The POT library brings the power of optimal transport (OT) to Python, helping researchers compare and transform probability distributions efficiently. OT is a mathematical framework widely used in machine learning, imaging, and statistics, with applications in domain adaptation, generative modeling, and distribution comparison (e.g., Wasserstein distance).
POT provides fast, scalable implementations of OT solvers. Whether you’re aligning datasets, computing barycenters, or improving ML models, POT makes OT accessible with NumPy, SciPy, and PyTorch support.
🔗 Check it out: POT on Synapse
POT is built with software sustainability in mind—it’s actively maintained, with well-defined dependencies, documentation, and test files ensuring long-term usability. If you’re interested in assessing software sustainability, check out the Cancer Complexity Toolkit, which is currently in development to evaluate best practices in research software.
