caching of Groebner basis not always done

[dkrenn]

            sage: R.<a,b> = QQ[]
            sage: I = R.ideal([a^2-a, b^2-b, a+b])
            sage: GB1 = I.groebner_basis(algorithm='libsingular:slimgb')
            sage: GB2 = I.groebner_basis()
            sage: GB1 is GB2

returns False, the Gröbner basis is recomputed. This is problematic, as a lot of methods (e.g. .variety) simply call .groebner_basis() and a precomputed Gröbner basis (maybe with a different algorithm than the default) is not used.

Component: algebra

Author: Daniel Krenn

Branch/Commit: u/dkrenn/cache-groebner @ 82d81c8

Issue created by migration from https://trac.sagemath.org/ticket/27328

[dkrenn]

Branch: u/dkrenn/cache-groebner

[dkrenn]

Commit: 8acaaf1

[dkrenn]

New commits:

8acaaf1

Trac #27328: use cache if possible and no algorithm given

[dkrenn]

comment:3

The proposed solution is now that if there is no algorithm given but something is cached, then this is taken.

[nbruin]

comment:4

The building of the cache is still reliant on cached_method. For instance, that means that if I.groebner_basis() is called and then I.groebner_basis(algorithm=<whatever the default is>), then the algorithm gets executed twice.

[dkrenn]

comment:5

Replying to @nbruin:

The building of the cache is still reliant on cached_method. For instance, that means that if I.groebner_basis() is called and then I.groebner_basis(algorithm=<whatever the default is>), then the algorithm gets executed twice.

Yes, this is indeed true. The problem is that I.groebner_basis() does not simply translate to I.groebner_basis(algorithm=<somealgorithmname>) at the beginning, but tries a couple of things which might lead to a result. Any ideas how this can be taken care of?

[tscrim]

comment:6

Replying to @dkrenn:

Replying to @nbruin:

The building of the cache is still reliant on cached_method. For instance, that means that if I.groebner_basis() is called and then I.groebner_basis(algorithm=<whatever the default is>), then the algorithm gets executed twice.

Yes, this is indeed true. The problem is that I.groebner_basis() does not simply translate to I.groebner_basis(algorithm=<somealgorithmname>) at the beginning, but tries a couple of things which might lead to a result. Any ideas how this can be taken care of?

The generic way around that is to instead have the default call self.groebner_basis(algorithm=<algorithmname>) with the chosen algorithm. The other thing you could do, and which might be better, is to explicitly set the cache in those cases.

[embray]

comment:7

Ticket retargeted after milestone closed (if you don't believe this ticket is appropriate for the Sage 8.8 release please retarget manually)

[dkrenn]

comment:8

Replying to @tscrim:

[...] The problem is that I.groebner_basis() does not simply translate to I.groebner_basis(algorithm=<somealgorithmname>) at the beginning, but tries a couple of things which might lead to a result. Any ideas how this can be taken care of?

The generic way around that is to instead have the default call self.groebner_basis(algorithm=<algorithmname>) with the chosen algorithm.

This is not what we want: We do want that algorithm='' tries a couple of things until it succeeds (or fails after all).

The other thing you could do, and which might be better, is to explicitly set the cache in those cases.

You mean setting the cache for algorithm='' to be the same as for algorithm='chosen'. This could work ;)

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from 8acaaf1 to 36ba0ee

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. New commits:

9d4ffa6	Trac #27445: groebner basis for multi-variate polynomial ring with no variables
6afce66	Trac #27445: polynomial rings over non-fields + major restructure of relevant code
df26e69	Merge tag '8.7' into t/27445/gb-no-var
eaaa150	Merge branch 't/27445/gb-no-var' into t/27328/cache-groebner
3234c66	Trac #27328: remove ignored/incompatibel/ununsed *args in groebner basis computations
c0b2168	Trac #27328: simplify parameter processing in groebner_basis
de2dc6c	Trac #27328: fixup prot parameter in submethod
35f231f	Trac #27328: enable "not tested" doctests as they work now
36ba0ee	Trac #27328: do proper caching of groebner_basis

[dkrenn]

comment:10

Replying to @dkrenn:

You mean setting the cache for algorithm='' to be the same as for algorithm='chosen'. This could work ;)

Works indeed; see current implementation.

[dkrenn]

comment:11

Replying to @nbruin:

The building of the cache is still reliant on cached_method. For instance, that means that if I.groebner_basis() is called and then I.groebner_basis(algorithm=<whatever the default is>), then the algorithm gets executed twice.

This is now fixed as well.

[dkrenn]

comment:12

I've changed the code so that all issues are solved.

Moreover, I've simplifed the way arguments are processed and handled, and cleaned up of what seems to be historically grown (and sometimes not even used or incompatible or even ignored).

Please review :)

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Branch pushed to git repo; I updated commit sha1. New commits:

d5e11d2	Trac #27328: fix doctests (sageinspect & cachefunc)
82d81c8	Trac #27328: remove example which does not test what it claims to test anymore

[7ed8c4ca-6d56-4ae9-953a-41e42b4ed313]

Changed commit from 36ba0ee to 82d81c8

[dkrenn]

comment:14

Forgot push; now updated :)

[embray]

comment:15

Moving tickets from the Sage 8.8 milestone that have been actively worked on in the last six months to the next release milestone (optimistically).

[embray]

comment:16

Ticket retargeted after milestone closed

[mkoeppe]

comment:17

Batch modifying tickets that will likely not be ready for 9.1, based on a review of the ticket title, branch/review status, and last modification date.

[kedlaya]

comment:19

It's been long enough that I'm getting a merge failure. Rebase?

[mkoeppe]

comment:20

Setting new milestone based on a cursory review of ticket status, priority, and last modification date.