CHANGELOG

# ChangeLog

## [3.9.3]  2022-04-21

-- MS Windows binary with all optional libraries
-- option NoHilbertBasisOutput added
-- short reference added
-- Bug fixes
-- Normaliz.lib for Singular updated

## [3.9.2]  2022-02-07

-- compilation for MS Windows under MSYS.
-- bug fixes and improvements
-- extension of sparse vectors to ranges of indices
-- unit_matrix as an input type.
-- output file with precomputed data as input for future computations
-- cone constructor from Normaliz input file
-- Slight modification of reduction.cpp to prevent errors in clang compilations

## [3.9.1] 2021-09-01

-- better handling of distributed computation
-- Python 2 builds dropped
-- bugfix and improvement of hollow triangulation
-- compatibility for operator < in dynamic_bitset with boost::dynamic_bitset

## [3.9.0] 2021-07-02

-- volume and integral by signed decomposition
-- variant ExploitIsosMult added to volume by descent
-- InputAutomorphisms and AmbientAutomorphisms
-- SHA256 hash values for certain data
-- Placing and pulling triangulations

## [3.8.10] 2021-02-05

-- change from Travis to GitHub actions
-- improvement of configure.ac for finding nauty
-- data structure changed for triangulations and decompositions
-- PyNormaliz added to manual

## [3.8.9] 2020-09-25

-- extreme rays in floating point available
-- TriangulationGenerators replace Generators

## [3.8.8] 2020-08-25

-- dual versions of face lattice, f-vector and incidence
-- rational lattices

## [3.8.7] 2020-07-25

-- improvements in finding dependencies
-- addition of IsEmptySemiopen and CoveringFace
-- source file structure changed
-- input in libnormaliz

## [3.8.6] 2020-06-17

-- tests for installation in SageMath


## [3.8.5] 2020-06-20

-- several small improvements
-- significant improvement in computation of integer hulls
-- refined triangulations added

## [3.8.4] 2020-02-20

-- completely revised precomputed_data
-- numerous fixes of small inaccuracies
-- especially for "mixed" input (generators and
   constraints simultaneously)
-- many additional tests and improvement of
   build infrastructure
-- new package layout for the releases
-- update of nauty and CoCoALib releases

## [3.8.3] 2019-11-26

-- repair of modifyCone

## [3.8.2] 2019-11-23

-- removed dependency on Boost by own bitset class
-- improvements in convex hull algorithm
-- internal representation of scalable renf_class_elem
   vector with integral coefficients
-- improvements in build infrastructure

## [3.8.1] 2019-09-30

-- bugfix codimension in face lattice output
-- update of Flint installation to avoid PIE clash
-- improve nauty installation

## [3.8.0] 2019-09-10

-- bugfix triangulation and cone decomposition
-- computation of automorphism groups
-- Computation goal Incidence added
-- SCIP removed

## [3.7.4] 2019-07-31

-- bugfix in face lattice computation
-- nmz_config.h indicates optional packages

## [3.7.3] 2019-06-09

-- "Dybamic" added
-- modifyCone added
-- VPATH builds in install scripts

## [3.7.2] 2019-04-25

-- update to PyNormaliz 2.2

## [3.7.1] 2019-04-23

-- output now a class of libnormaliz
-- introduction of nmz_config.h

## [3.7.0] 2019-04-15

-- QNormaliz realized within Normaliz
-- alignment of manual and code
-- NumberLatticePoints added
-- FaceLattice and FVector added
-- use of refurbished e-antic based on Flint 2.5.2

## [3.6.3] 2018-08-21

-- configure.ac back to "old" treatment of Flint

## [3.6.2] 2018-07-31

-- Better handling of degenerate polytopes

## [3.6.1] 2018-07-07

-- Improvements in build scripts
-- Introduction of NoGradingDenom and better treatment of polytopes
-- Refinements in the computation of volumes

## [3.6.0] 2018-05-31

-- QNormaliz with e-antic included in release
-- Computation goal EhrhartQuasipolynomial introduced, use of KeepOrder extended
-- Hilbert basis of recession cone can be set as precomputed data
-- improved installation and build system

## [3.5.4] 2018-04-13

-- triangulation, cone and Stanley decomposition available for cones over polytopes
   defined by inhomogeneous data
-- EhrhartSeries introduced as a computation goal
-- QNormaliz extended to algebraic polytopes (outside of release)

## [3.5.3] 2018-04-03

-- repair of performance loss
-- bug fixes

## [3.5.2] 2018-03-11

-- Descent algorithm for volumes of polytopes
-- recession rank also computed from Hilbert basis
-- install scripts improved for Mac OS
-- bug fixes

## [3.5.1] 2018-01-15

-- Uniqueness in output of not uniquely determined results (bases of sublattices, equations),
   necessary to make all tests yield the same results on all systems

## [3.5.0] 2017-12-11

-- LLL reduced coordinates for coordinate dependent algorithms
-- Expansion of Hilbert series and weighted Ehrhart series
-- projection of cones
-- Euclidean volumes
-- Flint as an optional library
-- Dockerfile
-- install scripts

## [3.4.1] 2017-11-15

-- bugfix

## [3.4.0] 2017-08-20

-- limit of parallelization via libnormaliz (and default limit of 8 threads)
-- floating point numbers in input
-- project-and-lift algorithm for lattice points in polytopes (default choice), also with a floating point variant
-- subdivision of large simplices by project-and-lift (default choice)
-- option to suppress output of extreme rays (can look rather ugly)
-- use of Scip optional, even if Normaliz has been built with Scip
-- fast Gorenstein test
-- restriction of number of significant coefficients of quasipolynomial
-- definition of semi-open parallelepipeds in input and output of their lattice points

## [3.3.0] 2017-05-15

-- inclusion of NmzIntegrate in libnormaliz
-- rational numbers in input
-- improvement of polynomial arithmetic
-- controlled interruption

## [3.2.1] 2017-02-22

-- automatic choice of symmetrization
-- change to HSOP from of Hilbert series possible after computation
-- fits PyNormaliz 1.4

## [3.2.0] 2017-01-01

-- constraints in symbolic format
-- a better implementation of Approximate and its use in the inhomogeneous case,
-- option Symmetzrize that produces symmetrized input for and runs nmzIntegrate and runs on it,
-- QNormaliz, a version of Normaliz using coordinates in an extension of the rational numbers
   (restricted to convex hull computations and triangulation),
-- further automatic choices of algorithmic variants.

## [3.1.4] 2016-11-26
- automatic choice of dual or primal algorithm (unless fixed by the user)
--extension of ConeProperty, constructors based on Matrix<Integer>, additional functions for retrieval of results

## [3.1.3] 2016-09-28
- bug fixes

## [3.1.2] 2016-08-25
- additional autotools build system
- sparse vectors and matrices in input files added
- choice of output directory added
- option HSOP added

## [3.1.1] 2016-04-05
- index of unit group extension added
- formatted vectors matrices in input files added
- "constraints" in input files added
- transposed matrices in input file added

## [3.1.0] 2016-02-04

- support for nonpointed cones / input of subspace
- new computation goals: IsIntegrallyClosed and WitnessNotIntegrallyClosed
- new computation goal: IntegerHull
- new computation goal: ConeDecomposition

## [3.0.0] - 2015-09-18
- new, more natural comfortable input syntax (with backward compatibility)
- standardization of output improved, especially for Hilbert series (not backward compatible)
- new computation goals: module generators over original monoid and class group
- Additional input types, in particular generators for lattices; free combination of generators and constraints
- computation goals can be set in the input file
- long options available
- automatic choice of integer type
- improved linear algebra with much better protection against overflows
- reduction of the arithmetical complexity by subdivision of large simplicial cones and bottom decomposition based on SCIP and approximation methods
- improvement of Fourier-Motzkin elimination by ordering the generators

## [2.12.2] - 2014-01-22
- bug fix in volume computation "-v"
- change max_rank_submatrix back to pre2.12 version, avoids some overflows

## [2.12.1] - 2014-10-23
- bug fix in dual algorithm
- workaround for compiler bug in intel windows compiler for -v

## [2.12.0] - 2014-10-17
- dual algorithm thoroughly revised
- internal parallelization of simplicial cones with large determinants
- improvement of linear algebra

## [2.11.2] - 2014
- improvement of intermediate reduction
- bug fix

## [2.11.1] - 2014
- bug fix

## [2.11.0] - 2014-04-30
- addition of inhomogeneous input possibilities
- Hilbert series of semiopen cones.
- integral approximation of rational polytopes.
- lattice points in polytopes via the dual algorithm.
- improvement in Fourier-Motzkin elimination by better use of pyramid decomposition.
- substantial improvement in computing ``large'' simplicial cones.

## [2.10.1] - 2013-06-27
- bug fix (wrong multiplicity in computation mode -v (volume) for some non-integral polytopes)

## [2.10.0] - 2013-05-13
- now avoids the production of duplicates of candidates for the Hilbert basis
- improvement of NmzIntegrate
- corrections in the output forwarded to NmzIntegrate