2.4:
- fix bug in relate_symanzik (thanks to Stephen Jones)

2.3:
- new: scan for super-convergent integrals
- improve handling of incomplete crossings
- improve compatibility with newer compilers
- fix bug for interference of e.g. s- and t-channel diagrams with fermions

2.2:
- new: export of graphs to Loopedia
- support GiNaC 1.7 API
- add example for method of finite integrals
- improve graph based shift finders
- improve support for Mac
- fix compilation issues with g++ 6.2.0 (thanks to Falko Dulat)

2.1:
- new: finite integral finder (pole resolution described in arXiv:1411.7392)
- new: support for dimensionally shifted integrals
- new: support for cut propagators
- new: configurable ordering for integrals
- new: family finder for maximization of permutation symmetry group
- new: Symanzik polynomial based integral identification
- new: configurable list of crossings
- new: export of configuration data, graphs and sector relations to mma
- new: find common parents for list of diagrams
- significantly improve support for (bi-)linear propagators
- significantly improve handling of inter-sector relations
- improve computation of differential equations
- finally works on Mac (with clang 6.0)
- fix critical bug: disable Laurent polynomials for Fermat since leads to wrong results with Fermat 4.25
- fix critical bug: crossings which change scale set to 1 gave false shift identities sometimes
- fix wrong sign of fermion propagator (required update of reduze.sty for Qgraf)
- fix bugs leading to suboptimal results with (graph based) shift finding
- change polarization sum for ghosts to automatically account for right overall sign
- and various smaller additions, improvements, and bug fixes
- thanks to Robert M. Schabinger, Stephen Jones, Matthias Kerner and many
  other people for useful feedback and help with bug fixes

2.0.9:
- fix treatment of incompletely written files on aborts
- fix treatment of invalid file contents
- fix gcc 4.7 compilation issue (Gentoo's Bugzilla Bug 432356)
- fix unusable deselect_recursively option for sector_selection

2.0.8:
- allow for dimension to be substituted via map_on_coeff (generate_identities)

2.0.7:
- fix generation of wrong crossing rules for numerical kinematic invariants
  (previously, in case where scalar_products are set to numerical values other
  than 0, the default setting generate_crossing=true could have caused wrong
  reduction results; now the program aborts)
- fix some sector symmetry shifts not beeing stored (setup_sector_mappings)
- fix bug preventing proper mapping of crossed masters (select_reductions)
- fix failure to match QGRAF diagrams with singular bridge (find_diagram_shifts)
- fix polarization sum issue for 2-pnt functions (compute_diagram_interferences)
- fix paranoid abort for internal-only masses (compute_differential_equations)
- fix and clean up integral and coefficient index types (MPI reductions)
- fix a potential memory leak (MPI reductions)
- abort if data size in MPI communication exceeds limit given by int type (2GB)
- use Fermat if possible also for general linear combinations (normalize)
- truncate very long log file names

2.0.6:
- fix critical bug causing errors in rules for crossings
  (may affect crossings which are not their inverse and which involve at least
  two kinematic invariants (possible for kinematics with >= 4 external legs),
  thanks to William Kilgore for reporting the bug)

2.0.5:
- add option to job setup_sector_mappings to find graphless zero sectors
  (e.g. sectors with bilinear propagators)
- add option to job reduce_sectors to keep temporary results of reduction
- fix rejection of permutation symmetries with bilinear propagators
- fix failure to perform polarization sum for outgoing vectors bosons

2.0.4:
- find and exploit shift relations between sectors which differ only by crossing
- allow permutation symmetries also in presence of non-standard propagators
- optimize symmetry shift finder
- fix application of shift relations for crossed sectors
- fix cmake not finding GiNaC via pkg-config
- fix incomplete crossing minimization and zero tests in sector mappings setup
- fix a missing replacement for symbol_to_replace_by_one in shift identities

2.0.3:
- fix recursive selections and zero recognition problems for crossed sectors
- optimize memory usage during reduction

2.0.2:
- fix crossed integral families ignoring original permutation symmetries
- fix segfault in progress report when reducing empty system of equations

2.0.1:
- workaround compile error with gcc 4.1.2
- fix aborts in graph algorithms for 2-point functions

2.0.0:
- major rewrite and extension of the first Reduze version
