Skip to content

Releases

v2.0.0: Python implementation, MIT license, Eq. 15 boundary handling

Choose a tag to compare

@Aekarkinli Aekarkinli released this 21 Nov 09:23

MDE (Multi-population based Differential Evolution): a parameter-free global optimizer.
It accompanies Karkinli, A.E. (2023), Neural Computing and Applications 35, 5193–5206, https://doi.org/10.1007/s00521-022-07969-w

New

  • Official Python implementation (mde_optimizer). Install it with
    pip install git+https://github.com/Aekarkinli/Multi-Population-based-Differential-Evolution-Algorithm.git@v2.0.0.
    It comes with tests, examples and continuous integration (Python 3.10 to 3.13).
  • Verification against MATLAB (verification/). With 5 functions, 25 runs each, the MATLAB and Python results show no statistically significant difference (Mann–Whitney U, p > 0.05 in all cells).
  • Library integrations submitted to mealpy, NiaPy, OptunaHub, EvoloPy and pyMetaheuristic.
  • CITATION.cff for GitHub's "Cite this repository" button.

Changed

  • algo_MDE.m: borderControl now follows Eq. 15 of the paper, so an out-of-bounds variable is re-sampled uniformly in [low, up]. The previous version could return values outside the box when low and up had the same sign. For symmetric boxes it clipped to the bound; both rules give statistically similar results.
  • ReadMe.m was renamed to example_circle_fit.m, and the plotted fitted circle is fixed.
  • The license changed from GPL-3.0 to MIT.

Citation

Karkinli, A.E. (2023). Detection of object boundary from point cloud by using multi-population based differential evolution algorithm. Neural Computing and Applications, 35, 5193–5206. https://doi.org/10.1007/s00521-022-07969-w