This toolbox has been created by
It is currently maintained by
The contributors to this library are:
- Rémi Flamary (EMD wrapper, Pytorch backend, DA classes, conditional gradients, WDA, weak OT, linear OT mapping, documentation)
- Nicolas Courty (Original sinkhorn, Wasserstein barycenters and convolutional barycenters, 1D wasserstein)
- Alexandre Gramfort (CI, documentation)
- Laetitia Chapel (Partial OT, Unbalanced OT non-regularized)
- Michael Perrot (Mapping estimation)
- Léo Gautheron (Initial GPU implementation)
- Nathalie Gayraud (DA classes)
- Stanislas Chambon (DA classes)
- Antoine Rolet (EMD solver debug)
- Erwan Vautier (Gromov-Wasserstein)
- Kilian Fatras (Stochastic solvers, empirical sinkhorn)
- Alain Rakotomamonjy (Greenkhorn)
- Vayer Titouan (Gromov-Wasserstein, Fused-Gromov-Wasserstein)
- Hicham Janati (Unbalanced OT, Debiased barycenters)
- Romain Tavenard (1D Wasserstein)
- Mokhtar Z. Alaya (Screenkhorn)
- Ievgen Redko (Laplacian DA, JCPOT)
- Adrien Corenflos (Sliced Wasserstein Distance)
- Tanguy Kerdoncuff (Sampled Gromov Wasserstein)
- Minhui Huang (Projection Robust Wasserstein Distance)
- Nathan Cassereau (Backends)
- Cédric Vincent-Cuaz (Graph Dictionary Learning, FGW, semi-relaxed FGW, quantized FGW)
- Eloi Tanguy (Generalized Wasserstein Barycenters, GMMOT)
- Camille Le Coz (EMD2 debug)
- Eduardo Fernandes Montesuma (Free support sinkhorn barycenter)
- Theo Gnassounou (OT between Gaussian distributions)
- Clément Bonet (Wassertstein on circle, Spherical Sliced-Wasserstein)
- Ronak Mehta (Efficient Discrete Multi Marginal Optimal Transport Regularization)
- Xizheng Yu (Efficient Discrete Multi Marginal Optimal Transport Regularization)
- Sonia Mazelet (Template based GNN layers)
- Laurène David (Low rank sinkhorn, Low rank Gromov-Wasserstein samples)
- Julie Delon (GMM OT)
This toolbox benefit a lot from open source research and we would like to thank the following persons for providing some code (in various languages):
- Gabriel Peyré (Wasserstein Barycenters in Matlab)
- Mathieu Blondel (original implementation smooth OT)
- Nicolas Bonneel (C++ code for EMD)
- Marco Cuturi (Sinkhorn Knopp in Matlab/Cuda)
POT has benefited from the financing or manpower from the following partners:
