Hurwitz zeta functions and Ramanujan's identity for odd zeta values (Q6112519)

The author calls \[ \frac{1}{2\pi i}\int_{(c)}\frac{\zeta(1-z)}{\cos \frac{\pi}{2} z}x^{-z}\,\mathrm{d}z=\frac{1}{e^{2\pi x}-1}, \quad x>0, 1<c<2 \tag{1} \] Ramanujan's kernel [\textit{E. C. Titchmarsh}, The theory of the Riemann zeta-function. 2nd ed., rev. by D. R. Heath-Brown. Oxford: Clarendon Press (1986; Zbl 0601.10026), p. 24]. The left-hand side is the hyperbolic cotangent function and has the partial fraction expansion \[ \frac{1}{e^{2\pi x}-1}=-\frac{1}{2}+\frac{1}{2\pi x}+\frac{x}{\pi}\sum_{n=1}^{\infty}\frac{1}{n^2+x^2}, \tag{2} \] which is known to be equivalent to the functional equation for the Riemann zeta-function. The author introduces its dual \[ \frac{1}{2\pi i}\int_{(c)}\frac{\zeta(1-z)}{\sin \frac{\pi}{2} z}x^{-z}\,\mathrm{d}z=\frac{1}{2\pi}\left(2\log x -\psi\left(ix \right)-\psi\left(-ix \right) \right), \quad \operatorname{Re}(x)>0, 1<c<2. \tag{3} \] This is to be compared with Kloosterman's kernel [\textit{E. C. Titchmarsh}, The theory of the Riemann zeta-function. 2nd ed., rev. by D. R. Heath-Brown. Oxford: Clarendon Press (1986; Zbl 0601.10026), p. 25, Equation 2.9.2] \[ \frac{1}{2\pi i}\int_{(c)}\frac{-\pi\zeta(1-z)}{\sin \pi z}x^{-z}\,\mathrm{d}z=\psi(x+1)-\log x, \quad |\arg x|<\pi, 0<c<1 \tag{4} \] used by \textit{A. Dixit} et al. [``Extended higher Herglotz functions. I: Functional equations'', Preprint, \url{arXiv:2107.02607}] in a similar context, yielding some similar results, e.g., their Theorem 3.2. The author introduces a two-parameter generalization of Ramanujan's kernel and its dual and obtain several Ramanujan type identities (well-known and involving the condition \(\alpha\beta=\pi^2, \alpha>0, \beta>0\)). For \(x>0, a\in \mathbb{C}, k\in \mathbb{N}\), (1) is generalized as in Definition 1.2 to \[ \Psi(x,a;k)=\frac{1}{2\pi i}\int_{(c)}\frac{\zeta(1-z,a)}{2k\cos \frac{\pi}{2}\frac{z-1+k}{k}}x^{-z}\,\mathrm{d}z, \quad x>0, 1<c<2 \tag{5} \] and \[ \Psi_\alpha(x,a;k)=\Psi\left(\frac{\alpha}{\pi}x,a;k\right) \tag{6} \] is used. Here \(\zeta(s,a)\) is the Hurwitz zeta-function (\(a\in \mathbb{C}\) is not a negative integer). The key is the contour integral (\(N\in \mathbb{N}\)) \[ Z(\alpha,a,b;k,N)=\frac{1}{2\pi i}\int_{(c)}\frac{\zeta(1-z,a)\zeta(2kN+2k-1+z,b)}{2k\cos \frac{\pi}{2}\frac{z-1+k}{k}}{\left( \frac{\alpha}{\pi}\right)}^{-z}\,\mathrm{d}z, \quad 1<c<2 \tag{7} \] which has the series expression \[ Z(\alpha,a,b;k,N)=\sum_{n=0}^{\infty} \frac{\Psi_\alpha(n+b,a,k)}{{(n+b)}^{2kN+2k-1}} \tag{8} \] By Cauchy's theorem, shifting the integration path to \(d = -c-2kN-2k+2\) we obtain \[ Z(\alpha,a,b;k,N)=\frac{1}{2\pi i}\int_{(c)}\frac{\zeta(1-z,a)\zeta(2kN+2k-1+z,n)}{2k\cos \frac{\pi}{2}\frac{z-1+k}{k}}{\left( \frac{\alpha}{\pi}\right)}^{-z}\,\mathrm{d}z+P(\alpha), \tag{9} \] where \(P(\alpha)\) is the sum of residues in the strip \(d<\sigma<c\). By the change of variable \(z\to -z - 2kN - 2k + 2\), the integral in (9) becomes \({(-1)}^N{\left(\frac{\alpha}{\pi}\right)}^{2kN+2k-2}\) \(Z(\beta,b,a;k,N)\). This leads to a generalization of Ramanujan's formula, Theorem 2.1. The dual of \(\Psi\) in (5) is defined by changing the denominator by \(2k\sin \frac{\pi}{2}\frac{z}{k}\) and leads to Theorem 2.6. Both of them and other theorems are also established by a new Dirichlet convolution technique [\textit{P. Chavan} et al., Ramanujan J. 59, No. 4, 1245--1285 (2022; Zbl 1498.11173)].