0.4.0: GPU tests, re-adding jit, improved NN OT solvers, and misc.

What's Changed

• Fix badges by @michalk8 in #211
• Feature/faster gw init by @michalk8 in #213
• Fix/device array check by @michalk8 in #216
• Remove support for Python 3.7 by @michalk8 in #227
• chore: add yapf config to pyproject.toml by @SauravMaheshkar in #230
• Remove functional API by @michalk8 in #222
• Docs/notebooks myst by @michalk8 in #233
• [ci skip] Add environment.yml for binder by @michalk8 in #236
• Use lax.cond when computing error in Sinkhorn by @michalk8 in #235
• Fix/debiased entropic map by @michalk8 in #238
• Misc/tox by @michalk8 in #239
• Re-add jit argument by @michalk8 in #221
• Add GPU test runner by @michalk8 in #242
• Remove old Transport interface by @michalk8 in #243
• Remove shape check in LRCGeometry by @michalk8 in #247
• Add regularized costs by @michalk8 in #244
• Check only whether problem can be LR by @michalk8 in #248
• Improvements to the neural dual solver by @bamos in #219

New Contributors

• @SauravMaheshkar made their first contribution in #230

Full Changelog: 0.3.1...0.4.0

0.3.1 with new cost features, different code layout, changes in potentials handling. Some solvers in function form deprecated.

What's Changed

• Fix jax.device_put for potentials by @michalk8 in #183
• Introduce primal_cost and dual_cost for Sinkhorn outputs (only primal for LR) for arbitrary geometries. by @marcocuturi in #184
• Feature/src layout by @michalk8 in #188
• Handle better inf with p parameter in SqPNorm by @marcocuturi in #189
• Feature/graph normalized laplacian by @michalk8 in #191
• Update installation instruction by @michalk8 in #195
• Feature/center potentials by @michalk8 in #194
• Remove jit from solvers by @michalk8 in #192
• Accelerating unbalanced OT by @michalk8 in #197
• Feature/depracate functional api by @michalk8 in #204
• Fixes numerical errors in Bures barycenter, and sqrtm, due to low default precision. by @marcocuturi in #205
• Update docstrings for deprecated functions by @michalk8 in #207

Full Changelog: 0.3.0...0.3.1

New organization for code; Changes in `costs` for `PointClouds`

The main changes in this version are twofold:

• Changes in PointCloud geometry, and more specifically handling of cost function. The power parameter that was used to optionally pass a c(x,y) = CostFn(x,y) ** power is now deprecated. One can add it manually by defining a custom CostFn. To compensate this change, a new class of translation invariant costs (TICost) has been created, from which most costs now inherit, defined as c(x,y) = h(x-y). Additionally, to handle Brenier's theorem, the user has the option of passing on the legendre transform of h, h_legendre.
• The core folder was too horizontal, containing various modules. It has been reorganized and split into 3 modules that make more sense, problems (to describe OT problems), solvers (solve them) and initializers (as optional modules to help solvers). The latter two have an arborescence that reflects that in problems.

What's Changed

• fix point_clouds.ipynb by @marcocuturi in #159
• Deprecate power in PointCloud, introduce TICost and use it to compute Entropic (Brenier) maps. by @marcocuturi in #167
• Misc/project structure by @michalk8 in #176

Full Changelog: 0.2.11...0.3.0

adaptive momentum by default, Euclidean is now SqEuclidean (W2), Euclidean is for W1 and uses `jnp.linalg.norm` to ensure differentiability

What's Changed

• default momentum with start=300 by @marcocuturi in #155
• Unbalanced gromov by @LaetitiaPapaxanthos in #128
• Euclidean / SqEuclidean and changes in power by @marcocuturi in #157

Full Changelog: 0.2.10...0.2.11

Fix various bugs and add Meta initializer

What's Changed

• Make GMM documentation more visible, fix bug in M step of EM algorithm by @geoff-davis in #144
• Feature/initializers as literals by @michalk8 in #148
• Fix fused_penalty and scale_cost in LRGW by @michalk8 in #147
• Add dtype property by @michalk8 in #150
• Meta OT Initializer by @bamos in #145
• fix power=1.0 using abs on cost values when needed by @marcocuturi in #153

New Contributors

• @geoff-davis made their first contribution in #144

Full Changelog: 0.2.9...0.2.10

0.2.9: new initializations for Sinkhorn, LR Sinkhorn and LR GW (including k-means++ implementations), entropic map, faster Sinkhorn implicit diff

What's Changed

• Feature/kmeans++ by @michalk8 in #120
• Feature/ Introduce initialization methods for Sinkhorn by @JTT94 in #98
• Pin sphinx-book-theme>=0.3.3 by @michalk8 in #123
• Additional Docstrings and Comments to NeuralDual-Module by @bunnech in #122
• Allow k-means to be differentiable when using while loop by @michalk8 in #130
• Feature/graph geometry by @michalk8 in #126
• LR Sinkhorn improvements by @meyerscetbon in #111
• Refactor GW initialization by @michalk8 in #133
• Change LR initializer to random by @michalk8 in #134
• Generalized k-means initialization for LR Sinkhorn/Gromov-Wasserstein by @michalk8 in #135
• Switch to one linear system per implicit differentiation by @ersisimou in #136
• Feature/disable warm start by @michalk8 in #138
• Add entropic map by @michalk8 in #142

Full Changelog: 0.2.8...0.2.9

Extension of barycenters to (Fused) GW, several bug fixes around LR Sinkhorn, parallel (segment) code for Sinkhorn, inits for ICNNs

What's Changed

• Refactoring problems.py by @marcocuturi in #86
• Feature/gw losses by @michalk8 in #88
• Fix/lr linear apply by @michalk8 in #91
• Addition of ICNN Initialization Schemes by @bunnech in #90
• Fix/point cloud apply cost fn by @michalk8 in #93
• Feature/lr gw apply output by @michalk8 in #96
• Remove legacy online option by @michalk8 in #97
• Feature/fast tests by @michalk8 in #101
• Fix/axis norm by @michalk8 in #103
• Generic LR cost decomposition by @michalk8 in #99
• Remove pytest-memray from test req, use only in CI by @michalk8 in #107
• demo and fixes for barycenters of GMMs by @ersisimou in #89
• Add (F)GW barycenters by @michalk8 in #87
• Feature/notebook tests by @michalk8 in #108
• [ci skip] Fix duplicate bibtex entry and a typo by @michalk8 in #109
• GW different cost ranks by @michalk8 in #113
• segment_sinkhorn by @marcocuturi in #114
• Create CODE_OF_CONDUCT.md by @marcocuturi in #118
• Fix/padded scaling by @michalk8 in #116

Full Changelog: 0.2.7...0.2.8

Fixing continuous barycenter, cleaning up docs and more generally the packaging of OTT.

What's Changed

• Pre-commit by @michalk8 in #71
• Fix/online gw by @michalk8 in #80
• Feature/improve packaging by @michalk8 in #82
• Ersi gauss bary by @ersisimou in #81

New Contributors

• @ersisimou made their first contribution in #81

Full Changelog: 0.2.6...0.2.7

Quite a few bug fixes on jitting / scaling of costs / convergence

What's Changed

• Fix jit issue with relative epsilon and copying epsilons by @JTT94 in #52
• Bug/scale cost by @MUCDK in #54
• ensure 3 geom objects returned when copying epsilons by @JTT94 in #56
• fixed logic of __call__ of LRSinkhorn to prevent uninterpretable error by @MUCDK in #57
• Project import generated by Copybara. by @LaetitiaPapaxanthos in #58
• General cost fn by @JTT94 in #61
• Set converged logic by @zoepiran in #62
• Fix LRSinkhornOutput.apply with batches by @michalk8 in #63
• Fix is_all_geoms_lr not returning by @michalk8 in #67
• Feature/quadratic problem scale by @michalk8 in #66
• Fix LR tests, fused_penalty property, tolerances by @michalk8 in #68
• add first version of SCOT demonstration notebook (Gromov-Wasserstein for multi-omics) by @antoinebelloir in #64
• Fix LR Gromov memory by @michalk8 in #70
• Fix scale_cost to LRGeometry, float/max-bound by @michalk8 in #72
• Update README to use Markdown math by @adrhill in #75
• Fix neural dual notebook by @lucaeyring in #76

New Contributors

• @JTT94 made their first contribution in #52
• @MUCDK made their first contribution in #54
• @antoinebelloir made their first contribution in #64
• @adrhill made their first contribution in #75
• @lucaeyring made their first contribution in #76

Full Changelog: 0.2.5...0.2.6

Bug in not_converged flag that was impacting convergence fixed.

What's Changed

• not converged logic by @zoepiran in #50

New Contributors

• @zoepiran made their first contribution in #50

Full Changelog: 0.2.4...0.2.5