Coincidences in generalized Fibonacci sequences (Q1946727)

Denote by \(F_n^{(k)}\) \((n\geq 2-k)\) the \(n\)th \(k\)-generalized Fibonacci number which is defined by \[ F_n^{(k)}=F_{n-1}^{(k)}+F_{n-2}^{(k)}+\cdots+F_{n-k}^{(k)} \] with \(F_{-(k-2)}^{(k)}=F_{-(k-3)}^{(k)}=\cdots=F_0^{(k)}=0\) and \(F_{1}^{(k)}=1.\) This notion generalizes the classical Fibonacci numbers as well as, for example, the Tribonacci and Tetranacci numbers. The authors show that the only nontrivial solutions of the Diophantine equation \(F_{n}^{(k)}=F_{m}^{(\ell)}\) in positive integers \(n, k, m, \ell\) with \(k> \ell\geq 2\) are \[ (n,k,m,\ell)\in \left\{(6,3,7,2),(11,7,12,3),(6,2,5,t)\right\} \] for all \(t\geq 4\). The method of proof is as follows. The starting point is to approximate \(F_n^{(k)}\) by means of the dominant root of its characteristic polynomial. In a second step, linear forms in logarithms of algebraic numbers (in Matveev's form) are used to establish bounds on \(m\) and \(n\) in terms of \(k\). For small \(k\), i.e., \(k\leq 1600\), the bounds are reduced by employing a lemma on continued fractions. This is a variation of a result due to Dujella and Pethő. For larger \(k\), besides the multiple use of the mentioned tools, the authors crucially use the fact that the dominant root can be determined with high accuracy, namely, it is always between \(2(1-2^{-k})\) and \(2\). The paper is very clearly written and shows a nice application of linear forms in logarithms coupled with the theory of continued fractions.