Diophantine equations with products of consecutive terms in Lucas sequences (Q2569017): Difference between revisions

Let \(r\), \(s\) be nonzero integers with \(r^2+4s\neq 0\), let \(\alpha\) and \(\beta\) (with \(| \alpha| \geq | \beta| \)) be the two roots of the equation \(x^2-rx-s=0\). Then the Lucas sequences of the first and second kind are denoted by \(u_n=(\alpha^n-\beta^n)/(\alpha-\beta)\) and \(v_n=\alpha^n+\beta^n\), respectively. If \(r+4s\neq 0\) and \(\gamma\) and \(\delta\) denote the roots of \(x^2-\sqrt rx-s=0\), then the corresponding Lehmer sequence is defined by \(w_n=(\gamma^n-\delta^n)/(\gamma-\delta)\) for odd \(n\) and by \((\gamma^n-\delta^n)/(\gamma^2-\delta^2)\) for even \(n\). In what follows, \(n\), \(d\), \(k\), \(m\), and \(y\) will always denote positive integers with \(m\geq 2\), \(y>1\), \(\gcd(n,d)=1\) and \(b\) will denote a nonzero integer. The paper under review studies the Diophantine equation \(u_n u_{n+d} \cdots u_{n+(k-1)d}=by^m\) (called \((D)\) in the sequel) in the unknowns \((n,d,k,b,y,m)\). Under the additional assumption that the largest prime factor of \(b\) is at most \(k\) (or at most \(2k\), if \(d>1\)), the authors prove that for every solution of \((D)\), the number \(k\) is bounded by an effectively computable number depending only on the sequence \(u_n\). Under the additional assumption that the largest prime factor of \(b\) is bounded by some positive constant \(P\), it is proved that \(\max\{n,d,k,| b| ,y,m\}\) is bounded by some effectively computable constant depending only on \(r\), \(s\), and \(P\). The two above results are also valid if the sequence \(u_n\) is replaced by the sequences \(v_n\) or \(w_n\). For the specific case that the \(u_n\) are the Fibonacci numbers, it is shown that \((D)\) has no solution with \(b=1\), \(n>1\), and \(k\geq 2\). Similarly, for an integer \(x>1\) and \(u_n=(x^n-1)/(x-1)\), there are no solutions of \((D)\) with \(b=1\), \(n>1\), \(k\geq 2\), and odd \(d\).