On an assertion of Riemann concerning the distribution of prime numbers (Q1187334)

The author discusses various conjectures and results about the sign and sign changes of \(\pi(x)-Li(x)\), where \(Li(x)\) is the usual integral logarithm. The Riemann hypothesis is equivalent to the integral from 2 to \(X\) of this difference being negative for \(X>X_ 0\). The author proves without any assumption that \[ \int^ \infty_ 1[\pi(x)- Li(x)]\exp\left(-{\log^ 2x\over y}\right)dx<-{c_ 2\over y}\exp\left({9\over 16} y\right)\text{ for }y>c_ 1, \] where \(c_ 1\) and \(c_ 2\) are explicitly calculable positive absolute constants. Thus in this sense, \(\pi(x)-Li(x)\) is negative on the average.