On the extremal theory of continued fractions (Q270215): Difference between revisions

For a number \(x \in (0,1)\), let \(a_n(x)\) denote the \(n\)th partial quotient in the simple continued fraction expansion of \(x\) and let \(S_N(x) = \sum_{n=1}^N a_n(x)\). \textit{A. Khintchine} [Compos. Math. 1, 361--382 (1934; Zbl 0010.34101)] proved that \((N \log N)^{-1} S_N(x)\) converges to \((\log 2)^{-1}\) in measure, but not almost everywhere. Much later, \textit{H. G. Diamond} and \textit{J. D. Vaaler} [Pac. J. Math. 122, 73--82 (1986; Zbl 0589.10056)] proved that if one removes the largest summand in \(S_n(x)\), the convergence takes place almost everywhere. In the present paper, the authors consider the distribution of the sum \(S_n(x)\), but with the largest \(d_N\) summands removed, where \(d_N \rightarrow \infty\) and \(d_N/N \rightarrow 0\). The precise asymptotics of the \(d_N\) largest terms are determined, and it is shown that the sum of the remaining terms has an asymptotically Gaussian distribution.