On a variant of Pillai's problem involving \(S\)-units and Fibonacci numbers (Q2161101): Difference between revisions

In the paper under review the author looks at integers \(c\) such that the Diophantine equation \(F_n-2^x\cdot 3^y=c\) has more than one integer solution \((n,x,y)\) with \(n\ge 2\). Using the subspace theorem, the author deduces that the set of such \(c\) is finite and extensive computations lead the author to conjecture the complete list of such \(c\). The extensive computations carried on reveal that the author's list is complete in the range \(|c|\le 10^{100}\). If one is more ambitious and asks for \(c\) such that the above equation has more than two solutions then the author determines all the positive \(c\) such that the equation has at least three solutions and all negative \(c\) such that the equation has at least four solutions. As a byproduct of his results he gets that the above equation has at most five solutions for any integer \(c\). The proofs use a variety of tools from Diophantine equations. For example, the existence of two solutions leads to an \(S\)-unit equation in four terms, which upon checking that the solutions of this \(S\)-unit equation are nondegenerate, the author can conclude by invoking the main finiteness result from the theory of nondegenerate solutions to \(S\)-unit equations. For the proof of the computational result the author shows using LLL reduction that if \((n,x,y)\) are such that \(|F_n-2^x3^y|<10^{100}\), then \(n\le 525\) and \(x,y\) are similarly bounded by rather small numbers after which the author performs a brute search computation in the remaining range. Finally, when three or four solutions are assumed for \(c>0\) (and \(c<0\), respectively), then the author uses effective methods such as the theory of linear forms in complex and \(p\)-adic logarithms and again LLL reduction to reduce the large bounds produced by the application of linear forms in logarithms. This is a very nice paper.