A generalization of continued fractions (Q640879)

Simple continued fractions are expressions of the form \(a_0+{1\over a_1+{\strut 1\over a_2+\cdots }}\), where \(a_i \in \mathbb{Z}\) and \(a_i \geq 1\) for \(i \geq 1\); the expansion may be finite or infinite. In this paper, the authors study what happens when the \(1\)'s in the numerators are replaced by a fixed but arbitrary positive integer \(N\). These expansions, which the authors call \(\roman{cf}_N\) expansions, have apparently been little studied. (Two obscure 19th-century French references which the authors missed are given on p.\ 185 of [\textit{C. Brezinski}, History of continued fractions and Padé approximants. Berlin etc.: Springer- Verlag (1991; Zbl 0714.01001)]. The authors give an algorithm to produce such expansions. In contrast with the case of simple continued fractions, the \(\roman{cf}_N\) expansion of a number is not unique, and in fact, every positive real number has infinitely many distinct expansions (for \(N\geq 3\)). The authors single out a particular kind of expansion, which maximizes the partial quotient chosen at each step; they call it the ``best'' expansion. The authors give a number of families of periodic expansions for quadratic irrationalities, but conjecture that such expansions are not always periodic. (One conjectured aperiodic example is the best \(\roman{cf}_7\) expansion of \(\sqrt{8}\).) The authors close with a discussion of Pell's equation in light of their continued fractions. Clearly, further study of these continued fractions is needed.