Smooth \(L^2\) distances and zeros of approximations of Dedekind zeta functions (Q2408133)

Let \(s=\sigma +it\). Gonek and Montgomery, building on previous work by Spira, showed that for \(X^{1/2+\epsilon}\ll t\ll X\), \(\zeta_X(1/2+it)\) approximates \(2\zeta(1/2+it)\), where \(\zeta_X(s)\) is the family of approximations given by \(\zeta_X(s) = F_X(s) + \chi(s)F_X(1-s)\), \(F_X(s) = \sum_{n\le X} n^{-s}\), and \(\chi(s) = \pi^{s-1/2} \Gamma((1-s)/2)/\Gamma(s/2)\). For larger \(t\), of size around \(X^2\), and \(0<\sigma<1\), it follows from the approximate functional equation that \(\zeta_X(s)\) approximates \(\zeta(s)\). Additionally, Spira showed that \(\zeta_1(s)\) and \(\zeta_2(s)\) satisfy the Riemann hypothesis, and demonstrated numerically that \(\zeta_3(s)\) does not. On the other hand, Gonek and Montgomery showed that almost all the zeros of \(\zeta_X(s)\) lie on the critical line and are simple. In the paper under review, the authors present similar results for Dedekind zeta functions, and consider additional questions related to the \(L^2\)-norm of these approximations to Dedekind zeta functions. Le \(K/\mathbb{Q}\) be a number field, \(\zeta_{K,X}(s) = F_{K,X}(s)+B(s)F_{K,X}(1-s)\), \(\zeta_K(s) = \sum_{n\ge 1} a(n)n^{-s}\) for \(\sigma >1\) the Dedekind zeta function of \(K\), \(B(s)\) the usual factor in the functional equation for \(\zeta_K(s)\), and \(F_{K,X}(s) = \sum_{n\le X} a(n)n^{-s}\) which is the analogue of \(F_X(s)\) in the case of zeta. If \(K/\mathbb{Q}\) is a Galois extension of degree \(n_0\), the authors establish for example that for \(T^{\epsilon}\le N\le M\le T^{n_0-\epsilon}\) and \(h(t)\) a suitable smooth compactly supported function, \begin{align*} \int_{-\infty}^{\infty} h(t/T) |\zeta_{K,N}(1/2+it)-\zeta_{K,M}(1/2+it)|^2\,dt\sim\\ 2T(\hat{h}(0)/n_0)C(K)(\log^{n_0} M-\log^{n_0} N), \end{align*} where \(C(K)\) is a certain constant that depends on the number field \(K\). Moreover, the authors study the nontrivial zeros of \(\zeta_{K,X}(s)\). They establish a zero-free region, give an equivalent condition for zeros to be on the critical line, and prove that for \(X\le T^{o(1)}\) and as \(T\to \infty\) almost all the zeros of \(\zeta_{K,X}(s)\) with imaginary part in \((0,T]\) lie on the critical line and are simple.