On the Diophantine equation \(px^2+q^{2m}=y^p\) (Q927720): Difference between revisions

The Diophantine equation \[ ax^2+c=by^n, \] where \(a, b, c, n\) are positive integers and \(n\geq 3\), has a very long and rich history. The literature is well furnished. In this paper, the author extends early results on the subject. See [Int. J. Math. Math. Sci. 25, 373--381 (2001; Zbl 0981.11010) and Tamkang J. Math. 31, 79--84 (2001; Zbl 0995.11026)]. She considers two primes \(p\), \(q\) and takes \(a=n=p\), \(b=1\), \(c=q^{2m}\), where \(m\) is a positive integer. First, she proves that if \(q=2\) the equation has no solution with \(x\) odd and \(p\not \equiv 3 \pmod 8\). Second, still with \(q=2\), she shows that the equation has no solution for all \(x, y\) and \(p>3\) if \(p\not \equiv 7 \pmod 8\). Finally, the author considers \(q>2\), \((x, y)=1\), \(m>0\) and proves that the equation has no solution if \(p > 5\), \(p\not \equiv 7 \pmod 8\), but when \(p=5\), it has two families of solutions given by \[ y=F_{3k},\; F_{3k+1}\;\; (\text{or}\;\; L_{3k},\; L_{3k+1}),\; k>1, \] where \(F_k\) is the \(k\)th term of the Fibonacci sequence and \(L_k\) is the \(k\)th term of the Lucas sequence. The proofs of all these results are based on a deep result of Bilu-Hanrot-Voutier on primitive divisors of Lehmer numbers. See [\textit{Y. Bilu, G. Hanrot, P. M. Voutier}, J. Reine Angew. Math. 539, 75--122 (2001; Zbl 0995.11010)]. Editorial remark: Unfortunately, the paper contains some errors pointed by out P. G. Walsh, particularly when \(p = 13\). For more details, see his review in MR2419187 (2009d:11057)]. Moreover, \textit{A. Laradji, M. Mignotte} and \textit{N. Tzanakis} found another error when \(p = 5\). This is the object of their paper [J. Number Theory 131, No. 9, 1575--1596 (2011)]. (revised version May 2011)