A note on the diophantine equation \(x^ 2+1=dy^ 4\) (Q1341239): Difference between revisions

The author proves the following interesting Theorem: Let \(d>0\) (not a perfect square), \(\varepsilon= u+v \sqrt{d}\), \(u,v>0\) be the least positive solution of Pell's equation \(x^ 2+1= dv^ 2\). If \(\varepsilon> 5\times 10^ 7\), then the title equation has at most one solution. This solution, if it exists, is found from the relation \(x+y^ 2 \sqrt{d}= \varepsilon^ \ell\), where \(\ell\) is the square-free part of \(v\). The proof is based on six elegant and not very technical lemmas and on a sharp lower bound for linear forms in two logarithms due to Mignotte and Waldschmidt. Remarks: (1) It would be very interesting to apply, instead of the above mentioned result of Mignotte and Waldschmidt, a recent one due to Laurent, Mignotte and Nesterenko [Formes linéaires en deux logarithmes et déterminants d'interpolation, preprint]. (2) The author's result is better than Ljunggren's, which says that ``if the fundamental unit in \(\mathbb{Z} [\sqrt{d}]\) does not coincide with that of \(\mathbb{Q} [\sqrt{d}]\), then the title equation has at most two solutions'', but, obviously, only under the assumptions on \(\varepsilon\) mentioned in the author's theorem; the author, therefore, exaggerates in claiming that ``\dots\ a stronger result, obtained by a simpler proof, is given''. (3) The author mentions a paper by \textit{R. Steiner} and \textit{N. Tzanakis} [J. Number Theory 37, No.2, 123-132 (1991; Zbl 0716.11016)] in which the explicit solution of \(x^ 2+1= 2y\) is given. Obviously, this equation does not fall in the class considered in the paper under review, because it does not fulfill the assumptions of the author's theorem. Therefore, the author at least exaggerates again in saying that Steiner and Tzanakis' solution ``\dots\ is only a special case of [his] formulas (15).