Analytic continuation of multiple zeta functions (Q4942893)

The series \[ \zeta(s_1,\dots,s_d) := \sum_{0<n_1<\dots<n_d} {1\over n_1^{s_1}\cdots n_d^{s_d}} \] is absolutely convergent in the domain \(\text{Re}(s_d)>1\), \(\sum_{j=1}^d \text{Re}(s_j)>d\) and defines an entire function there, the so-called multiple zeta function (its values at the integers are sometimes called Euler-Zagier sums). For some years it has been believed that for \(d>1\) (and in striking contrast to the \(d=1\) case) these functions do not possess an analytic continuation beyond the area of convergence. In the present paper this belief is proved to be wrong. The author uses generalized functions to prove that \(\zeta(s_1,\dots,s_d)\) can be analytically extended to a meromorphic function on all of \(\mathbb C^n\) with possible (and then simple) poles at \(s_d=1\) and at \(s_j+\dots+s_d=d-j+2-l\) for positive integers \(l\) and \(1\leq j < d\). He also computes the residues at these poles. The reason why the analytic continuation was believed not to exist are certain apparent contradictions which are now explained by the existence of these poles.