Laurent expansion of harmonic zeta functions (Q2195213)

Ramanujan had written in Chapter VI of his Notebook 2 about `the constant of a series'. For a convergent series, the constant is the sum of the series but Ramanujan was attempting to make sense of the ``constant of a divergent series'' which he says is roughly like the ``center of gravity'' of the series. If \(f\) is a smooth function on the positive reals, he introduces the symbol \[\phi(x) = f(1) + f(2) + \cdots + f(x)\] which is supposed to the solution of the difference equation \[\phi(x) - \phi(x-1) = f(x), \phi(0)=0.\] He writes the Maclaurin series \[\phi(x) = C + \int f(x) \,dx + \frac{1}{2} f(x) + \sum_{n \geq 1}\frac{(-1)^n B_{2n}}{(2n)!} f^{(2n-1)}(x)\] where \(B_{2n}\)'s are the Bernoulli numbers and where he says that the constant \(C\) is like the center of gravity of the series. This somewhat vague method -- known as the Ramanujan summation method -- has been made precise later. Many authors including the first author here have developed this method and applied it with success. In the paper under review, the authors consider an analytic function \(f\) on the right half-plane \(\{z : \Re(z) > 0 \}\). The corresponding zeta function is assumed to be defined as a Dirichlet series by \(\zeta_f(s) = \sum_n \frac{f(n)}{n^s}\) that is convergent in a right half-plane \(\Re(s)> \alpha\) and has a meromorphic continuation with a pole of some order \(m\) at \(s=a\) for some \(a\). That is, in a neighbourhood of \(a\), \[\zeta_f(s) = \sum_{n=1}^m \frac{b_n}{(s-a)^n} + C_a + O(s-a).\] In this paper, the authors study how the constant \(C_a\) is related to the series \(\sum_n \frac{f(n)}{n^a}\) in the sense of Ramanujan summation. In the case \(f \equiv 1\), we have the Riemann zeta function, and the constant \(C_1\) is Euler's constant \(\gamma\). The authors study the case \(f(n) = H_n\), the harmonic numbers. In this case, the Dirichlet series \(\sum_n \frac{H_n}{n^s}\) has a meromorphic continuation with a double at \(s=1\) and simple poles at \(s-0,-1,-3,-5, \cdots\). They express the constants \(C_a\) for each pole \(a\), in terms of \(\sum_n \frac{H_n}{n^a}\) in the sense of Ramanujan summation. For instance, they determine \[C_0 = \frac{\gamma + 1}{2}, C_1 =\frac{\gamma^2 + \zeta(2)}{2}.\] They also deal with generalized harmonic numbers in place of \(H_n\)'s.