[Axiom-developer] Proving Axiom Correct

[daly]

Interactive Theorem Proving in Industry (Intel speaker)
http://events.inf.ed.ac.uk/Milner2012/J_Harrison-html5-mp4.html

This talk was part of the Milner symposium.

This talk mentions the idea of proving computer algebra
results. He notes that some classes of problems that can
be checked easily (aka have a certificate) exist.

For instance, factorization problems can easily be checked
by expanding the factorization back to the input.

Indefinite integration can be checked by differentiation.

But definite integration is harder and there is no obvious
way to produce a certificate.

He conjectures that there may be an NP-like class for
computer algebra where you can't check in polynomial time.

Altogether an interesting talk.

Tim