The OVERHAUL (#78)

* Complete refactor of the Regularized structures:
  - Only one struct named `Regularized`, every regularized layer is a particular case of it
  - Specific constructors for `SparseArgmax`, `SoftArgmax`, and `RegularizedFrankWolfe`
  - Now we can also use `Regularized` with a custom optimizer (we may need to test this feature)

* The OVERHAUL

* Fix tutorial in tests, more docs

* Bump version

* Fix Léo's remarks

* Fix imports

* typo

* Don't export compress_dist

---------

Co-authored-by: BatyLeo <leo.baty67@gmail.com>

Author: gdalle

.gitignore

@@ -3,8 +3,10 @@
 *.jl.mem
 /Manifest.toml
 /docs/build/
+/docs/src/index.md
+/docs/src/tutorial.md
 /test/profiling.jl
-
+.vscode
 *.tar.gz
 *.zip
 *.csv

CITATION.bib

@@ -2,7 +2,7 @@ @misc{InferOpt.jl
 	author  = {Guillaume Dalle, Léo Baty, Louis Bouvier and Axel Parmentier},
 	title   = {InferOpt.jl},
 	url     = {https://github.com/axelparmentier/InferOpt.jl},
-	version = {v0.4.0},
-	year    = {2022},
+	version = {v0.5.0},
+	year    = {2023},
 	month   = {7}
 }

Project.toml

@@ -1,14 +1,13 @@
 name = "InferOpt"
 uuid = "4846b161-c94e-4150-8dac-c7ae193c601f"
 authors = ["Guillaume Dalle", "Léo Baty", "Louis Bouvier", "Axel Parmentier"]
-version = "0.5.0-DEV"
+version = "0.5.0"
 
 [deps]
 ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
 DifferentiableFrankWolfe = "b383313e-5450-4164-a800-befbd27b574d"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
-SimpleTraits = "699a6c99-e7fa-54fc-8d76-47d257e15c1d"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 ThreadsX = "ac1d9e8a-700a-412c-b207-f0111f4b6c0d"
@@ -22,7 +21,6 @@ InferOptFrankWolfeExt = "DifferentiableFrankWolfe"
 [compat]
 ChainRulesCore = "1"
 DifferentiableFrankWolfe = "0.1.2"
-SimpleTraits = "0.9"
 StatsBase = "0.33, 0.34"
 TestItemRunner = "0.2.2"
 ThreadsX = "0.1.11"

README.md

@@ -22,7 +22,7 @@ To install the stable version, open a Julia REPL and run the following command:
 julia> using Pkg; Pkg.add("InferOpt")
 ```
 
-To install the development version, run this command instead:
+To install the development version (*recommended for now*), run this command instead:
 
 ```julia
 julia> using Pkg; Pkg.add(url="https://github.com/axelparmentier/InferOpt.jl")
@@ -33,3 +33,11 @@ julia> using Pkg; Pkg.add(url="https://github.com/axelparmentier/InferOpt.jl")
 If you use our package in your research, please cite the following paper:
 
 > [Learning with Combinatorial Optimization Layers: a Probabilistic Approach](https://arxiv.org/abs/2207.13513) - Guillaume Dalle, Léo Baty, Louis Bouvier and Axel Parmentier (2022)
+
+## Related packages
+
+The following libraries implement similar functionalities:
+
+- [ImplicitDifferentiation.jl](https://github.com/gdalle/ImplicitDifferentiation.jl): automatic differentiation of implicit functions 
+- [DiffOpt.jl](https://github.com/jump-dev/DiffOpt.jl): differentiating convex optimization programs w.r.t. program parameters
+- [JAXopt](https://github.com/google/jaxopt): hardware accelerated, batchable and differentiable optimizers in JAX