CBX v.1.0.0

This version provides new features and optimizers.

Torch backend

The handling of torch tensors was improved, and more functions were added to torch_utils.

Polarized CBO: Bungert, L., Roith, T., & Wacker, P. (2024). Polarized consensus-based dynamics for optimization and sampling. Mathematical Programming, 1-31.
Mirror CBO: Bungert, L., Hoffmann, F., Kim, D. Y., & Roith, T. (2025). MirrorCBO: A consensus-based optimization method in the spirit of mirror descent. arXiv preprint arXiv:2501.12189.
Adamized CBO: Chen, J., Jin, S., & Lyu, L. (2020). A consensus-based global optimization method with adaptive momentum estimation. arXiv preprint arXiv:2012.04827.

CBXPy now supports constrained optimization. Next to the modules regularizers and constraints the following methods were added:

Drift constrained CBO: Carrillo, J. A., Jin, S., Zhang, H., & Zhu, Y. (2024). An interacting particle consensus method for constrained global optimization. arXiv preprint arXiv:2405.00891.
Penalized CBO: Borghi, G., Herty, M., & Pareschi, L. (2023). Constrained consensus-based optimization. SIAM Journal on Optimization, 33(1), 211-236.
Hypersurface CBO: Fornasier, M., Huang, H., Pareschi, L., & Sünnen, P. (2020). Consensus-based optimization on hypersurfaces: Well-posedness and mean-field limit. Mathematical Models and Methods in Applied Sciences, 30(14), 2725-2751.

This release incorporates all changes that were made, within the JOSS review.

The names of the plotting routines were changed to be consistent with Python naming conventions.
Unit test, now check for correctness of optimizers.
Minor Bug fixes.

The main updates in this release:

allow multi-dimensional input spaces. Particles are stile of size (M,N,d) but d can be a tuple now.
remodeled termination criteria. Allowing termination per run now.
Noise models are not part of the CBX class anymore and can be specified as a callable.
Exposed norm, normal, copy and compute_consensus function
Added proper implementation of effective step size scheduler

The first release of CBXPy provides the basic functionality and structure of the packages.