Hello,

I am going to improve the functionality of nef partitions and fix some
issues in the current implementation
(sage.geometry.lattice_polytope.NEFPartition), which was written
before I understood what are nef partitions and included into Sage
before mandatory refereeing. Worst issues:
- it is allowed to create a nef partition as a sequence of numbers
without specifying the associated polytope, but without the polytope
these numbers are meaningless;
- during pickling the reference to the polytope is dropped even if it
was given, because polytopes are supposed to restore this reference,
but it will not work for a standalone nef partition.

So I want to do the following:
1) deprecate NEFPartition class;
2) replace it with NefPartition derived from SageObject rather than
Sequence;
3) require the polytope (Delta_polar) during construction and perform
a check that the input is indeed a nef partition (unless check=False
was given), do not interfere with pickling;
4) remove "polytope" method, since there are many polytopes related to
a nef partition and such a name is confusing;
5) add methods "Delta", "nabla", "Delta_polar" and "nabla_polar":
while the last two can be obtained as "Delta().polar()", it would be
nice to include a docstring for them explaining their meaning in nef
partitions;
6) "Delta" and "nabla" will take an optional argument so that they can
return "Delta_i" and "nabla_i" polytopes as well;
7) deprecate "LatticePolytopeClass.mif" since this functionality
belongs to a nef partition and in any case it should have a more
meaningful name.

Regarding names of different polytopes, I want to follow p.63 of
"Mirror Symmetry and Algebraic Geometry" by Cox & Katz, i.e. the
starting point is assigning a part to each vertex of Delta_polar,
these parts together with the origin span nabla_i's, whose Minkowski
sum is nabla. On the polar side, Delta is the Minkowski sum of
Delta_i's and their convex hull is nabla_polar.

Any objections or other comments?
Andrey

-- 
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to 
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org

Reply via email to