Raymond Toy writes:
>>>>>>"Seth" == Seth Goldberg writes:
>
>
> Seth> How this arose.
> Seth> I was noodling around and asked for
> Seth> integrate(sin(x^2),x);
> Seth> I got back some interesting form in terms of erf
> Seth> (as an alternative to the usual textbook form).
> Seth> This checked out OK on differentiation.
> Seth> However, it wouldn't plot, and wouldn't evaluate for x=1.
> Seth> Using rectform or ev(...,numer) doesn't seem to help.
>
> Seth> The problem seems already present in erf(%i)
>
> This is a deficiency in maxima. It doesn't know how to calculate erf
> for complex args.
>
> I've looked around for an algorithm to do this, but haven't found one
> yet. There was a patch for octave that used Rybicki's method for
> complex args, but it had some unexplained constants that I wanted to
> understand, and my local university library didn't seem to have the
> journal with Rybicki's article, so I didn't do anything about it.
>
> File this as a bug, if you like.
>
> Ray
>
Hi Raymond!
I just want to remind that quite a bit ago you promised
to take a look at gamma-lanczos for gamma function at
complex argument. gamma-lanczos code is already hare
but AFAIK is used for real gamma's arguments only.
In general I still don't understand what is preferred way
to calculate complex valued expression involving
elementary or special functions numerically?
It seems that rectform is required for expression with
elementary functions. Is is so for special functions.
Which special functions can be evaluated numerically
by Maxima CVS? At real arguments? At complex arguments?
--
Vadim V. Zhytnikov