Differential calculus for linear operators represented by finite signed measures and applications (Q2434154)

Derivatives of linear operators have different forms depending on the differentiability requirements on the functions under consideration. Take the classical Szász operator \(L\) defined as \[ L\phi(t) := \sum^{\infty}_{k=0} \phi(k)\frac{e^{-t}t^{k}}{k!},\quad t\geq 0, \tag{1} \] where \(\phi\) is any real function defined on \([0, \infty )\) for which the preceding series makes sense. Let \(0\leq s\leq t.\) If \(\phi\) is differentiable, it is well known that \[ L\phi(t)-L\phi(s)=\int^{t}_{s} D\phi'(u)du, \] where \(D\) is the Szász-Kantorovich operator defined by \[ Df(t):=\sum^{\infty}_{k=0}\frac{e^{-t}t^{k}}{k!} \int^{1}_{0}f(k+x)dx. \tag{2} \] The aim of this paper is to extend the facts above to the framework of a differential calculus for linear operators represented by a family of finite signed measures, that is, linear operators \(L \colon \mathcal{M}(J)\rightarrow \mathcal{M}(I)\) having the form \[ L\phi(t):=\int_{J}\phi(x)d\mu_{t}(x),\quad t\in I,\, \phi \in \mathcal{M}(J), \] where \(I\) and \(J\) are two real intervals, \(\mathcal{M}(J)\) stands for the set of measurable functions on \(J\) and \((\mu_{t})_t\in I\) is a family of finite signed measures on \(J\). Having in mind differentiation formulas such as (1) and (2), the author introduces the notion of a \(g\)-differentiable linear operator, \(g\) being a right-continuous nondecreasing function on \(J\). Depending on the choice of \(g\), the differential calculus works for non-smooth functions and under weak integrability conditions. For linear operators represented by stochastic processes, the author provides a characterization criterion of \(g\)-differentiability in terms of characteristic functions of the random variables involved. The author also considers various illustrative examples. As an application, the author obtains an efficient algorithm to compute the Riemann zeta function \(\zeta(z)\) with a geometric rate of convergence which improves exponentially as \(\mathfrak{R}(z)\) increases.