Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[WIP] New API solve_sample #562

Closed
wants to merge 4 commits into from
Closed

[WIP] New API solve_sample #562

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a73ad08
add exact line-search for (f)gw solvers with kl_loss (#556)
cedricvincentcuaz Nov 4, 2023
1071759
add kl_loss to all semi-relaxed (f)gw solvers (#559)
cedricvincentcuaz Nov 4, 2023
fe20bc6
[WIP] Add new features to unbalanced solvers (#551)
6Ulm Nov 6, 2023
1ece2d8
fix conflict with RELEASES.md (#561)
cedricvincentcuaz Nov 6, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
163 changes: 17 additions & 146 deletions RELEASES.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -13,6 +13,9 @@
+ 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)

#### Closed issues
- Fix line search evaluating cost outside of the interpolation range (Issue #502, PR #504)
Expand All @@ -34,143 +37,13 @@ Many other bugs and issues have been fixed and we want to thank all the contribu


#### New features
- Gaussian Gromov Wasserstein loss and mapping (PR #498)
- Template-based Fused Gromov Wasserstein GNN layer in `ot.gnn` (PR #488)
- Make alpha parameter in semi-relaxed Fused Gromov Wasserstein differentiable (PR #483)
- Make alpha parameter in Fused Gromov Wasserstein differentiable (PR #463)
- Added the sparsity-constrained OT solver to `ot.smooth` and added `projection_sparse_simplex` to `ot.utils` (PR #459)
- Add tests on GPU for master branch and approved PR (PR #473)
- Add `median` method to all inherited classes of `backend.Backend` (PR #472)
- Update tests for macOS and Windows, speedup documentation (PR #484)
- Added Proximal Point algorithm to solve GW problems via a new parameter `solver="PPA"` in `ot.gromov.entropic_gromov_wasserstein` + examples (PR #455)
- Added features `warmstart` and `kwargs` in `ot.gromov.entropic_gromov_wasserstein` to respectively perform warmstart on dual potentials and pass parameters to `ot.sinkhorn` (PR #455)
- Added sinkhorn projection based solvers for FGW `ot.gromov.entropic_fused_gromov_wasserstein` and entropic FGW barycenters + examples (PR #455)
- Added features `warmstartT` and `kwargs` to all CG and entropic (F)GW barycenter solvers (PR #455)
- Added entropic semi-relaxed (Fused) Gromov-Wasserstein solvers in `ot.gromov` + examples (PR #455)
- Make marginal parameters optional for (F)GW solvers in `._gw`, `._bregman` and `._semirelaxed` (PR #455)
- Add Entropic Wasserstein Component Analysis (ECWA) in ot.dr (PR #486)
- Added feature Efficient Discrete Multi Marginal Optimal Transport Regularization + examples (PR #454)

#### Closed issues

- Fix gromov conventions (PR #497)
- Fix change in scipy API for `cdist` (PR #487)
- More permissive check_backend (PR #494)
- Fix circleci-redirector action and codecov (PR #460)
- Fix issues with cuda for ot.binary_search_circle and with gradients for ot.sliced_wasserstein_sphere (PR #457)
- Major documentation cleanup (PR #462, PR #467, PR #475)
- Fix gradients for "Wasserstein2 Minibatch GAN" example (PR #466)
- Faster Bures-Wasserstein distance with NumPy backend (PR #468)
- Fix issue backend for ot.sliced_wasserstein_sphere ot.sliced_wasserstein_sphere_unif (PR #471)
- Fix issue with ot.barycenter_stabilized when used with PyTorch tensors and log=True (PR #474)
- Fix `utils.cost_normalization` function issue to work with multiple backends (PR #472)
- Fix pression error on marginal sums and (Issue #429, PR #496)

#### New Contributors
* @kachayev made their first contribution in PR #462
* @liutianlin0121 made their first contribution in PR #459
* @francois-rozet made their first contribution in PR #468
* @framunoz made their first contribution in PR #472
* @SoniaMaz8 made their first contribution in PR #483
* @tomMoral made their first contribution in PR #494
* @12hengyu made their first contribution in PR #454

## 0.9.0
*April 2023*

This new release contains so many new features and bug fixes since 0.8.2 that we
decided to make it a new minor release at 0.9.0.

The release contains many new features. First we did a major
update of all Gromov-Wasserstein solvers that brings up to 30% gain in
computation time (see PR #431) and allows the GW solvers to work on non symmetric
matrices. It also brings novel solvers for the very
efficient [semi-relaxed GW problem
](https://pythonot.github.io/master/auto_examples/gromov/plot_semirelaxed_fgw.html#sphx-glr-auto-examples-gromov-plot-semirelaxed-fgw-py)
that can be used to find the best re-weighting for one of the distributions. We
also now have fast and differentiable solvers for [Wasserstein on the circle](https://pythonot.github.io/master/auto_examples/plot_compute_wasserstein_circle.html#sphx-glr-auto-examples-plot-compute-wasserstein-circle-py) and
[sliced Wasserstein on the
sphere](https://pythonot.github.io/master/auto_examples/backends/plot_ssw_unif_torch.html#sphx-glr-auto-examples-backends-plot-ssw-unif-torch-py).
We are also very happy to provide new OT barycenter solvers such as the [Free
support Sinkhorn
barycenter](https://pythonot.github.io/master/auto_examples/barycenters/plot_free_support_sinkhorn_barycenter.html#sphx-glr-auto-examples-barycenters-plot-free-support-sinkhorn-barycenter-py)
and the [Generalized Wasserstein
barycenter](https://pythonot.github.io/master/auto_examples/barycenters/plot_generalized_free_support_barycenter.html#sphx-glr-auto-examples-barycenters-plot-generalized-free-support-barycenter-py).
A new differentiable solver for OT across spaces that provides OT plans
between samples and features simultaneously and
called [Co-Optimal
Transport](https://pythonot.github.io/master/auto_examples/others/plot_COOT.html)
has also been implemented. Finally we began working on OT between Gaussian distributions and
now provide differentiable estimation for the Bures-Wasserstein [divergence](https://pythonot.github.io/master/gen_modules/ot.gaussian.html#ot.gaussian.bures_wasserstein_distance) and
[mappings](https://pythonot.github.io/master/auto_examples/domain-adaptation/plot_otda_linear_mapping.html#sphx-glr-auto-examples-domain-adaptation-plot-otda-linear-mapping-py).

Another important first step toward POT 1.0 is the
implementation of a unified API for OT solvers with introduction of [`ot.solve`](https://pythonot.github.io/master/all.html#ot.solve)
function that can solve (depending on parameters) exact, regularized and
unbalanced OT and return a new
[`OTResult`](https://pythonot.github.io/master/gen_modules/ot.utils.html#ot.utils.OTResult)
object. The idea behind this new API is to facilitate exploring different solvers
with just a change of parameter and get a more unified API for them. We will keep
the old solvers API for power users but it will be the preferred way to solve
problems starting from release 1.0.0.
We provide below some examples of use for the new function and how to
recover different aspects of the solution (OT plan, full loss, linear part of the
loss, dual variables) :
```python
#Solve exact ot
sol = ot.solve(M)

# get the results
G = sol.plan # OT plan
ot_loss = sol.value # OT value (full loss for regularized and unbalanced)
ot_loss_linear = sol.value_linear # OT value for linear term np.sum(sol.plan*M)
alpha, beta = sol.potentials # dual potentials

# direct plan and loss computation
G = ot.solve(M).plan
ot_loss = ot.solve(M).value

# OT exact with marginals a/b
sol2 = ot.solve(M, a, b)

# regularized and unbalanced OT
sol_rkl = ot.solve(M, a, b, reg=1) # KL regularization
sol_rl2 = ot.solve(M, a, b, reg=1, reg_type='L2')
sol_ul2 = ot.solve(M, a, b, unbalanced=10, unbalanced_type='L2')
sol_rkl_ukl = ot.solve(M, a, b, reg=10, unbalanced=10) # KL + KL

```
The function is fully compatible with backends and will be implemented for
different types of distribution support (empirical distributions, grids) and OT
problems (Gromov-Wasserstein) in the new releases. This new API is not yet
presented in the kickstart part of the documentation as there is a small change
that it might change
when implementing new solvers but we encourage users to play with it.

Finally, in addition to those many new this release fixes 20 issues (some long
standing) and we want to thank all the contributors who made this release so
big. More details below.


#### New features
- Added feature to (Fused) Gromov-Wasserstein solvers inherited from `ot.optim` to support relative and absolute loss variations as stopping criterions (PR #431)
- Added feature to (Fused) Gromov-Wasserstein solvers to handle asymmetric matrices (PR #431)
- Added semi-relaxed (Fused) Gromov-Wasserstein solvers in `ot.gromov` + examples (PR #431)
- Added the spherical sliced-Wasserstein discrepancy in `ot.sliced.sliced_wasserstein_sphere` and `ot.sliced.sliced_wasserstein_sphere_unif` + examples (PR #434)
- Added the Wasserstein distance on the circle in ``ot.lp.solver_1d.wasserstein_circle`` (PR #434)
- Added the Wasserstein distance on the circle (for p>=1) in `ot.lp.solver_1d.binary_search_circle` + examples (PR #434)
- Added the 2-Wasserstein distance on the circle w.r.t a uniform distribution in `ot.lp.solver_1d.semidiscrete_wasserstein2_unif_circle` (PR #434)
- Added Bures Wasserstein distance in `ot.gaussian` (PR ##428)
- Added Generalized Wasserstein Barycenter solver + example (PR #372), fixed graphical details on the example (PR #376)
- Added Free Support Sinkhorn Barycenter + example (PR #387)
- New API for OT solver using function `ot.solve` (PR #388)
- Backend version of `ot.partial` and `ot.smooth` (PR #388 and #449)
- Added argument for warmstart of dual potentials in Sinkhorn-based methods in `ot.bregman` (PR #437)
- Added parameters method in `ot.da.SinkhornTransport` (PR #440)
- `ot.dr` now uses the new Pymanopt API and POT is compatible with current
Pymanopt (PR #443)
- Added CO-Optimal Transport solver + examples (PR #447)
- Remove the redundant `nx.abs()` at the end of `wasserstein_1d()` (PR #448)
- Backend version of `ot.partial` and `ot.smooth` (PR #388)
- Added argument for warmstart of dual vectors in Sinkhorn-based methods in `ot.bregman` (PR #437)

#### Closed issues

Expand Down Expand Up @@ -198,11 +71,9 @@ PR #413)
- Fix an issue where the parameter `stopThr` in `empirical_sinkhorn_divergence` was rendered useless by subcalls
that explicitly specified `stopThr=1e-9` (Issue #421, PR #422).
- Fixed a bug breaking an example where we would try to make an array of arrays of different shapes (Issue #424, PR #425)
- Fixed an issue with the documentation gallery section (PR #444)
- Fixed issues with cuda variables for `line_search_armijo` and `entropic_gromov_wasserstein` (Issue #445, #PR 446)


## 0.8.2
*April 2022*

This releases introduces several new notable features. The less important
but most exiting one being that we now have a logo for the toolbox (color
Expand Down Expand Up @@ -346,7 +217,7 @@ a [Generative Network
(GAN)](https://PythonOT.github.io/auto_examples/backends/plot_wass2_gan_torch.html),
for a [sliced Wasserstein gradient
flow](https://PythonOT.github.io/auto_examples/backends/plot_sliced_wass_grad_flow_pytorch.html)
and [optimizing the Gromov-Wasserstein distance](https://PythonOT.github.io/auto_examples/backends/plot_optim_gromov_pytorch.html). Note that the Jax backend is still in early development and quite
and [optimizing the Gromov-Wassersein distance](https://PythonOT.github.io/auto_examples/backends/plot_optim_gromov_pytorch.html). Note that the Jax backend is still in early development and quite
slow at the moment, we strongly recommend for Jax users to use the [OTT
toolbox](https://github.com/google-research/ott) when possible.
As a result of this new feature,
Expand All @@ -358,7 +229,7 @@ Pointwise Gromov
Wasserstein](https://PythonOT.github.io/auto_examples/gromov/plot_gromov.html#compute-gw-with-a-scalable-stochastic-method-with-any-loss-function),
Sinkhorn in log space with `method='sinkhorn_log'`, [Projection Robust
Wasserstein](https://PythonOT.github.io/gen_modules/ot.dr.html?highlight=robust#ot.dr.projection_robust_wasserstein),
ans [debiased Sinkhorn barycenters](https://PythonOT.github.ioauto_examples/barycenters/plot_debiased_barycenter.html).
ans [deviased Sinkorn barycenters](https://PythonOT.github.ioauto_examples/barycenters/plot_debiased_barycenter.html).

This release will also simplify the installation process. We have now a
`pyproject.toml` that defines the build dependency and POT should now build even
Expand Down Expand Up @@ -499,15 +370,15 @@ are coming for the next versions.

#### Closed issues

- Add JMLR paper to the readme and Mathieu Blondel to the Acknowledgments (PR
- Add JMLR paper to the readme and Mathieu Blondel to the Acknoledgments (PR
#231, #232)
- Bug in Unbalanced OT example (Issue #127)
- Clean Cython output when calling setup.py clean (Issue #122)
- Various Macosx compilation problems (Issue #113, Issue #118, PR#130)
- EMD dimension mismatch (Issue #114, Fixed in PR #116)
- 2D barycenter bug for non square images (Issue #124, fixed in PR #132)
- Bad value in EMD 1D (Issue #138, fixed in PR #139)
- Log bugs for Gromov-Wasserstein solver (Issue #107, fixed in PR #108)
- Log bugs for Gromov-Wassertein solver (Issue #107, fixed in PR #108)
- Weight issues in barycenter function (PR #106)

## 0.6.0
Expand Down Expand Up @@ -538,9 +409,9 @@ a solver for [Unbalanced OT
barycenters](https://github.com/rflamary/POT/blob/master/notebooks/plot_UOT_barycenter_1D.ipynb).
A new variant of Gromov-Wasserstein divergence called [Fused
Gromov-Wasserstein](https://pot.readthedocs.io/en/latest/all.html?highlight=fused_#ot.gromov.fused_gromov_wasserstein)
has been also contributed with examples of use on [structured
has been also contributed with exemples of use on [structured
data](https://github.com/rflamary/POT/blob/master/notebooks/plot_fgw.ipynb) and
computing [barycenters of labeled
computing [barycenters of labeld
graphs](https://github.com/rflamary/POT/blob/master/notebooks/plot_barycenter_fgw.ipynb).


Expand Down Expand Up @@ -601,7 +472,7 @@ and [free support](https://github.com/rflamary/POT/blob/master/notebooks/plot_fr
implementation of entropic OT.

POT 0.5 also comes with a rewriting of ot.gpu using the cupy framework instead of
the unmaintained cudamat. Note that while we tried to keep changes to the
the unmaintained cudamat. Note that while we tried to keed changes to the
minimum, the OTDA classes were deprecated. If you are happy with the cudamat
implementation, we recommend you stay with stable release 0.4 for now.

Expand All @@ -625,7 +496,7 @@ and new POT contributors (you can see the list in the [readme](https://github.co
* Stochastic OT in the dual and semi-dual (PR #52 and PR #62)
* Free support barycenters (PR #56)
* Speed-up Sinkhorn function (PR #57 and PR #58)
* Add convolutional Wasserstein barycenters for 2D images (PR #64)
* Add convolutional Wassersein barycenters for 2D images (PR #64)
* Add Greedy Sinkhorn variant (Greenkhorn) (PR #66)
* Big ot.gpu update with cupy implementation (instead of un-maintained cudamat) (PR #67)

Expand Down Expand Up @@ -676,15 +547,15 @@ This release contains a lot of contribution from new contributors.
* new notebooks for emd computation and Wasserstein Discriminant Analysis
* relocate notebooks
* update documentation
* clean_zeros(a,b,M) for removing zeros in sparse distributions
* clean_zeros(a,b,M) for removimg zeros in sparse distributions
* GPU implementations for sinkhorn and group lasso regularization


## V0.2
*7 Apr 2017*

* New dimensionality reduction method (WDA)
* Efficient method emd2 returns only transport (in parallel if several histograms given)
* Efficient method emd2 returns only tarnsport (in paralell if several histograms given)



Expand Down Expand Up @@ -725,4 +596,4 @@ It provides the following solvers:
* Optimal transport for domain adaptation with group lasso regularization
* Conditional gradient and Generalized conditional gradient for regularized OT.

Some demonstrations (both in Python and Jupyter Notebook format) are available in the examples folder.
Some demonstrations (both in Python and Jupyter Notebook format) are available in the examples folder.
Loading
Loading