On 22 May 2017 at 13:01, Waldek Hebisch <[email protected]> wrote:
> Bill Page wrote:
>>
>> Just to confirm in FriCAS the 'totalGroebner' routine which apparently
>> uses degree reverse lexical ordering takes much less time:
>>
>> Time: 1718.35 (EV) + 0.53 (OT) = 1718.88 sec
>>
>> and finds a basis with 28 polynomials (although that is still a
>> lot longer than Singular needs for this computation).
>
> There is a simple trick to make this about 10-20 time faster:
> clear denominator and use coefficients in the ring.  I profiled
> the groebner basis computation and 95% of time goes into
> gcd-s needed for fraction arithmetic.  I also did computation
> over a ring and then fraction of time spent computing gcd-s
> went down to 5%.
>

Thank you for this idea.  By multiplying through by the parameter
domain denominators and changing to a polynomial domain over the ring
instead of the quotient field I can compute the 'totalGroebner' for
this case in just 89 seconds - much better - still not the speed of
Singular but quite useable. As you say it seems that FriCAS does spend
a lot of time computing canonical representation for fractions when
there is no immediate need to do so.

>> But as you say, even with this result I am a bit stuck since there I
>> could not find any convenient way to change the ordering of a given
>> basis.
>>
>> In computing 'groebnerFactorize' is there some way to keep the basis
>> in degree reverse lexical order? Maybe it I use some other polynomial
>> type rather than 'Polynomial'?
>
> IIUC 'groebnerFactorize' uses order from the polynomial type:
>
> ++   The term ordering is determined by the polynomial type used.
>
> OTOH I am not sure how term order affects the result you will get...
>

I am only looking for some 'groebnerFactorize' result for any
convenient term ordering.  Still, by the documentation I am left
wondering just which polynomial domain I should use to get degree
reverse lexical order?

-- 
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 https://groups.google.com/group/fricas-devel.
For more options, visit https://groups.google.com/d/optout.

Reply via email to