I have recently commited a few other changes related to
factorizatiom. Good news is that now Expression,
Algebraicnumber and SAE (for fields) export
PolynomialFactorizationExplicit, so most domain automatically
gets routines for polynomial factorization. This means
that few old factorization packages became redundant and
could be removed. Also, in several places in algebra which
need factorization various ad hock tricks were uses instead
of calling general factorizer. Some of those tricks were
incomplete, like ignoring algebraic nature of roots when
factoring expression or using squareFree when we needed
real factorizer. I removed most of those tricks (some
remain as it will require work to properly modify
affected code). In particular, factorizer have now no
fallback to squareFree: when factor is called and there
is no factorizer, then we get library error.
Not so good is that now running the test suite takes more
time. Some of this is because we are doing better job:
factorization is more expensive that squareFree (or
doing nothing), factoring in algebraic extensions is
significantly more expensive than factoring which ignores
algberics. But some of this is because general factorizers
are slower than more specialized ones. So we need to look
at performance regressions and fix them. Also, we are using
different code, which in the past did not get much use,
so some new bugs are likely to show up.
Currently algorithmically we are using rather old methods
for factorization. I hope add a bunch of algorithmic
improvements in the future.
--
Waldek Hebisch
[email protected]
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/fricas-devel.
For more options, visit https://groups.google.com/d/optout.
