On the \((C,1)\) summability method of improper integrals (Q2513536): Difference between revisions

Given a real-valued continuous function \(f\) on \([0, \infty)\), the authors define \(s(x)=\int_0^x f(t)\,dt\) and \(\sigma(x)={1\over x}\int_0^x s(t)\,dt\). Here, \(\sigma(x)\) is called the \((C,1)\) mean of \(s(x)\) [\textit{G. H. Hardy}, Divergent series. 2nd ed. New York, NY: Chelsea (1991; Zbl 0897.01044)]. The Kronecker identity is defined by \(s(x)-\sigma(x)=\nu(x)\), where \(\nu(x)={1\over x}\int_0^x t f(t)\,dt\), here \(\nu(x)\) is called a generator of the integral \(\int_0^\infty f(t)\,dt\). It is easily seen that \(x{d\over dx}\sigma(x)=\nu(x)\). For each integer \(m\geq0\), the authors define \(\sigma_m(x)\) by \(\sigma_m(x)={1\over x}\int_0^x \sigma_{m-1}(t)\,dt\) for \(m\geq1\), \(\sigma_0(x)=\sigma(x)\), and define \(\nu_m(x)={1\over x}\int_0^x \nu_{m-1}(t)\,dt\) for \(m\geq1\), \(\nu_0(x)=\nu(x)\). The classical control modulo of \(s(x)=\int_0^x f(t)\,dt\) is defined by \(\omega_0(x)=xf(x)\) and the general control modulo of integer order \(m\geq1\) of \(s(x)\) is defined by \(\omega_m(x)=\omega_{m-1}(x)-\sigma(\omega_{m-1}(x))\). The authors note that the concept of the classical and general control modulo for sequences of real and complex numbers were introduced by \textit{M. Dik} [Math. Morav. 5, 57--94 (2001; Zbl 1046.40004)]. For a function \(f\), the authors define \[ \left(x{d\over dx}\right)_m f(x)=\left(x{d\over dx}\right)_{m-1}\left(x{d\over dx}f(x)\right)=x{d\over dx}\left(\left(x{d\over dx}\right)_{m-1}f(x)\right), \] where \((x{d\over dx})_0f(x)=f(x)\), and \((x{d\over dx})_1f(x)=x{d\over dx}f(x)\). The authors [Appl. Math. Lett. 24, No. 6, 891--896 (2011; Zbl 1221.40009)] proved that for each integer \(m\geq1\), \[ \omega_m(x)=\left(x{d\over dx}\right)_m \nu_{m-1}(x). \] A real-valued function \(s(x)\) is slowly oscillating in the sense of Stanojević [\textit{İ. Çanak} and \textit{Ü. Totur}, Math. Comput. Modelling 55, No.~3--4, 1558--1561 (2012; Zbl 1255.40006)] if \[ \lim_{\lambda\to1^+}\limsup_{x\to\infty}\max_{x\leq t \leq \lambda x}|s(t)-s(x)|=0. \tag{01} \] An equivalent reformulation of ({01}) can be given as follows: \[ \lim_{\lambda\to1^-}\limsup_{x\to\infty}\max_{\lambda x\leq t \leq x}|s(t)-s(x)|=0. \] The integral \(\int_0^\infty f(t)\,dt\) is \((C, 1)\) summable to a finite number \(L\) if \[ \lim_{x\to\infty}\sigma(x)=L \tag{02} \] exists. But there are \((C,1)\) summable integrals which fail to converge as improper integrals. In this paper, the authors' aim is to obtain the slow oscillation of \(s(x)\) from \((C,1)\) summability of \(\nu(x)\) and one-sided boundedness of the general control modulo of integer order \(m\geq1\) with respect to a function satisfying an appropriate condition. Furthermore, they recover convergence of \(\int_0^\infty f(t)\,dt\) as a corollary.