Fw: problem with command normal (from normal.lib)
Burcin Erocal
Burcin
Begin forwarded message:
Date: Fri, 15 Apr 2011 14:47:08 -0300
From: Santiago Laplagne <slap...@dm.uba.ar>
To: philipp...@business.uzh.ch
Cc: Frank Seelisch <seel...@mathematik.uni-kl.de>,
bur...@mathematik.uni-kl.de Subject: Re: problem with command normal
(from normal.lib)
Dear Philipp,
Frank has forwarded my your question about normal.lib.
Thank you very much for your detailed feedback!
We have made a change in the default settings for procedure normal.
In older versions, the first step was to compute the minimal associated
primes. Now, we have changed this to compute only an equidimensional
decomposition.
Computing the equidimensional decomposition is usually faster than
computing the minimal associated primes.
However, in some cases as in your example, computing the normalization
of the equidimensional components can be much slower than computing
the normalization of the prime components.
This produces the differences of timings in your example.
We have made this change because we didnt want the proc to get stuck
in the computation of the min ass primes, which is not part of the
normalization algorithm.
But I will add a comment in the manual, suggesting the user to use the
option "prim" when the associated primes are easy to compute.
Best regards,
Santiago