*December 2023*
This new release contains several new features and bug fixes. Among the new features
we have a new solver for estimation of nearest Brenier potentials (SSNB) that can be used for OT mapping estimation (on small problems), new Bregman Alternated Projected Gradient solvers for GW and FGW, and new solvers for Bures-Wasserstein barycenters. We also provide a first solver for Low Rank Sinkhorn that will be ussed to provide low rak OT extensions in the next releases. Finally we have a new exact line-search for (F)GW solvers with KL loss that can be used to improve the convergence of the solvers.
We also have a new `LazyTensor` class that can be used to model OT plans and low rank tensors in large scale OT. This class is used to return the plan for the new wrapper for `geomloss` Sinkhorn solver on empirical samples that can lead to x10/x100 speedups on CPU or GPU and have a lazy implementation that allows solving very large problems of a few millions samples.
We also have a new API for solving OT problems from empirical samples with `ot.solve_sample` Finally we have a new API for Gromov-Wasserstein solvers with `ot.solve_gromov` function that centralizes most of the (F)GW methods with unified notation. Some example of how to use the new API below:
python
Generate random data
xs, xt = np.random.randn(100, 2), np.random.randn(50, 2)
Solve OT problem with empirical samples
sol = ot.solve_sample(xs, xt) Exact OT betwen smaples with uniform weights
sol = ot.solve_sample(xs, xt, wa, wb) Exact OT with weights given by user
sol = ot.solve_sample(xs, xt, reg= 1, metric='euclidean') sinkhorn with euclidean metric
sol = ot.solve_sample(xs, xt, reg= 1, method='geomloss') faster sinkhorn solver on CPU/GPU
sol = ot.solve_sample(x,x2, method='factored', rank=10) compute factored OT
sol = ot.solve_sample(x,x2, method='lowrank', rank=10) compute lowrank sinkhorn OT
value_bw = ot.solve_sample(xs, xt, method='gaussian').value Bures-Wasserstein distance
Solve GW problem
Cs, Ct = ot.dist(xs, xs), ot.dist(xt, xt) compute cost matrices
sol = ot.solve_gromov(Cs,Ct) Exact GW between samples with uniform weights
Solve FGW problem
M = ot.dist(xs, xt) compute cost matrix
Exact FGW between samples with uniform weights
sol = ot.solve_gromov(Cs, Ct, M, loss='KL', alpha=0.7) FGW with KL data fitting
recover solutions objects
P = sol.plan OT plan
u, v = sol.potentials dual variables
value = sol.value OT value
for GW and FGW
value_linear = sol.value_linear linear part of the loss
value_quad = sol.value_quad quadratic part of the loss
Users are encouraged to use the new API (it is much simpler) but it might still be subjects to small changes before the release of POT 1.0 .
We also fixed a number of issues, the most pressing being a problem of GPU memory allocation when pytorch is installed that will not happen now thanks to Lazy initialization of the backends. We now also have the possibility to deactivate some backends using environment which prevents POT from importing them and can lead to large import speedup.
New features
+ Added support for [Nearest Brenier Potentials (SSNB)](http://proceedings.mlr.press/v108/paty20a/paty20a.pdf) (PR #526) + minor fix (PR 535)
+ Tweaked `get_backend` to ignore `None` inputs (PR 525)
+ Callbacks for generalized conditional gradient in `ot.da.sinkhorn_l1l2_gl` are now vectorized to improve performance (PR 507)
+ The `linspace` method of the backends now has the `type_as` argument to convert to the same dtype and device. (PR 533)
+ The `convolutional_barycenter2d` and `convolutional_barycenter2d_debiased` functions now work with different devices.. (PR 533)
+ New API for Gromov-Wasserstein solvers with `ot.solve_gromov` function (PR 536)
+ New LP solvers from scipy used by default for LP barycenter (PR 537)
+ Update wheels to Python 3.12 and remove old i686 arch that do not have scipy wheels (PR 543)
+ Upgraded unbalanced OT solvers for more flexibility (PR 539)
+ Add LazyTensor for modeling plans and low rank tensor in large scale OT (PR 544)
+ Add exact line-search for `gromov_wasserstein` and `fused_gromov_wasserstein` with KL loss (PR 556)
+ Add KL loss to all semi-relaxed (Fused) Gromov-Wasserstein solvers (PR 559)
+ Further upgraded unbalanced OT solvers for more flexibility and future use (PR 551)
+ New API function `ot.solve_sample` for solving OT problems from empirical samples (PR 563)
+ Wrapper for `geomloss`` solver on empirical samples (PR 571)
+ Add `stop_criterion` feature to (un)regularized (f)gw barycenter solvers (PR 578)
+ Add `fixed_structure` and `fixed_features` to entropic fgw barycenter solver (PR 578)
+ Add new BAPG solvers with KL projections for GW and FGW (PR 581)
+ Add Bures-Wasserstein barycenter in `ot.gaussian` and example (PR 582, PR 584)
+ Domain adaptation method `SinkhornL1l2Transport` now supports JAX backend (PR 587)
+ Added support for [Low-Rank Sinkhorn Factorization](https://arxiv.org/pdf/2103.04737.pdf) (PR #568)
Closed issues
- Fix line search evaluating cost outside of the interpolation range (Issue 502, PR 504)
- Lazily instantiate backends to avoid unnecessary GPU memory pre-allocations on package import (Issue 516, PR 520)
- Handle documentation and warnings when integers are provided to (f)gw solvers based on cg (Issue 530, PR 559)
- Correct independence of `fgw_barycenters` to `init_C` and `init_X` (Issue 547, PR 566)
- Avoid precision change when computing norm using PyTorch backend (Discussion 570, PR 572)
- Create `ot/bregman/`repository (Issue 567, PR 569)
- Fix matrix feature shape in `entropic_fused_gromov_barycenters`(Issue 574, PR 573)
- Fix (fused) gromov-wasserstein barycenter solvers to support `kl_loss`(PR 576)