A priori error estimates of multidimensional \(g\)-fraction approximations (Q811723)

For the multidimensional \(g\)-fraction \[ \dfrac{1}{1} \begin{aligned}{}\\+\end{aligned} \sum_{i_1=1}^N\dfrac{g_{i(1)}z_{i_1}}{1} \begin{aligned}{}\\+\end{aligned} \dots \begin{aligned}{}\\+\end{aligned} \sum_{i_k=1}^N\dfrac{g_{i(k)}(1-g_{i(k-1)})z_{i_k}}{1} \begin{aligned}{}\\+\end{aligned} \dots,\tag{1} \] where \( z=(z_1,z_2,\dots,z_N)\in {\mathbb C}^N\), \(\sum_{i=1}^N z_i\neq -1\), \(g_{i(k)} = \pi_{i(k)}(1+\pi_{i(k)})^{-1}\), \(\pi_{i(k)}>0\), \(k = \overline{1,\infty}\), \(i_p = \overline{1,N}\), \(p = \overline{1, k}\), estimates of the convergence speed are established in some bounded domains.