Re: [Axiom-developer] Re: hyperdoc

[root]

> That's one of the things I'll be interested to see how CATS deals with
> - there (as I understand it, please correct me if I'm wrong Tim) the
> goal is not to be able to swap inputs and outputs, but to have as
> nearly as possible the same question posed, and see if the systems
> agree mathematically?

In fact, the idea behind CATS is that we take a well-defined math
area, choose some problems with a known solutions and write up the 
problems and their solutions. Then, for each CAS, there is a section
that sets up the problem and an encoding of the expected answer so
the testing can be automated. There is no assumption that every
CAS is capable of handling every test case, nor any assumption 
about common structure/function/etc. 

CATS is trying to suggest a way to pool the mathematical expertise
needed to solve a problem and present it so that each CAS can 
consider the problem and know they have a correct solution. Of 
course, the issue of a "correct" solution is not always clear and
often a point of debate but that's why the mathematical explanation
section exists.

See Jeffrey, D.J and Norman, A.C
"Not seeing the roots for the branches: multivalued functions in
computer algebra" SIGSAM Bulletin (Association for Computing 
Machinery) Vol 38, Number 3; Sept 2004 Issue 149 pp57-66

for the kind of analysis that might be expected.

Arthur Norman probably has a copy of the paper on the web somewhere.

t