Ideas from continued fraction theory extended to Padé approximation and generalized iteration (Q1591307)

A continued fraction \(b_0+K(a_n/b_n)\) where \(a_n, b_n\) are polynomials in \(z\in\mathbb C\), is said to correspond to a formal power series \(L(z)=\sum_{n=0}^\infty c_nz^n\) at the origin if its approximants \(S_k(0)\) has Maclaurin expansions of the form \(\sum_{n=0}^{n_k} c_nz^n+\dots\) where \(n_k\to\infty\) as \(k\to\infty\). If \(n_k\) is large enough, then \(S_k(0)\) is also a Padé approximant to \(L\) by definition. As well documented, spurious poles of \(S_k(0)\) can create problems in Padé approximation. In the continued fraction theory, such problems have to a certain extent been solved by the introduction of modified approximants and general convergence. In the present paper these concepts are applied in the setting of Padé approximation, and some of their advantages are explored.