>>>>> "Richard" == Richard Fateman <fateman@cs.berkeley.edu> writes:
Richard> Not ever having changed the limit program, I will accept
Richard> this, but I find it sad. I would hope that the limit
Richard> programs would use material already on the list and not
Richard> have to be changed themselves. In particular, limit
Richard> should use differentiation, simplification, and (esp. in
Richard> the variant tlimit) taylor series. If there is some
Richard> other knowledge of a function that is not encoded in
Richard> these programs and is specific to the limit activity, it
Richard> would be nice to understand what that might be!
I may not be remembering everything correctly. Of course, part of the
problem was figuring out what routine was actually doing the limit
computations.
Also, I might have had modify the code because the elliptic functions
take 2 args, unlike most of the other functions, and the arg you
usually want the limit of is the second one. In jacobi_sn(m, x), you
usually want the limit as x approaches some value, not m.
I'll try to look through the code again and reverse engineer the
changes I did and see what kind of info I can get.
Ray