On the Diophantine equation \(x^2+5^a\cdot 11^b=y^n\) (Q617797)

The authors find all solutions of the title equation in integers \(a\geq 0,~b\geq 0,~n\geq 3\), \(x>0\) and \(y>0\) with \(x\) and \(y\) coprime except when \(xab\) is odd. When \(n=4\), the proof is elementary, while for \(n\geq 5\) with \(n\neq 6\), the proof uses the theory of primitive divisors of Lucas--Lehmer sequences. The hardest case is \(n=3\), which is reduced to finding all \(\{5,11\}\)-integral points on a collection of \(36\) elliptic curves. Magma dealt successfully with \(35\) of them. For the last one, the original equation is converted into a Thue--Mahler equation, which in turn is solved using linear forms in \(p\)-adic logarithms to get some huge bounds on the potential solutions, followed by LLL to reduce such bounds and a sieve to deal with the small solutions. When \(xab\) is odd, the original equation implies that the \(n\)th term of a certain (parametric) binary recurrence is free of primes larger than \(11\), but unfortunately the binary recurrences involved are not Lucas--Lehmer sequences, so one cannot apply the theory of primitive divisors to deduce that \(n\) must be very small.