On 16 May 2017 at 16:13, Waldek Hebisch <[email protected]> wrote:
> ...
> One trick is to substitute integers for some parameters.
> After
>
> oV := OrderedVariableList([mp,mq,mr,yqp,yrp,yrq])
> evl := enumerate()$oV
> ideq2a := [map(c +-> eval(c, evl, [3, 5, 7, 9, 11, 13]), p) for p in ideq2];
>
> I was able to compute groebner basis of the evaluated ideal
> in about 5 seconds.

Wow, that is remarkable.  I would not have expected this magnitude of speed up.

> For almost all possible values the evaluated ideal
> has Groebner basis with the same structure as the
> orignal one.

OK, interesting but not so immediately useful.

> It is possible to reconstruct (and verify) Groebner basis
> for the full ideal from Groebner basis for evaluated ideal,
> but ATM FriCAS contains no such code.
>

Surely such reconstruction is likely to require a lot of computation,
no?  Do you have a reference to any publications on this subject?  Is
there already an implementation in some other system?

> groebnerFactorize for evaluated ideal takes 224.17 sec.
> I was also able to do groebnerFactorize after evaluating
> IIRC only 3 variables.  I am not sure if it possible to
> reconstruct full result of groebnerFactorize from evaluated
> result, but at least you should be able to get some
> quantitative data like number of subsystems (1086) and
> degrees.
>

There seems to be quite a lot of literature on algorithms for
triangular decomposition.

Reconstruction sounds interesting but I am skeptical if it can be made
efficient.

> BTW: After evaluating variables one could retract coefficients
> to rational numbers and then reduce modulo a prime.  For
> Groebner basis this should give additianal speedup.  For
> groebnerFactorize reducing modulo prime is dangerous,
> because then a lot of univariate polynomials will be reducible,
> so it may change structure of basis quite a lot.
>

Thank you for these ideas.

Bill.

-- 
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