lfunthetainit

Function: lfunthetainit
Section: l_functions
C-Name: lfunthetainit
Prototype: GDGD0,L,b
Help: lfunthetainit(L,{tdom},{m=0}): precompute data for evaluating
  the m-th derivative of theta functions with argument in domain tdom
  (by default t is real >= 1).
Doc: Initalization function for evaluating the $m$-th derivative of theta
 functions with argument $t$ in domain \var{tdom}. By default (\var{tdom}
 omitted), $t$ is real, $t \geq 1$. Otherwise, \var{tdom} may be

 \item a positive real scalar $\rho$: $t$ is real, $t \geq \rho$.

 \item a nonreal complex number: compute at this particular $t$; this
 allows to compute $\theta(z)$ for any complex $z$ satisfying $|z|\geq |t|$
 and $|\arg z| \leq |\arg t|$; we must have $|2 \arg z / d| < \pi/2$, where
 $d$ is the degree of the $\Gamma$ factor.

 \item a pair $[\rho,\alpha]$: assume that $|t| \geq \rho$ and $|\arg t| \leq
 \alpha$; we must have $|2\alpha / d| < \pi/2$, where $d$ is the degree of
 the $\Gamma$ factor.

 \bprog
 ? \p500
 ? L = lfuncreate(1); \\ Riemann zeta
 ? t = 1+I/2;
 ? lfuntheta(L, t); \\ direct computation
 time = 30 ms.
 ? T = lfunthetainit(L, 1+I/2);
 time = 30 ms.
 ? lfuntheta(T, t); \\ instantaneous
 @eprog\noindent The $T$ structure would allow to quickly compute $\theta(z)$
 for any $z$ in the cone delimited by $t$ as explained above. On the other hand
 \bprog
 ? lfuntheta(T,I)
  ***   at top-level: lfuntheta(T,I)
  ***                 ^--------------
  *** lfuntheta: domain error in lfunthetaneed: arg t > 0.785398163397448
 @eprog
 The initialization is equivalent to
 \bprog
 ? lfunthetainit(L, [abs(t), arg(t)])
 @eprog