Re: Handling branch cuts for hypergeometric functions

[Richard Fateman , Sat, 12 Feb 2005 09:42:39 -0800]:
> No, I re-read what you wrote, and now I understand what you mean by
> typical values.  I think this could be useful. The difficulty
> is that the conditions can become extremely elaborately nested as the
> computation continues, and so there is a need for simplification.
> For example, if the answer is stated as 3 different forms for x>0, x=0,
> and x<0, but  the forms are equal for x>0 and x<0, then the simplifier
> should make two forms, one for x=0 and one for x#0.

Just to clarify a bit more: The way I see it, if you say the "typical"
value of x is 1, then only the case for x>0 will be returned.

> (I do not expect that the CL condition system will help, unfortunately).

(I thought that the ask...-functions might raise a condition that
causes the question to be asked by default, in which case the
with_typical_values() form might provide some other restart that
provides the answer.  But then my only acquaintance with the condition
system is from reading some of Peter Seibel's "Practical Common
Lisp", and that was a couple of months ago.)

Albert.