A continual analog of continued fractions (Q1067123)

G is a set. With \(\tau_ 1,...,\tau_ i\in G\), \(\tau^ i\) is the set \(\{\tau_ 1,...,\tau_ i\}\) and \(G^ i\) is the space of \(\tau^ i\) \((1\leq i<\infty)\). Integrals are in the sense of Lebesgue and over G in all cases. \({\mathbb{C}}\) is the complex plane. With \(b_ 0\in {\mathbb{C}}\) and \(a_ i,b_ i: G^ i\to D\) where D is a prescribed subset of points in \({\mathbb{C}}\) \[ C_ n=b_ 0+\int \frac{a_ 1(\tau^ 1)d\tau_ 1}{b_ 1(\tau^ 1)+}...\int \frac{a_ n(\tau^ n)d\tau \quad_ n}{b_ n(\tau^ n)} \] may formally be determined by a recursive process of division, integration and addition. The expansion \(b_ 0+\int G| a_ i/b_ i]\) of which the \(C_ n\) are convergents is an integral continued fraction. (For a more general presentation and further results and references, see \textit{Yu. R. Batyuk} and the second author [Dokl. Akad. Nauk Ukr. SSR, Ser. A 1984, No. 7, 6-8 (1984; Zbl 0571.40003)]. Conditions which suffice to ensure that the convergents \(C_ n\) are well defined and converge to a finite limit are given. Neither existence nor convergence is affected by the term \(b_ 0\) which we omit henceforth. The special form of \(\int G| a_ i/1]\) in which \(b_ i(\tau^ i)=1\) is denoted by \(\int G| a_ i/1]\) and that in which \(b_ i(\tau^ i)=Kd_ i(\tau^ i)\), K being constant, by \(\int G| a_ i/Kd_ i]\). \(\mu\) (G) is the measure of the set G. The notations \([\tau^ i\in G^ i]\) and \((\tau^ i\in G^ i)\) indicate that an associated condition holds for all \(\tau^ i\in G^ i\) and for almost all \(\tau^ i\in G^ i\) respectively. A sum such as \(\sum M_ i\) denotes \(\sum^{\infty}_{i=1}M_ i.\) (A) Let \(0<\mu(G)<\infty\) and \(c_ i: G^ i\to {\mathbb{R}}\) \((1\leq i<\infty)\). If either (i) \(0<c_ i(\tau^ i)\leq M_ i\) \((\tau^ i\in G^ i\); \(1\leq i<\infty)\) and \(\sum M_ i<\infty\) or (ii) \(m_{i+1}\leq c_ i(\tau^ i)c_{i+1}(\tau^{i+1})\) \((\tau^{i+1}\in G^{i+1}\); \(1\leq i<\infty)\) and \(\sum m_{i+1}=\infty\), then \(\int G| c_ i/1]\) converges. (B) With \(a_ i,b_ i: G^ i\to {\mathbb{R}}\), let \(\int \{a_{i+1}(\tau^{i+1})/b_{i+1}(\tau^{i+1})\}d\tau_{i+1}\leq 1\) and \(b_ i(\tau^ i)\geq d>0\) \([\tau^ i\in G^ i;1\leq i<\infty]\). \(\int G| a_ i/b_ i]\) converges. (C) Let \(0<\mu(G)<\infty\). With \(a_ i,d_ i: G^ i\to {\mathbb{C}}\), let \(| d_ i(\tau^ i)| \geq 1+| a_ i(\tau^ i)|\) \([\tau^ i\in G^ i;1\leq i<\infty]\). \(\int G| a_ i/\mu (G)d_ i]\) converges absolutely and, denoting its value or that of any of its convergents by y, \(| y| \leq 1.\) (D) With \(a_ i,b_ i: G^ i\to {\mathbb{C}}\), let \(| b_ i(\tau^ i)| \geq 1+\int | a_{i+1}(\tau^{i+1})| d\tau_{i+1}\) \((\tau^ i\in G^ i;1\leq i<\infty)\). \(\int G| a_ i/b_ i]\) converges absolutely and, with y as above, \(| y| \leq 1.\) (E) Let the mapping \(e_ i(z): G^ i\to {\mathbb{C}}\) be defined for all \(z\in D\subseteq {\mathbb{C}}\) and let \(| \int e_ i(z;\tau^ i)d\tau_ i| \leq \alpha \leq 1/4\) \((i=1\) and \([\tau^{i-1}\in G^{i-1};1\leq i<\infty])\). \(1/\{1+\int G| e_ i(z)/1]\}\) converges uniformly over d and, with y as above, \(| y-\eta^ 2(\eta^ 2-4\alpha^ 2)^{- 1}| \leq 2\alpha \eta (\eta^ 2-4\alpha^ 2)^{-1}\) where \(\eta =1+v(1-4\alpha)\). (The constant \(\alpha =1/4\) and the region \(| y- 4/3| \leq 2/3\) are natural in the sense that the former cannot be increased and the latter cannot be reduced in size.)