|
C.1 News and changes
NEWS in SINGULAR::PLURAL 2-1-99
The current version 2-1-99 is the second pre-release
of the upcoming release series version 2-1. Therefore, some
command names are still subject to change.
Singular::Plural version 2-1 is generally much faster than any previous
experimental version of Singular::Plural, due to a rewrite of major parts
of the Singular kernel (e.g., the kernel has a new memory manager
and a new polynomial arithmetic) and due to new and better implemented algorithms (e.g., for two-sided Groebner bases and generally, for computations
in factor algebras).
Besides theses internal changes, Singular::Plural version 2-1 offers many new
features and functionalities.
Efficiency Improvements
The following table shows some timings which compare the efficincy of
Singular:Plural version 0-9-9 and Singular version 2-1-2.
All times are in seconds
and were obtained on HP 160 workstation with 512 MB RAM running HP-UX 10.20.
Example | 0-9 | 2-1 | Speedup |
AnnFD-sl2-4 | 3.34 | 0.92 | 3.6 |
TwoGB-g2-3 | 24577 | 165 | 149 |
ucha4 | 12.9 | 8.5 | 1.5 |
ucha2 | 44.3 hrs | 9 hrs | 4.9 |
(Commutative) Factorizing algorithms revisited
Starting with version 2-0-4, we use NTL (of Victor Shoup) for factoring
univariate polynomials. The multivariate factoring code in libfac/factory
does now also work over algebraic extension fields.
General Changes
- Emacs user interface
- the recommended interface for using SINGULAR::PLURAL
EPlural
- new program for an out-of-the-box, pre-customized Emacs which runs
Plural
- The online help system
- choose in which browser the on-line help is displayed;
wildcard expansion of help topics
new layout of html manual pages
- Source code debugger
- interactive debugging of procedures written in the SINGULAR
language
- http://www.singular.uni-kl.de/plural
- New WWW home-site of PLURAL
SINGULAR::PLURAL libraries
- lieA_lib Plural definitions of several series of enveloping algebras of simple Lie algebras
Internal Changes
- new data structures for monomials and polynomials
- Makes polynomial arithmetic significantly faster: Especially for block -
or weighted orderings.
Bucket representation of polynomials in std and NF;
Ring change during std-related computations
(resulting in more speed and less space consumption).
- new memory management
- Results is less memory usage, faster memory allocation/free, less
fragmentation, much better locality of reference.
|