_{1}})}\alpha^{-n_{1}}b^{\ell_{3}}-1\Biggr|<1.4\,\alpha^{-(n_{1}-n_{3})}. Applying Theorem 3 to inequality ( 18 ) with parameters B=n_{1}+2,\quad\eta_{1}:=\frac{\sqrt{5}(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{1}-d_{3}))}{(b-1)(1+\alpha^{n_{2}-n_{1}})},\quad\eta_{2}:=\alpha,\quad\eta_{3}:=b,\quad b_{1}=1, b_{2}:=-n_{1},\quad b_{3}:=\ell_{3},\quad s:=3, and noting that h^{\prime}(\eta_{1})<1.85\times 10^{41}\,\log^{4}b\,\log^{3}(n_{1}+2) and |\Lambda_{7}|\neq 0 , we conclude that (n_{1}-n_{3})\log\alpha<8.7\times 10^{54}\,\log^{5}b\,\log^{4}(n_{1}+2). Step 3.8 . We consider Case 2 and show that under the assumption that \ell_{1}\log b<1.5\times 10^{14}\log^{2}b\log(n_{1}+2), we obtain that \min\{(n_{1}-n_{2})\log\alpha,\ell_{2}\log b\}<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2). Using ( 1 ) and ( 4 ), we obtain \displaystyle(b-1)\alpha^{n_{1}}-\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)b^{\ell_{2}+\ell_{3}} \displaystyle=-\sqrt{5}\left((d_{2}-d_{3})b^{\ell_{3}}+d_{3}\right) \displaystyle-\sqrt{5}(b-1)(F_{n_{2}}+F_{n_{3}})+(b-1)\beta^{n_{1}} \displaystyle\left|(b-1)\alpha^{n_{1}}-\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)b^{\ell_{2}+\ell_{3}}\right| \displaystyle<\sqrt{5}(b-1)\left(b^{\ell_{3}}+1\right) \displaystyle+2\sqrt{5}(b-1)\alpha^{n_{2}-1}+(b-1)\alpha^{-n_{1}} \displaystyle<1.5\sqrt{5}(b-1)b^{\ell_{3}}+2.77(b-1)\alpha^{n_{2}} \displaystyle<6.13(b-1)\max\{b^{\ell_{3}},\alpha^{n_{2}}\} Dividing both sides of the above inequality by (b-1)\alpha^{n_{1}} , we get \displaystyle\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{b-1}.\alpha^{-n_{1}}.b^{\ell_{2}+\ell_{3}}-1\right| \displaystyle<\max\left\{\frac{6.13b^{\ell_{3}}}{\alpha^{n_{1}}},\alpha^{-(n_{1}-n_{2})}\right\}. Using the fact that \frac{1}{\alpha^{n_{1}}}<\frac{1.86}{b^{\ell_{1}+\ell_{2}+\ell_{3}-1}} , we obtain (19) \displaystyle\left|\Lambda^{\prime}_{1}\right|=\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{b-1}.\alpha^{-n_{1}}.b^{\ell_{2}+\ell_{3}}-1\right|<5.7\max\left\{b^{-(\ell_{2}-1)},\alpha^{-(n_{1}-n_{2})}\right\} We apply Theorem 3 to inequality ( 19 ) by taking B=n_{1}+2 \eta_{1}:=\left(\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\right)}{b-1}\right),\eta_{2}:=\alpha,\eta_{3}:=b,b_{1}=1,b_{2}:=-n_{1},b_{3}:=\ell_{3},s:=3. In this case we have h^{\prime}(\eta_{1})<1.6\times 10^{14}\log^{2}b\log(n_{1}+2). and |\Lambda^{\prime}_{1}|\neq 0 . Therefore, we get \min\{(n_{1}-n_{2})\log\alpha,\ell_{2}\log b\}<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2). At this stage, we consider the following two sub-cases for Case 2: Case 2A: \min\{(n_{1}-n_{2})\log\alpha,\,\ell_{2}\log b\}=\ell_{2}\log b Case 2B: \min\{(n_{1}-n_{2})\log\alpha,\,\ell_{2}\log b\}=(n_{1}-n_{2})\log\alpha. We will analyze these sub-cases in the subsequent steps. Step 3.9 . We consider Case 2A and show that under the assumption that \ell_{1}<1.5\times 10^{14}\log^{2}b\log(n_{1}+2)\text{ and }\ell_{2}\log\alpha<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2), we obtain that (n_{1}-n_{2})\log\alpha<1.76\times 10^{41}\log^{4}b\log^{3}(n_{1}+2). Using ( 1 ) and ( 4 ), we obtain \displaystyle(b-1)\alpha^{n_{1}} \displaystyle-\sqrt{5}\Bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\Bigr)b^{\ell_{3}} \displaystyle=-\sqrt{5}\,d_{3}-\sqrt{5}(b-1)(F_{n_{2}}+F_{n_{3}})+(b-1)\beta^{n_{1}}, \displaystyle\Bigl|(b-1)\alpha^{n_{1}} \displaystyle-\sqrt{5}\Bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\Bigr)b^{\ell_{3}}\Bigr| \displaystyle<\sqrt{5}(b-1)+2\sqrt{5}(b-1)\alpha^{n_{2}-1}+(b-1)\alpha^{-n_{1}}, \displaystyle<2.77(b-1)\alpha^{n_{2}}. Dividing both sides of the above inequality by (b-1)\alpha^{n_{1}} , we get (20) \displaystyle\left|\Lambda^{\prime}_{2}\right|=\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\right)}{b-1}.\alpha^{-n_{1}}.b^{\ell_{3}}-1\right|<2.77\alpha^{-(n_{1}-n_{2})} We apply Theorem 3 to inequality ( 20 ) by taking B=n_{1}+2 \eta_{1}:=\left(\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\right)}{b-1}\right),\eta_{2}:=\alpha,\eta_{3}:=b,b_{1}=1,b_{2}:=-n_{1}, b_{3}:=\ell_{3},s:=3. In this case we have h^{\prime}(\eta_{1})<3.75\times 10^{27}\log^{3}b\log^{2}(n_{1}+2) and |\Lambda^{\prime}_{2}|\neq 0 . Therefore, we get (n_{1}-n_{2})\log\alpha<1.76\times 10^{41}\log^{4}b\log^{3}(n_{1}+2). Step 3.10 . We consider Case 2A and show that under the assumption that \ell_{1}<1.5\times 10^{14}\log^{2}b\log(n_{1}+2),\,\,\,\ell_{2}\log\alpha<7.48\times 10^{27}\log^{3}b\log^{2}(n_{1}+2)\text{ and } (n_{1}-n_{2})\log\alpha<1.76\times 10^{41}\log^{4}b\log^{3}(n_{1}+2), we obtain that (n_{1}-n_{3})\log\alpha<8.27\times 10^{54}\log^{5}b\log^{4}(n_{1}+2). Using ( 1 ) and ( 4 ), we obtain \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}} \displaystyle-\sqrt{5}\Bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\Bigr)b^{\ell_{3}} \displaystyle=-\sqrt{5}\,d_{3}-\sqrt{5}(b-1)F_{n_{3}}+(b-1)(\beta^{n_{1}}+\beta^{n_{2}}), \displaystyle\Bigl|(b-1)\alpha^{n_{1}}\bigl(1+\alpha^{n_{2}-n_{1}}\bigr) \displaystyle-\sqrt{5}\Bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\Bigr)b^{\ell_{3}}\Bigr| \displaystyle<\sqrt{5}(b-1)+\sqrt{5}(b-1)\alpha^{n_{3}-1}+2(b-1)\alpha^{-n_{3}}, \displaystyle<1.4(b-1)\alpha^{n_{3}}. Dividing both sides of the above inequality by (b-1)\alpha^{n_{1}}\left(1+\alpha^{n_{2}-n_{1}}\right) , we get (21) \displaystyle|\Lambda^{\prime}_{3}|=\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{1}-d_{3})\right)}{(b-1)\left(1+\alpha^{n_{2}-n_{1}}\right)}.\alpha^{-n_{1}}.b^{\ell_{3}}-1\right|<1.4.b^{-(n_{1}-n_{3})}. and therefore (n_{1}-n_{3})\log\alpha<8.27\times 10^{54}\log^{5}b\log^{4}(n_{1}+2). Step 3.11 . We consider Case 2B and prove that, under the assumptions \ell_{1}<1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2)\quad\text{and}\quad(n_{1}-n_{2})\log\alpha<7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2), one obtains the bound \min\{(n_{1}-n_{3})\log\alpha,\ \ell_{2}\log b\}<3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2). We note that Step 3.11 is analogous to Step 3.5 , with the roles of \ell_{1}\log b and (n_{1}-n_{2})\log\alpha interchanged. We now divide Case 2B into the following sub-subcases. Case 2B-A: \min\{(n_{1}-n_{3})\log\alpha,\ \ell_{2}\log b\}=(n_{1}-n_{3})\log\alpha Case 2B-B: \min\{(n_{1}-n_{3})\log\alpha,\ \ell_{2}\log b\}=\ell_{2}\log b. These sub-subcases will be treated in the subsequent steps. By applying the same arguments as in Steps 3.6 and 3.7 , we obtain the results stated in Steps 3.12 and 3.13 . Step 3.12 . We consider Case 2B-A and prove that, under the assumptions \ell_{1}\log b<1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2),\qquad(n_{1}-n_{2})\log\alpha<7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2), and (n_{1}-n_{3})\log\alpha<3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2), one obtains \displaystyle\ell_{2}\log b<8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2). We apply again Theorem 3 to derive an upper bound for \ell_{2}\log b . The argument is entirely analogous to Case 1B-A. In particular, we obtain \displaystyle|\Lambda_{6}|=\left|\left(\frac{(b-1)\bigl(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\bigr)}{\sqrt{5}\bigl(d_{1}b^{\ell_{1}}-(d_{1}-d_{2})\bigr)}\right)\alpha^{n_{1}}b^{\ell_{2}+\ell_{3}}-1\right|<1.6\,b^{-\ell_{2}+1}, with the same parameter setting as in Case 1B-A. Consequently, Theorem 3 yields \displaystyle\ell_{2}\log b<8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2). Step 3.13 . We consider Case 2B-B and prove that, under the assumptions \ell_{1}\log b<1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2),\qquad(n_{1}-n_{2})\log\alpha<7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2), and \ell_{2}\log b<3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2), one obtains \displaystyle(n_{1}-n_{3})\log\alpha<8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2). We again apply Theorem 3 to derive an upper bound for (n_{1}-n_{3})\log\alpha . The argument is entirely analogous to that of Case 1B-B. In particular, we obtain \displaystyle\left|\frac{\sqrt{5}\bigl(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{1}-d_{3})\bigr)}{(b-1)\bigl(1+\alpha^{n_{2}-n_{1}}\bigr)}\alpha^{-n_{1}}b^{\ell_{3}}-1\right|<1.4\,\alpha^{-(n_{1}-n_{3})}, with the same parameter setting as in Case 1B-B. Consequently, Theorem 3 yields \displaystyle(n_{1}-n_{3})\log\alpha<8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2). Table 4 summarizes the bounds obtained so far. Upper bound of Case 1A Case 1B-A Case 1B-B (n_{1}-n_{2})\log\alpha 1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2) 1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2) 1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2) (n_{1}-n_{3})\log\alpha 7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2) 8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2) 3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2) \ell_{1}\log b 1.7\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2) 7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2) 7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2) \ell_{2}\log b 8.27\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2) 3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2) 8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2) Upper bound of Case 2A Case 2B-A Case 2B-B (n_{1}-n_{2})\log\alpha 1.76\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2) 7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2) 7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2) (n_{1}-n_{3})\log\alpha 8.27\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2) 3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2) 8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2) \ell_{1}\log b 1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2) 1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2) 1.5\times 10^{14}\log^{2}b\,\log(n_{1}+2) \ell_{2}\log b 7.48\times 10^{27}\log^{3}b\,\log^{2}(n_{1}+2) 8.7\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2) 3.6\times 10^{41}\log^{4}b\,\log^{3}(n_{1}+2) Table 4. Summary of the upper bounds obtained in all cases Step 3.14 . Assuming that the bounds given in Table 4 hold, we prove that n_{1}<1.51\times 10^{82}\log^{11}b. Proof. Using ( 1 ) and ( 4 ), we obtain \displaystyle(b-1)\alpha^{n_{1}}+(b-1)\alpha^{n_{2}}+(b-1)\alpha^{n_{3}}-\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\right)b^{\ell_{3}} \displaystyle\hskip 17.00024pt=-\sqrt{5}d_{3}+(b-1)(\beta^{n_{1}}+\beta^{n_{2}}+\beta^{n_{3}}), \displaystyle\Bigl|(b-1)\alpha^{n_{1}}\bigl(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\bigr)-\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\right)b^{\ell_{3}}\Bigr| \displaystyle\hskip 17.00024pt<\sqrt{5}(b-1)+3(b-1)\alpha^{-n_{3}}<2.24(b-1). Dividing both sides of the above inequality by (b-1)\alpha^{n_{1}}\bigl(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\bigr), we obtain (22) \displaystyle|\Lambda|=\left|\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\right)}{(b-1)\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)}\alpha^{-n_{1}}b^{\ell_{3}}-1\right|<2.24\,\alpha^{-n_{1}}. We now apply Theorem 3 to inequality ( 22 ) with B=n_{1}+2,\quad\eta_{1}:=\frac{\sqrt{5}\left(d_{1}b^{\ell_{1}+\ell_{2}}-(d_{1}-d_{2})b^{\ell_{2}}-(d_{2}-d_{3})\right)}{(b-1)\left(1+\alpha^{n_{2}-n_{1}}+\alpha^{n_{3}-n_{1}}\right)}, \eta_{2}:=\alpha,\qquad\eta_{3}:=b,\qquad b_{1}=1,\quad b_{2}:=-n_{1},\quad b_{3}:=\ell_{3},\quad s:=3. In this case we have h^{\prime}(\eta_{1})<8.28\times 10^{54}\log^{5}b\,\log^{4}(n_{1}+2),\qquad|\Lambda|\neq 0. Therefore, n_{1}\log\alpha<3.86\times 10^{68}\log^{6}b\,\log^{5}(n_{1}+2). Using the inequality \log(n_{1}+2)<2\log n_{1} , we obtain n_{1}<1.6\times 10^{69}\log^{6}b\,\log^{5}n_{1}. Next, we apply Lemma 1 with parameters \ell:=5,\qquad H:=1.6\times 10^{69}\log^{6}b,\qquad L:=n_{1}. Hence, n_{1}<2^{5}\times 1.6\times 10^{69}\log^{6}b\left(159.35+6\log\log b\right)^{5}. This implies n_{1}<3.09\times 10^{82}\log^{11}b. Since b\geq 2 , we have 159.35+6\log\log b<227\log b, and the desired bound follows. Therefore, Theorem 1 is proved. ∎ 4. Proof of theorem 2 In the final step, we reduce the large upper bound for n_{1} obtained in Theorem 1 by repeatedly applying Lemma 2 . More precisely, we explicitly determine all sums of three Fibonacci numbers that are concatenations of three repdigits in base b , for 2\leq b\leq 10 . Within this range, we deduce that every solution of equation ( 3 ) satisfies n_{1}<3\times 10^{86}. We first consider inequality ( 7 ) and recall that \Gamma_{1}=\log\left(\frac{b-1}{d_{1}\sqrt{5}}\right)+n_{1}\log\alpha-(\ell_{1}+\ell_{2}+\ell_{3})\log b. For technical reasons, we assume that \min\{n_{1}-n_{2},\;n_{1}-n_{3},\;\ell_{1},\;\ell_{2}\}\geq 24. If this condition is not satisfied, we proceed according to the following cases: \bullet If n_{1}-n_{2}<24 while \ell_{1},\ell_{2},n_{1}-n_{2}\geq 24 , then we consider inequality ( 10 ), that is, we move to Step 3.2 . \bullet If n_{1}-n_{2}<24 and n_{1}-n_{3}<24 while \ell_{1},\ell_{2}\geq 24 , then we consider inequality ( 13 ), that is, we move to Step 3.3 ; afterwards, we consider inequality ( 15 ), that is, we proceed to Step 3.4 . \bullet If n_{1}-n_{2}<24 and \ell_{1}<24 while \ell_{2},n_{1}-n_{3}\geq 24 , then we consider inequality ( 16 ), that is, we move to Step 3.5 , and afterwards proceed to Step 3.11 . \bullet If n_{1}-n_{2}<24 , n_{1}-n_{3}<24 , and \ell_{1}<24 while \ell_{2}\geq 24 , then we consider inequality ( 17 ), that is, we move to Step 3.6 , and afterwards proceed to Step 3.12 . \bullet If n_{1}-n_{2}<24 , \ell_{1}<24 , and \ell_{2}<24 while n_{1}-n_{3}\geq 24 , then we consider inequality ( 18 ), leading to Step 3.7 and Step 3.13 . \bullet If \ell_{1}<24 while n_{1}-n_{2},n_{1}-n_{3},\ell_{2}\geq 24 , then we consider inequality ( 19 ), which leads to Step 3.8 . \bullet If \ell_{1}<24 and \ell_{2}<24 while n_{1}-n_{2},n_{1}-n_{3}\geq 24 , then we consider inequality ( 20 ), that is, we move to Step 3.9 ; afterwards, we consider inequality ( 21 ), that is, we proceed to Step 3.10 . \bullet If all n_{1}-n_{2},n_{1}-n_{3},\ell_{1},\ell_{2}<24 , then we consider inequality ( 22 ), which leads to Step 3.14 . Step 4.1 . We prove that \ell_{1}\leq 301 or n_{1}-n_{2}\leq 438 . We begin by considering inequality ( 7 ). Since we assume that \min\{n_{1}-n_{2},n_{1}-n_{3},\ell_{1},\ell_{2}\}\geq 24 , we obtain \left|\Lambda_{1}\right|=\left|\mathrm{e}^{\Gamma_{1}}-1\right|<\frac{1}{4} , hence \left|\Lambda_{1}\right|<\frac{1}{2} . Moreover, since the inequality \left|x\right|<2\left|\mathrm{e}^{x}-1\right| holds for all x such that \left|x\right|<\frac{1}{2} , we deduce that \left|\Gamma_{1}\right|<15.7b\max\{b^{-\ell_{1}},\alpha^{-(n_{1}-n_{2})}\}. Consequently, we obtain 0<\left|(\ell_{1}+\ell_{2}+\ell_{3})\left(\frac{\log b}{\log\alpha}\right)-n_{1}+\frac{\log\left(\frac{d_{1}\sqrt{5}}{b-1}\right)}{\log\alpha}\right|<\max\left\{32.63b^{-(\ell_{1}-1)},\,32.63b\,\alpha^{-(n_{1}-n_{2})}\right\}. To apply Lemma 2 to the above inequality, we choose \tau:=\frac{\log b}{\log\alpha},\quad\mu:=\frac{\log\left(\frac{d_{1}\sqrt{5}}{b-1}\right)}{\log\alpha}, (A,B)=(32.63,b)\ \text{or}\ (32.63b,\alpha),\quad w:=\ell_{1}-1\ \text{or}\ n_{1}-n_{2},\quad\text{with }1\leq d_{1}\leq b-1. Since \ell_{1}+\ell_{2}+\ell_{3}6M does not fulfill the condition \varepsilon>0 , we consider the next convergent until the condition is satisfied. Thus we obtain Table 5 . Table 5. b 2 3 4 5 6 7 8 9 10 q_{t} q_{172} q_{164} q_{176} q_{171} q_{178} q_{170} q_{162} q_{158} q_{175} \varepsilon\geq 0.373 0.264 0.254 0.337 0.268 0.283 0.051 0.182 0.436 \ell_{1}-1 300 189 150 129 116 107 101 95 90 n_{1}-n_{2}\leq 434 434 434 432 433 434 438 435 435 Therefore, 1\leq\ell_{1}\leq\frac{\log\left(32.63q_{172}/0.373\right)}{\log 2}+1\leq 301, or 1\leq n_{1}-n_{2}\leq\frac{\log\left(261.04q_{162}/0.051\right)}{\log\alpha}+1\leq 438. Consequently, we distinguish the following cases: Case 1: n_{1}-n_{2}\leq 438 Case 2: \ell_{1}\leq 301. Step 4.2 . We consider Case 1 and prove that, under the assumption n_{1}-n_{2}\leq 438 , one has n_{1}-n_{3}\leq 456 or \ell_{1}\leq 308 . In this step, we consider inequality ( 10 ) and assume that \ell_{1},\,n_{1}-n_{3}\geq 24 . Recall that \Gamma_{2}=\log\left(\frac{(b-1)\left(1+\alpha^{\,n_{2}-n_{1}}\right)}{\sqrt{5}\,d_{1}}\right)+n_{1}\log\alpha-(\ell_{1}+\ell_{2}+\ell_{3})\log b. From inequality ( 10 ) we obtain |\Gamma_{2}|<19.86b\max\left\{b^{-\ell_{1}},\,\alpha^{-(n_{1}-n_{3})}\right\}. Hence, (23) \displaystyle 0< \displaystyle\left|(\ell_{1}+\ell_{2}+\ell_{3})\left(\frac{\log b}{\log\alpha}\right)-n_{1}+\frac{\log\!\left(\frac{\sqrt{5}\,d_{1}}{(b-1)\left(1+\alpha^{-(n_{1}-n_{2})}\right)}\right)}{\log\alpha}\right| \displaystyle<\max\left\{41.28\,b^{-(\ell_{1}-1)},\,41.28\,b\,\alpha^{-(n_{1}-n_{3})}\right\}. Since the hypotheses of Lemma 2 are satisfied, we apply it to inequality ( 23 ) with \tau:=\frac{\log b}{\log\alpha},\qquad\mu:=\mu_{k}=\frac{\log\!\left(\frac{\sqrt{5}\,d_{1}}{(b-1)\left(1+\alpha^{-k}\right)}\right)}{\log\alpha}, (A,B)=(41.28,b)\ \text{or}\ (41.28b,\alpha),\qquad w:=\ell_{1}-1\ \text{or}\ n_{1}-n_{3}, with 1\leq d_{1}\leq b-1 and k=n_{1}-n_{2}\in[0,438] . Since \ell_{1}+\ell_{2}+\ell_{3}