Sage/Singular Days Projects
Contents
-
Sage/Singular Days Projects
- Update Singular spkg
- Update the Singular interface to work with libpolys
- Fix the building of spielwiese Singular (libpolys)
- Refactored Singular should pass it's test suite
- Interface to Singular's modules
- Review Simon's letterplace patches
- Plural interface
- Exponent overflow
- Use libpolys for function field arithmetic
- FLINT integration in Singular
- Beginner tickets
- Fast containers for objects and morphisms of a category
- Improve/Review citation library
- Fix reference counting of Singular rings
- Review PolyBoRi 0.8.0 patch
- Review linbox for modn dense matrices patch
Update Singular spkg
People: Simon, Burcin, Martin
working on prerelease of Sage-4.7.2.alpha3 dated Sept 19, 2012 md5sum: ad021855ae07ecdd1cdcc7b973c9cdba
#10903 update Singular
#10810 Singular doesn't build on ARM
#11769 Singular doesn't build on Ubuntu Oneiric
#10902 factorization of multivariate polynomials
Update the Singular interface to work with libpolys
People: Volker, Burcin, Oleksandr
- There were many changes to libSingular as part of the refactoring effort. The Sage interface needs to be adapted to these. This should be a matter of renaming the functions that are called in Sage and adding the current ring as an argument (modulo bugs in libpolys).
- create spkg for refactored Singular
search/replace calls to libSingular in the Sage library to get the interface to compile & link
Fix the building of spielwiese Singular (libpolys)
People: Volker, Oleksandr
parallel building make -j used to fail because of generated sources (e.g. feOpt.inc)
make install should install everything needed for make distcheck
tests run via make check should be able to find resources (e.g. gftables) - cleate symlink or set Singular environment variables?
Refactored Singular should pass it's test suite
People: Burcin, Oleksandr, Hans, Volker
Bug hunt through Singular kernel code. An adventure no hacker can resist.
build refactored Singular: http://github.com/Singular2/singular2.github.com/wiki/New-TODOs
run test suite with debug version of Singular which dumps stack traces on error. (kernel/cntrl.cc should be modified to define DEBUG_STOP instead of INTERACTIVE)
list of test results (temporarily)
- Monday morning: 595 failing tests
- Tuesday morning: 543 failing tests
- Wednesday: 486 failing tests
- Thursday: 390 failing tests
Compiler warnings
People:
There are lots of compiler warnings...
Plural/SCA
People: Oleksandr
- Bug hunt at PLURAL/SCA kernel code: (left) std etc...
Try out some dynamic modules
People: Oleksandr
Try building & using dynamic modules
Blackbox objects
People: Oleksandr
Try building & using blackbox objects, e.g. coefficient domain as a wrapper for coeffs
Interface to Singular's modules
People: Hiep, Burcin, Simon
- This shouldn't be so hard since libSingular uses the same type for polynomials and module elements
Review Simon's letterplace patches
People: Simon, Burcin, Alexander (offsite)
Plural interface
People: Burcin, Simon, Alexander (offsite)
#4539 plural wrapper is rebased to 11068 implementation of one and two sided ideals of noncommutative rings, so that one has left ideals and two-sided ideals of g-algebras.
- The left or two-sided Gröbner bases can be used to compute normal forms (that has been missing before).
- Pickling works (has been missing), the category test suites pass.
Since g-algebras are now unique parents (unless uniqueness is destroyed on purpose), id(self) is used as hash.
- Doc tests pass (says Simon), but the last patch does not apply cleanly (says Alexander, but Simon can't confirm it). Apart from that, it is ready for review.
To do: Currently, quotients of g-algebras are implemented by the general framework of #11068. It should instead wrap what Singular offers, but that shall be on a different ticket.
Exponent overflow
People: Simon
Working on the Plural interface, we found that Sage does not raise an overflow error when it should. Example: With P.<x,y>=QQ[], y^2^30 returns a correct result; in P.<x,y,z>=QQ[], an overflow occurs, but instead of raising an error, y^2^30 returns zero.
- It is fixed, combining the old test used in Sage and the test used in Singular.
Use libpolys for function field arithmetic
People: Burcin
- libpolys has better function field arithmetic now, we should wrap that in Sage.
This is also relevant #9054 create a class for basic function field arithmetic
FLINT integration in Singular
People: Fredrik, MartinL, Christian
- Add missing utility and conversion functions to FLINT
- Use FLINT in Singular
- Benchmark against NTL version
Beginner tickets
People: Hiep, Burcin to practice preparing and submitting patches, for people new to the Sage development process:
http://trac.sagemath.org/sage_trac/ticket/11850 choose the random elements of the ideals.
.hilbert_series(), .hilbert_function() methods of ideals http://trac.sagemath.org/sage_trac/ticket/11855.
http://trac.sagemath.org/sage_trac/ticket/11864 add .saturation(), .blow_up_ideal() methods in the multivariate polynomial ideal class
Fast containers for objects and morphisms of a category
People: Simon
See #10667 Morphisms and Objects of Categories
Improve/Review citation library
People: Niels, Burcin, (and somebody else?)
#11851 #11852 split off pybtex package to a separate ticket (submit as an optional package first)
- Add more examples of citations
add sage's citation
- Open a ticket for matplotlib not to include an extra version of pyparsing.
Fix reference counting of Singular rings
People: Volker, Burcin, Martin
Review PolyBoRi 0.8.0 patch
People: Martin, Alexander (offsite)
Martin reviewed it mathematically. The patch was rebased to 4.7.2alpha3, the spkg got the improved sse2-flags detection from 0.8.1. (Final) tests are running currently.