Opened 5 years ago
Closed 5 years ago
#23805 closed enhancement (fixed)
adding SchemeMorphism_points to coercion framework
Reported by:  bhutz  Owned by:  

Priority:  minor  Milestone:  sage8.1 
Component:  algebraic geometry  Keywords:  
Cc:  Merged in:  
Authors:  Ben Hutz  Reviewers:  Paul Fili, Adam Towsley 
Report Upstream:  N/A  Work issues:  
Branch:  74347ea (Commits, GitHub, GitLab)  Commit:  74347ea0b5fd48c7e52b1e8c8eb0cb11bc9c4d52 
Dependencies:  Stopgaps: 
Description (last modified by )
This is part of meta ticket #23047. The goal is to have scheme points able to be coerced when that coercion makes sense. Essentially SchemeHomeset_points needs a _coerce_map_from_ function and possibly updates to _element_constructor_.
For example, we should be able to coerce between the following objects
 the same space type where the base rings have a coercion between them. i.e. P^{1}(ZZ) > P^{1}(QQ)
 a points of a subscheme to points of the ambient space
 base ring elements to Affine 1 space.
Change History (13)
comment:1 Changed 5 years ago by
 Branch set to u/bhutz/scheme_point_coercion
 Commit set to c020db966dd35e0d7effaf08d64b403a9490548b
 Status changed from new to needs_review
comment:2 Changed 5 years ago by
 Description modified (diff)
comment:3 Changed 5 years ago by
 Status changed from needs_review to needs_work
 There should be no coercion from list or tuple, just a conversion.
 No coercion if the variable names on the projective/affine space are different (for consistency with multivariate polynomials).
 For products of projective spaces and toric varieties, you need to check more than just that the relative dimensions are the same.
comment:4 Changed 5 years ago by
 Commit changed from c020db966dd35e0d7effaf08d64b403a9490548b to 6d26bccd251542d7d2bd7c41d01faf8220c0eef4
Branch pushed to git repo; I updated commit sha1. New commits:
6d26bcc  23805: fixes from first review

comment:5 Changed 5 years ago by
Thanks for the comments. I've tried to better match what polynomial rings are doing. I don't think even a conversion from tuple/lists actually fits here very well. So I've no longer allowed that behind the scenes. i.e. being able to say P1([0,1]) == [0,1] does not work.
However, I feel out of my depth for Toric varieties. I'm not sure what type of situations it would actually be valid to coerce. Consequently, I taken to conservative approach and not done coercion for those. Would it be better to raise a NotImplementedError? than just allowing _coerce_map_from_ to return None? or does this need to wait for someone who can deal with the Toric cases?
comment:6 Changed 5 years ago by
I agree that P1([0,1]) == [0,1]
should not work, but again, make sure you keep coercion and conversion straight. The fact that P1([0,1])
works is a conversion, and is fine. the implicit construction described by the equality test you mention is a coercion, and is not.
As for toric varieties, it's fine leaving it as None
. That just means that there is no coercion now; one can be added later if appropriate.
comment:7 Changed 5 years ago by
 Status changed from needs_work to needs_review
comment:8 Changed 5 years ago by
 Status changed from needs_review to needs_work
I ran some functionality tests, the following examples should give true, but don't:
PQ.<x,y> = ProjectiveSpace(QQ,1) PC.<x,y> = ProjectiveSpace(CC,1) PQ([1,0])==PC([1,0]) , PC(CC).has_coerce_map_from(PQ(QQ))
and
R.<x> = PolynomialRing(QQ) K.<i> = NumberField(x^2+1) A1.<x> = AffineSpace(QQ,1) A2.<x> = AffineSpace(K,1) p1 = A1(1) p2 = A2(1) p1==p2 , A2(K).has_coerce_map_from(A1(QQ))
comment:9 Changed 5 years ago by
 Status changed from needs_work to needs_review
Nevermind, those examples do work. My apologies.
comment:10 Changed 5 years ago by
 Reviewers set to Paul Fili
 Status changed from needs_review to positive_review
I reviewed the code and I tried some rather sophisticated examples (couple of examples below), wasn't able to find any problems Looks good to me.
sage: R.<X> = PolynomialRing(QQ) sage: K.<a> = NumberField(X^2  2, embedding=QQbar(sqrt(2))) sage: P.<x,y> = ProjectiveSpace(QQbar, 1) sage: Q.<x,y> = ProjectiveSpace(K, 1) sage: P(sqrt(2),1) == Q(a,1) sage: P(sqrt(2),1) (1.414213562373095? : 1) sage: Q(a, 1) (a : 1) sage: P(sqrt(2),1) == Q(a,1) True sage: P(QQbar).has_coerce_map_from(Q(K)) True
and
sage: R.<t> = PolynomialRing(QQ) sage: K.<a> = NumberField(t^2  6, embedding=QQbar(sqrt(6))) sage: P.<x,y,z,w> = ProjectiveSpace(QQbar, 3) sage: X = P.subscheme(y^2*z  x*(xz)*(xw) ); X Closed subscheme of Projective Space of dimension 3 over Algebraic Field defined by: x^3 + x^2*z + y^2*z + x^2*w  x*z*w sage: Y = P.subscheme([y^2*z  x*(xz)*(xw), w  2*z]); Y Closed subscheme of Projective Space of dimension 3 over Algebraic Field defined by: x^3 + x^2*z + y^2*z + x^2*w  x*z*w, (2)*z + w sage: Y.change_ring(K) Closed subscheme of Projective Space of dimension 3 over Number Field in a with defining polynomial t^2  6 defined by: x^3 + x^2*z + y^2*z + x^2*w  x*z*w, 2*z + w sage: P = Y([3,a,1,2]); P (3/2 : 1.224744871391589? : 1/2 : 1) sage: X(P) verbose 0 (3370: multi_polynomial_ideal.py, groebner_basis) Warning: falling back to very slow toy implementation. verbose 0 (3370: multi_polynomial_ideal.py, groebner_basis) Warning: falling back to very slow toy implementation. verbose 0 (3370: multi_polynomial_ideal.py, groebner_basis) Warning: falling back to very slow toy implementation. (3/2 : 1.224744871391589? : 1/2 : 1) sage: Q = X([3,a,1,2]) sage: P == Q verbose 0 (3370: multi_polynomial_ideal.py, groebner_basis) Warning: falling back to very slow toy implementation. True sage: Q == X(P) # should be true True sage: Q2 = X([3,a,1,2]) sage: Q2 == P # should not be true False
comment:11 Changed 5 years ago by
 Commit changed from 6d26bccd251542d7d2bd7c41d01faf8220c0eef4 to 74347ea0b5fd48c7e52b1e8c8eb0cb11bc9c4d52
 Status changed from positive_review to needs_review
Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:
74347ea  23805: add another subscheme example

comment:12 Changed 5 years ago by
 Reviewers changed from Paul Fili to Paul Fili, Adam Towsley
 Status changed from needs_review to positive_review
It worked appropriately for every example that I came up with.
comment:13 Changed 5 years ago by
 Branch changed from u/bhutz/scheme_point_coercion to 74347ea0b5fd48c7e52b1e8c8eb0cb11bc9c4d52
 Resolution set to fixed
 Status changed from positive_review to closed
Here is a first implementation of coercion for scheme points. Comments welcome.
New commits:
23805: coercion for scheme points