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.
New optimizers
- 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.
Constrained Optimization
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.