Function: intnum
Section: sums
C-Name: intnum0
Prototype: V=GGEDGp
Help: intnum(X=a,b,s,{tab}): numerical integration of s from a to b with 
 respect to X. a (and similarly b) is coded as follows. It can be a scalar: f
 is assumed to be C^infty at a. It can be a two component vector [a[1],a[2]],
 where a[1] is the scalar, and a[2] is the singularity exponent (in ]-1,0]),
 logs being neglected. It can be a one component vector [1] or [-1] meaning
 +infty or -infty, slowly decreasing functions. It can be a two component
 vector [[1], z] or [[-1], z], where [1] or [-1] indicates +infty or -infty
 and z is coded as follows. If z is zero, slowly decreasing. If z is real
 positive, exponentially decreasing, of the type exp(-zX). If z<-1, very
 slowly decreasing like X^(-z). If z is complex nonreal, real part is ignored
 and if z = r+I*s then if s>0, cosine oscillation exactly cos(sX), while if
 s<0, sine oscillation exactly sin(sX). If f is exponentially decreasing times
 oscillating function, you have a choice, but it is in general better to
 choose the oscillating part. Finally tab is either 0 (let the program choose 
 the integration step), a positive integer m (choose integration step
 1/2^m), or a table tab precomputed with intnuminit (depending on the type of
 interval: compact, semi-compact or R, very slow, slow, exponential, or cosine
 or sine-oscillating decrease).

