Arithmetic progressions of zeros of the Riemann zeta function (Q2581380): Difference between revisions

It is widely believed that there are no arithmetic progressions of zeros of the Riemann zeta-function \(\zeta(s)\). In this paper the author proves the following result. Let \({1\over 2} \leq D < 1, \, p>0,\, N \geq 2\), and suppose that \(\zeta(D+inp)=0\) for \(0<| n| <N\). Then \(N<60({p\over 2\pi})^{{1\over D}-1}\log p\). Moreover, if \(D={1\over 2}\) then \(N<13 p\), and if \(D<0.96\) then \(N<80({p\over 2\pi})^{{1\over D}-1}\). For the proof the author defines \[ T(x)={p\over 2\pi}\sum_{| n| <N}\left(1-{| n| \over N}\right){x^{D+inp}-1\over D+inp} \quad \quad (x\geq 1), \] \[ \nu_{T}(x)=\sum_{k\leq x}T({x\over k}), \] and shows the explicit formula \[ \nu_{T}(x)={p\over 2\pi}\sum_{| n| <N}\left(1-{| n| \over N}\right)\left[{x\over 1-D-inp}+{x^{D+inp}\zeta(D+inp)\over D+inp}\right] + O(1). \] It is seen that \(T(x)\) is increasing function, and it is sharply increasing at \(x=e^{{2\pi m\over p}}\) (\(m\): a positive integer), if \(N\) is large. Clearly \(\nu_{T}(x)\) increases at least as sharply as \(T(x)\) at \(x=e^{{2\pi m\over p}}\). If \(\zeta(D+inp)=0\) for all \(n\), \(0<| n| <N\), the explicit formula gives that the increase of \(\nu_{T}(x)\) at \(x=e^{{2\pi m\over p}}\) is small. Thus a bound for \(N\) is found.