What's new in JaCoP 4.1.0
May 20, 2014
- In this release we introduce floating-point variables (FloatVar) and constraints on these variables. The methods used in floating point domain of JaCoP are based on floating point intervals and consistency methods build around them. We provide a set of basic arithmetic constraints as well as square root, absolute value, trigonometric constraints (sin, cos,tan, asin, acos, atan), exponential constraint and natural logarithm. Moreover element constraint and equation between integer and float variables makes it possible to build models that mix different JaCoP variables. We have also included experimental implementation of several features that is not consider as final and can change in future.
- The floating point domain is fully integrated with JaCoP solver. The same search methods can be used as well as floating point variables can be used as cost functions for minimization. We provide also specialized optimization methods that are specially developed for floating point variables.
- The released version provides also flatzinc interpreter for floating point variables. Minizinc models containing float variables can be compiled using standard mzn2fzn compiler and used by JaCoP solver.
- This release fixes also few bugs.