Published online by Cambridge University Press: 01 October 2016
The Li coefficients $\unicode[STIX]{x1D706}_{F}(n)$ of a zeta or
$L$ -function
$F$ provide an equivalent criterion for the (generalized) Riemann hypothesis. In this paper we define these coefficients, and their generalizations, the
$\unicode[STIX]{x1D70F}$ -Li coefficients, for a subclass of the extended Selberg class which is known to contain functions violating the Riemann hypothesis such as the Davenport–Heilbronn zeta function. The behavior of the
$\unicode[STIX]{x1D70F}$ -Li coefficients varies depending on whether the function in question has any zeros in the half-plane
$\text{Re}(z)>\unicode[STIX]{x1D70F}/2.$ We investigate analytically and numerically the behavior of these coefficients for such functions in both the
$n$ and
$\unicode[STIX]{x1D70F}$ aspects.