Perturbed Padé approximation with high accuracy. (Q1421273)

Perturbed Padé approximation was introduced by the author in [Int. J. Comput. Math. 74, No. 2, 247--253 (2000; Zbl 0953.41011)]. This paper deals with a special case. A perturbed approximant to \(f(x)=c_0+c_1x+\cdots\) is constructed starting from the Padé equations for an approximant of the form \(p_{1,1}(x)=(a_0+a_1x)/(1+b_1x)\). It is obtained by solving the equations \(a_0=c_0\), \(a_1=c_1+c_0b_1\) and \(c_1+c_1b_1=0\). If the zero in the last right hand side is replaced by some \(\varepsilon_n(x)\) which depends on \(n\), \(x\) and the given coefficients \(c_k\), then, solving for the coefficients of \(p_{1,1}\) leads to a perturbed approximant \(p_{1,1}\). Here, the perturbation is chosen such that the perturbed approximant \(p_{1,1}\) is equal to \(\sum_{l=0}^n c_kx^k\).