v1.0.0

What's Changed

General:

• Added cmake option to disable asserts in the library independent from the vlaue of NDEBUG c8b3d5e
• Added initialize method to the interface of each of the genetic operators which did not previously have it.
  These are called at the start of a GA run. 290b396
• Improved multithreaded performance 6c09e28
• Added the AnyObjective algorithm wrapper to simplify the definition of algorithms which work with arbitrary
  numbers of objectives 9946e47

Support for mixed encodings:

• The generateCandidate function must now be defined as part of the specialization of GaTraits<T> instead of
  the derived GA class when defining new encodings 389bf2c
• Parts of the Algorithm, Selection, and Replacement interfaces now work with vectors of candidate indices
  instead of pointers to candidate solutions 905e2ff
• The context parameter in every genetic operator interface was changed to GaInfo uniformly instead of using the
  typed GA<T> classes in some places 3705d3c
• The attributes field was removed from the candidate solutions f57bcfe
• Added support for mixed-encodings e17351d

Reproducibility related improvements:

• Fixed some bugs related to the reproducibility of the results of GAs when using DLLs on Windows 38623d7, 23b0d21
• Fixed some general bugs regarding the reproducibility of the results of GA runs 3d8bfd7, 75dad5c, 0bcd740
• Improved the testing used to check for the reproducibility of the results of the GAs 77f7ae8
• Guarantee that the results of GA runs are reproducible across platforms a1c5836, ae3c2e2
• Added the cmake option GAPP_DETERMINISM for better control over the handling of compiler flags required for
  reproducible runtime results across debug and release builds 44067f5

Full Changelog: v0.3.0...v1.0.0

v0.3.0

What's Changed

• Implemented support for better constraint handling 38ad061
• Changes to ensure the reproducibility of results 79b44dd, b615ebd
• Added attributes field to the candidates for better customizability 00f627e
• Added the ability to cache the fitness values of candidates in order to reduce the number fitness function calls 1ccd823
• Added the ability to change the number of threads used by the algorithms 024230f

• Added new non-dominated sorting algorithm to improve the performance of the multi-objective algorithms 71267e2
• Improved performance of some binary and permutation crossover operators 327b04d, 32b0353
• Improved performance of the pseudo-random number generators 97af863
• Other various small performance improvements (selection operators and other general improvements)

• Improved documentation and code examples

• Added macOS to the list of tested platforms c3467a2
• Added gcc14, clang16,17,18 to the list of tested compilers d8089ab
• Added arm64 to the list of tested targets d8089ab
• Added x86 (32 bit) to the list of tested targets cdaeeb0
• Fixed some issues on ARM d8089ab

• Added a conanfile for the project db5f69d
• Replaced CMakeSettings.json with CMakePresets.json 4ff4f9e

Full Changelog: v0.2.0...v0.3.0