Acceleration methods for series: a probabilistic perspective (Q347005)

In the paper under review, the authors apply a probabilistic approach to accelerate convergence of some well-known series. This approach is based on the differential calculus for the negative binomial process developed in [the first author, Acta Math. Hung. 138, No. 1--2, 44--82 (2013; Zbl 1289.60016)] and extends the classical Euler transformation formula. As a result, the authors obtain accelerated series representation for \(\log(1+a)\) and Catalan's constant \(K\) given by \[ \log(1+a)=\frac{4(a+2)}{a}\sum_{j=0}^{\infty}(-1)^jc(j)\left(\frac{a^2}{a^2+8(a+1)}\right)^{j+1} \] and \[ K=\frac{2}{3}\sum_{j=0}^{\infty}d(j)\frac{1}{3^j} \] where \(c(j)\) and \(d(j)\) are evaluated in terms of moments of two independent random variables having the uniform distribution on \([0,1]\). It is shown that \(c(j)\) and \(d(j)\) satisfy simple recurrence equations (with initial conditions \(c(0)=d(0)=1\)). A similar approach is applied to the values of alternating zeta function, Stieltjes constants, and Euler constant, which gives accelerated series with more involved coefficients. For the Stieltjes constants, the authors recover a series given by \textit{M. W. Coffey} [Analysis, München 30, No. 4, 383--409 (2010; Zbl 1235.11079)].