[Help-gsl] Hurwitz zeta function: code: precision

[Jerome BENOIT]

Hello !

In view to computer derivatives of the Hurwitz zeta function
I am attempting to understand the code implemented in specfunc/zeta.c
for computing $\zeta(s,a)$ for $s>1$ and $a>0$.
I can reproduce the involved mathematics (see [1]).

My concern is the choice for the parameters kmax and jmax.
I guess that once kmax is chosen, jamx can be easily deduced;
so the issue reduces to the choice of kmax.
I have noticed that kmax is not involved in the err computation:
is it a bug or related to the choice kmax=10 (and why) ?

Any hint is welcome.

Thanks in advance,
Jerome



NB: the citation ``Moshier, p 400'' is a little short (and hard to find both 
with Google and in libraries)

[1] 
https://math.stackexchange.com/questions/917100/numerical-evaluation-of-hurwitz-zeta-function/917938?noredirect=1#comment5300709_917938