Re: [Maxima-discuss] ML for symbolic math

[Raymond T. <toy...@gm...>]

>>>>> "Richard" == Richard Fateman <fa...@gm...> writes:

    Richard> deciding "interesting" facts or locations to plot.
    Richard> Singular  points, asymptotes, zeroes, .. branch cuts
    Richard> In complex plane for funcs of a complex variable..

I always look to an example from Kahan in his Much Ado About Nothing's
Sign paper where he gives two functions in the complex plane.  Your
task is to determine if they are actually equivalent.  He said it's
unlikely you'd discover that they're different because they are
identical except over a small region on the complex plane.  The best
way is to do the analysis of the functions and looking at all the
branch cuts to see where they're different.  Could ML figure this out
without being trained on this example?

    Richard> All would be prospects.  

    Richard> I would certainly be happy to see progress on these
    Richard> topics.

    Richard> Can ML do this?

    Richard> Uh, lets try something concrete . . Zerofinding perhaps?

    Richard> Or factoring?  ( this WAS tried byBG Claybrook, 1973.  Never heard of it? Why do you  think?)

Zerofinding or factoring?  Of course ML is way more powerful today
than it was in 1973 since computers are a zillion times faster with a
zillion times more memory.