Be notified of new releases
Create your free GitHub account today to subscribe to this repository for new releases and build software alongside 50 million developers.Sign up
Changes relative to 3.8.0:
- Dependency on Boost replaced by own dynamic_bitset class (written by Max Horn).
- Use of precomputed data newly designed.
- Several little bugs fixed, especially in connection with "exotic" input. Several small improvements in the code.
- Build infrastructure cleaned up and simplified. Much closer to GNU standard now.
- New versions of CoCoALib and nauty integrated.
- Package structure of the release simplified: one package for source distribution, one for binary distribution (depending on platform).
The source distribution of Normaliz contains PyNormaliz 2.10, which is independently available at
The binary distribution contains jNormaliz.
If you do not want to build Normaliz yourself, download the binary package made for your platform and extract it in a directory of your choice. Otherwise (or in addition) download the source and extract it (in the same directory).
BUGFIX: As Jonathan Kliem noticed, there is a bug in the computation of affine dimension and recession rank for inhomogeneous precomputed data if the polyhedron is not full dimensional or has no vertices. The problem is fixed with cone.cpp below. Please replace source/libnormaliz/cone.cpp by this version.
This is a purely technical realease that fixes an incompatibility between libnormaliz and PyNormaliz in version 3.8.2.
For binaries see 3.8.2.
IMPORTANT: the first versions of the source tar.gz and zip did not contain download.sh. They have now been updated.
In any case you can copy it into the Normaliz directory and make it executable by
chmod a+x download.sh.
-- Some improvement in convex hull computation for algebraic polytopes.
-- Code cleaned up and formatted; also in the installation files.
-- own dynamic_bitset class (by Max Horn), removing the dependency on Boost
No change in the user interface.
IMPORTANT: the file download.sh is missing from the tar and the zip packages below. Copy it from the release of 3.3.3 into the Normaliz directory and make it executable by
chmod a+x download.sh.
This release is a minor update to 3.8.0. It fixes the codimension bug of 3.8.0 and improves some install scripts.
This release adds the computation of automorphism groups to Normaliz. Various types of automorphism groups can be computed: combinatorial, rational/algebraic, integral and euclidean. It is not yet possible to use the automorphism groups in the computation of other invariants, which is the main reason for their introduction.
The option to use Scip has been removed.
A binary for Mac OS will follow as soon as possible.
CAUTION: the codimension in face lattice output is too latre by 1. Correction: replace line cone.cpp:7340 by
This release fixes the bug reported in issue #238.
Ther status of nauty and automorphisms has not changed yet (see 3.7.3)
This release adds the computation goals Dynamic and Static and the libnormaliz function modifyCone. These are mainly interseting for the interactive use of libnormaliz. PyNormaliz has been extended accordingly.
IMPORTANT NOTE: This release contains already part of the computation and exploitation of automorphism groups. This computation requires the optional package nauty, http://users.cecs.anu.edu.au/~bdm/nauty. Automorphism groups are not an official part of this release. With improvements and extensions they will be released in version 3.8.0. With the exception of Appendix D they do not yet appear in the manual.
Binaries as in 3.7.1 (MacOS and Linux) and 3.7.0 (MS Windoes)
This release fixes several small problems in Normaliz and PyNormaliz related to algebraic polyhedra.
For MS Windows binary see 3.7.0.
Algebraic polyhedra: these can now be computed directly in Normaliz so that QNormaliz has become obsolete. PyNormaliz has been extended accordingly. The computation of algebraic polyhedra is based on the package e-antic, version 0.1.2 by V. Delecroix (http://www.labri.fr/perso/vdelecro/e-antic/).
Face lattice and f-vector can now be computed by Normaliz.
NumberLatticePoints is a new computation goal that lets Normaliz count lattice points in polytopes without storing them.
NOTE: If libnormaliz was built with e-antic and, therefore, -DENFNORMALIZ had been used, you must also include
-DENFNORMALIZ in any compile command that involves libnormaliz since some header files depend on it.