The unreasonable effectualness of continued function expansions

From MaRDI portal



Abstract: Many generalizations of continued fractions, where the reciprocal function has been replaced by a more general function, have been studied, and it is often asked whether such generalized expansions can have nice properties. For instance, we might ask that algebraic numbers of a given degree have periodic expansions, just as quadratic irrationals have periodic continued fractions; or we might ask that familiar transcendental constants such as e or pi have periodic or terminating expansions. In this paper, we show that there exist such generalized continued function expansions with essentially any desired behavior.


Many generalizations of continued fractions having nice properties have been studied. In this paper the author shows that the \(f\)-function introduced by \textit{B. H. Bissinger} [Bull. Am. Math. Soc. 50, 868--876 (1944; Zbl 0060.16302)] can be chosen so that the expansions of prescribed real numbers can have essentially any desired behaviour. He proves the following. Theorem 1. For any two real numbers \(x\), \(y\in(0,1)\), there exists a function \(f\) such that the \(f\)-expansion of \(x\) is the same as the usual continued fraction expansion of \(y\). Theorem 2. There exists a function \(f\) such that the \(f\)-expansion of any rational or quadratic irrational terminates. Theorem 3. There exists a function \(f\) such that the \(f\)-expansion of a real number \(x\) is periodic if and only if \(x\) is a cubic irrational number. Theorem 4. There exists a function \(f\) such that, simultaneously for every integer \(d\geq 1\) a real number \(x\) is algebraic of degree \(d\) if and only if the \(f\)-expansion of \(x\) terminates with the integer \(d+1\). In all four theorems, the cardinality of the set of functions \(f\) satisfying the given property is that of the continuum. A surprising generalization of Theorem 4 is also mentioned. A particular one-parameter family of the power functions \(f_\alpha(x)=x^{-\alpha}\) (\(\alpha>0\)) is considered.











This page was built for publication: The unreasonable effectualness of continued function expansions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4658690)