Large strings of consecutive smooth integers (Q651390): Difference between revisions

For positive integers \(n\) and \(k\) put \(\Pi_{n,k}=n(n+1)\cdots (n+k-1)\). By a theorem of Sylvester, \(\Pi_{n,k}\) is divisible by a prime \(p>k\) whenever \(n>k\). Following Erdős, put \(f(k)\) to be the smallest positive integer such that \(\Pi_{n,f(k)}\) is divisible by a prime \(p>k\) for all \(n>k\). Prior to this paper, values \(f(k)\) had been tabulated for all \(k\leq 178\). In this paper, the author extends this calculation to all \(k\leq 268\). The project involved searching for all ``smooth'' solutions \(y\) of a very large number of Pell type equations \(x^2-dy^2=1\). The author succeeds by combining in a clever way an ingenious method due to \textit{M. Bauer} and \textit{M. A. Bennett} [Math. Comput. 77, No. 264, 2455--2459 (2008; Zbl 1211.11102)] to reduce the number and size of such Pell equations, with a method used in a paper by the author and the reviewer to work with the so-called compact representation of a solution. This method in turn requires a subexponential algorithm of \textit{J. Buchmann} [Sémin. Théor. Nombres, Paris/Fr. 1988--89, Prog. Math. 91, 27--41 (1990; Zbl 0727.11059)] which in general is only conditional on the GRH, but the dependence on GRH can be removed by doing a certain amount of extra computations in case only smooth solutions are of interest. The new values found are \(f(k)=14\) for all \(k\in [179,222]\) and \(f(k)=16\) for all \(k\in [223, 268]\).