Distribution of values of a real function. Means, moments, and symmetry (Q1813540)

For a measurable function \(h: [0,\infty)\to\mathbb{R}\) the author defines the absolute asymptotic mean of order \(k\) \((k\in\mathbb{R})\) by \[ m_ k=m_ k(h)=\lim_{x\to\infty} {1\over x} \int_ 0^ x | h(t)|^ k dt \] and the signed asymptotic mean of order \(k\) \((k\in\mathbb{R})\) by \[ \mu_ k=\mu_ k(h)=\lim_{x\to\infty} {1\over x} \int_ 0^ x \text{sgn}(h(t))| h(t)|^ k dt \] (if they exist) as natural generalizations of the standard asymptotic mean of order \(k\) \[ M_ k=M_ k(h)=\lim_{x\to\infty} {1\over x} \int_ 0^ x (h(t))^ k dt. \] Similarly he defines the absolute moment and the signed moment of order \(k\) \((k\in\mathbb{R})\) for the distribution function \(D_ h\) of the values of \(h\). Under the assumption that \(D_ h\) exists the author shows the equivalence of moments of means. Further he proves that if the means of \(h\) exist on an infinite and bounded set \(L\) of values of \(k\) then they exist on the interval \(I=(0,\sup L)\). He derives the existence of \(D_ h\) and shows that on \(I\) the moments of \(D_ h\) coincide with the means of \(h\). Finally he gives sufficient conditions for \(D_ h\) to be symmetric. His results are applicable to the error terms related to Euler's function \(\varphi\) and the sum-of-divisors function \(\sigma\). A further application is given to functions represented by certain trigonometrical series. The class of these functions contains the remainder term of the prime number theorem provided the Riemann hypothesis is true.