Re: Maple Vs Mathematica debugging / cost to write a system

Caerwyn Esgydiad wrote:



But does it have to be that way, or would it be possible to make a CAS
that used more human-like methods?



i have thought of doing this myself, using genetic algorithms alongside a
neural network to literally train a computer to carry out integrations in a
similar manner to how a human would, then be able to report on the
successful and unsuccessful routes.  may do it within the next year or so.


What makes you think that humans use either genetic algorithms or neural
networks?  So far as I can tell, these cute names  ("genetic", "neural")
are merely cute names with vague analogies to the correspondingly named
biological concepts. But surely you don't think humans learn calculus
by altering their DNA, and I doubt that human need such excessive training
as neutral networks.  Doing calculus problems looks like a classic application of
symbolic processing.

Good luck.
RJF



.