Change in MC++ 5.0 [from MC++ 4.0]
- Updated spectral bound class (specbnd.hpp)
- Updated sparse Chebyshev model class (scmodel.hpp, sicmodel.hpp) with tailored Remez algorithm
- Updated ellipsoidal calculus and image classes (ellipsoid.hpp, ellimage.gpp) with tailored Remez algorithm
- Updated superposition relaxation class (supmodel.hpp) with new capability
- Updated DAG class (ffunc.hpp) and polyhedral relaxation (polimage.hpp) with hash maps
- Added an external operation for DAG embeddings (ffdagext.hpp)
- Moved quadratisation (squad.hpp), variable elimination (selim.hpp) and reduction constraint (sred.hpp) to sister project CANON
- All linear algebra support based on Armadillo library, kept cpplapack as legacy code
- Exposed all classes and types using pybind11
- Added python scripts and Jupyter notebook tutorials

Change in MC++ 4.0 [from MC++ 3.0]
- Changed external operation mechanism to not rely on variadic class templates
- Added external operations for linear terms (fflin.hpp), sparse polynomials (ffpol.hpp), neural networks (ffmlp.hpp, ffmlpreg.hpp)
- Further support of MC++ types in Python using pybind11

Change in MC++ 3.0 [from MC++ 2.3]
- Support of forward and backward automatic differentiation in DAG, no longer reliant on FADBAD++
- Removed depency propagation in DAG, now needing evaluation in mc::FFDep type
- Enabled external operations using variadic class templates
- Support of MC++ types in Python using pybind11

Change in MC++ 2.3 [from MC++ 2.2]
- Move semantic enabled in DAG and several set arithmetics
- Support of external operations in DAG
- Expression lifting and elimination in DAG
- Expression creator from DAG, e.g. for expert to GAMS models
- Improved interval superposition arithmetic

Change in MC++ 2.2 [from MC++ 2.1]
- Support of Boost interval library (in addition to PROFIL/BIAIS and Filib++)
- Processing of sparse expressions in DAG
- Quadratization of multivariate polynomials
- Sparse Chebyshev model arithmetic (both real- or interval-valued coefficients)
- Interval supersition arithmetic

Change in MC++ 2.1 [from MC++ 2.0]
- FADBAD++ library updated to include hyperbolic function (https://github.com/modsim/CADET-semi-analytic/tree/master/ThirdParty/FADBAD++)
- CPPLapack library updated to version 2015.05.11-1 (http://cpplapack.sourceforge.net/doc/html/index.html)
- Functions added in the DAG and computer algebra classes: cosh, sinh, tanh, lmtd, rlmtd, xlog, prod
- Support of n-ary operations in DAG (prod)
- Variadic templates for flexible composition, differentiation and evaluation methods in DAG
- Detection of quadratic, polynomial and rational dependence in DAG, in addition to linear and nonlinear dependence
