CHANGELOG

# ChangeLog

## [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 otional, even if Normaliz has been built with Scip
-- fast Gorenstein test
-- restriction of number of significant coefieicients 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
-- imrovement of polynomial arithmetic
-- controlled interruption

## [3.2.1] 2017-02-22

-- automatic choice of symmetzrization
-- change to HSOP form 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 (estricted 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