On Ramanujan's \(Q\)-function (Q1894389)

The main result proved in the paper solves a problem posed by Ramanujan in 1913 saying that \(e^n/2\) permits a representation of the form \[ {1\over 2} e^n = \Theta (n) {n^n \over n!} + \sum^{n - 1}_{\nu = 0} {n^\nu \over \nu!}\qquad \text{with}\qquad \Theta (n) = {1 \over 3} + {4 \over 135 \bigl( n + k(n) \bigr)} \] and \(k(n)\) lying between \(8/45\) and \(2/21\). In particular, this implies \(\Theta (n) \in [1/3,1/2]\). In order to prove this result, the study provides a detailed analysis of the \(Q\)-function given by \[ Q(n) = {n! \over n^n} \sum^{n - 1}_{\nu = 0} {n^\nu \over n!}. \] This function also surfaces in the analysis of a variety of algorithms and in several discrete probability problems.