Tauberian constants for power series methods in summability (Q1260899)

A power series method of summability is generated by a fixed sequence \((p_ n)\), \(p_ 0>0\), \(p_ n\geq 0\) (\(n=1,2,\dots\)), such that \(p(t):=\sum_{n=0}^ \infty p_ n t^ n\) has radius of convergence \(R=1\) or \(R=\infty\). A sequence \((s_ n)\) is said to be summable by the method if \(p_ s(t):=\sum_{n=0}^ \infty p_ n s_ n t^ n\) converges for \(| t|<R\) and if \(\sigma_ s(t):= p_ s(t)/p(t)\to s\) when \(0<t\to R_ -\). The definition includes many well-known summability methods. This paper concerns Tauberian results, namely theorems of the form \([\sigma_ s(t)\to s\) and \((s_ n)\in T] \Rightarrow s_ n\to s\). The Tauberian set \(T\) is often defined by an order condition of the form \(a_ k:=s_ k-s_{k-1}= O(h_ k)\) and we can then look for optimal sequences \(t_ n\to R_ -\) yielding the smallest (Tauberian constant) \(A\geq 0\) such that \(\limsup_{n\to\infty} | \sigma_ s(t_ n)-s_ n|\leq A\limsup_{k\to\infty} | a_ k/h_ k|\). Tauberian constants for several classes of power series methods are given in the paper. For example, in Theorem 1 the generating sequence \((p_ n)\) is assumed to be of the form \(p_ n\sim n^ \beta L(n)\) where \(\beta>-1\) and \(L(\cdot)\) is a slowly varying function, and a parameter \(\alpha\in (0,\infty)\) is assumed such that \(n=n(\alpha)\to\infty\), \(c(\alpha)\to c>0\), \(t(\alpha):=1- c(\alpha)/n(\alpha)\to 1-\), as \(\alpha\to\infty\). Under these conditions, and with the Tauberian condition \(a_ k=O(1/k)\), Theorem 1 concludes that \[ \limsup_{\alpha\to\infty} |\sigma_ s(t(\alpha))- s_{n(\alpha)}|\leq A(c) \limsup_{k\to\infty} | ka_ k|, \] and an explicit value for \(A(c)\) is obtained. An alternative exponential form of the generating sequence \((p_ n)\) (which includes the Borel summability method, for example) leads to a second theorem, again with an explicit value for the Tauberian constant. Finally, the consequences are explored, of replacing an order-type Tauberian condition by an oscillation-type condition on \((s_ n)\).