Merge branch 'master' into feature/alldifferent

.github/workflows/TagBot.yml

@@ -0,0 +1,15 @@
+name: TagBot
+on:
+  issue_comment:
+    types:
+      - created
+  workflow_dispatch:
+jobs:
+  TagBot:
+    if: github.event_name == 'workflow_dispatch' || github.actor == 'JuliaTagBot'
+    runs-on: ubuntu-latest
+    steps:
+      - uses: JuliaRegistries/TagBot@v1
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          ssh: ${{ secrets.DOCUMENTER_KEY }}

LICENSE

@@ -1,19 +1,28 @@
-Copyright (c) 2020 "Ilan Coulon", "Félix Chalumeau"
+BSD 3-Clause License
 
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
+Copyright (c) 2021: CORAIL-research
 
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
 
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+1. Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its
+   contributors may be used to endorse or promote products derived from
+   this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Project.toml

@@ -1,7 +1,7 @@
 name = "SeaPearl"
 uuid = "c13076dc-bdcd-48ba-bc88-4b44c2587ab3"
 authors = ["Ilan Coulon", "Félix Chalumeau"]
-version = "0.1.0"
+version = "0.2.0"
 
 [deps]
 DataStructures = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"

README.md

@@ -1,29 +1,22 @@
-# SeaPearl
+# SeaPearl.jl
 
-[![Build Status](https://travis-ci.com/ilancoulon/SeaPearl.jl.svg?token=txSsK23gqDP8efBDxJzv&branch=master)](https://travis-ci.com/ilancoulon/SeaPearl.jl)
+SeaPearl is a Constraint Programming solver that can use Reinforcement Learning agents as value-selection heuristics, using graphs as inputs for the agent's approximator. It is to be seen as a tool for researchers that gives the possibility to go above and beyond what has already been done with it.
 
-Hybrid solver using Constraint programming and Reinforcement learning. 
-Ilan Coulon, Félix Chalumeau & Quentin Cappart. 
+The paper accompanying this solver can be found on the [arXiv](https://arxiv.org/abs/2102.09193v1). If you use SeaPearl in your research, please cite our work.
 
-## Useful links 
+The RL agents are defined using [ReinforcementLearning.jl](https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl), their inputs are dealt with using [GeometricFlux.jl](https://github.com/FluxML/GeometricFlux.jl) and [Flux.jl](https://github.com/FluxML/Flux.jl). The CP part, inspired from [MiniCP](http://www.minicp.org/), is focused on readability. The code is meant to be clear and modulable so that researchers could easily get access to CP data and use it as input for their ML model.
 
-Here are useful links 
+## Installation
 
-### User interface:
-Documentation of the modeling language we want to use, called JuMP http://www.juliaopt.org/JuMP.jl/v0.19.0/
+```julia
+]add SeaPearl
+```
 
-### Constraint Programming:
-The architecture is inspired by miniCP, which is a CP solver in Java. Here is its documentation: http://www.minicp.org/
-One can also check slides used to present miniCP's structure: https://school.a4cp.org/summer2017/slidedecks/MiniCP.pdf
+## Use
 
-### Reinforcement Learning:
-We are using native julia packages ReinforcementLearning.jl https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl
-The types were inheriting are in ReinforcementLearningBase.jl and ReinforcementLearningCore.jl and some interesting examples can be found in ReinforcementLearningZoo.jl and ReinforcementLearningEnvironments.jl
+Working examples can be found in [SeaPearlZoo](https://github.com/corail-research/SeaPearlZoo).
 
-### Machine Learning:
-They are two main packages for ML at the moment. KNet.jl & Flux.jl
-We are planning to use Flux.jl. Github repo : https://github.com/FluxML/Flux.jl & documentation : https://fluxml.ai/
+## Contribution
+
+All PRs and issues are welcome.
 
-### Working on graphs:
-Even if this part is still under debate, we might be working with graphs. In this case, we will use LightGraphs.jl https://github.com/JuliaGraphs/LightGraphs.jl
-And the machine learning on graphs would be eased by GeometricFlux.jl https://github.com/yuehhua/GeometricFlux.jl

docs/make.jl

@@ -7,9 +7,10 @@ makedocs(
     format = Documenter.HTML(),
     modules = [SeaPearl],
     pages = ["Home" => "index.md",
-    "Trailer" => "trailer.md",
-    "Constraint Programming solver" =>
-        ["Variables" => "CP/int_variable.md"],
+    "Constraint Programming solver" => [
+        "Variables" => "CP/int_variable.md",
+        "Trailer" => "CP/trailer.md",
+    ],
     "Building Models" =>
       ["Basics" => "models/basics.md"],
     "Community" => "community.md"],

docs/src/CP/int_variable.md

@@ -1,11 +1,11 @@
 # Variables
 
-The implementation of variables in SeaPearl is heavily inspired on [MiniCP](https://minicp.readthedocs.io/en/latest/learning_minicp/part_2.html). If you have some troubles understanding how it works, you can get more visual explanations by reading their [slides](https://inginious.org/course/minicp/domains).
+## Integer variables
+
+The implementation of integer variables in SeaPearl is heavily inspired on [MiniCP](https://minicp.readthedocs.io/en/latest/learning_minicp/part_2.html). If you have some troubles understanding how it works, you can get more visual explanations by reading their [slides](https://inginious.org/course/minicp/domains).
 
 The variables are all a subset of `AbstractIntVar`.
-```@docs
-SeaPearl.AbstractIntVar
-```
+
 
 Every `AbstractIntVar` must have a unique `id` that you can retrieve with `id`.
 ```@docs
@@ -15,14 +15,14 @@ SeaPearl.assign!(::SeaPearl.AbstractIntVar, ::Int)
 SeaPearl.assignedValue
 ```
 
-## IntVar
+### IntVar
 
 ```@docs
 SeaPearl.IntVar
 SeaPearl.IntVar(::Int, ::Int, ::String, ::SeaPearl.Trailer)
 ```
 
-## IntDomain
+### IntDomain
 
 ```@docs
 SeaPearl.AbstractIntDomain
@@ -44,3 +44,16 @@ SeaPearl.updateBoundsFromRemovedVal!
 SeaPearl.minimum
 SeaPearl.maximum
 ```
+
+If you want to express some variations of an integer variable ``x`` (for example ``-x`` or ``a x`` with ``a > 0``) in a constraint, you can use the `IntVarView` types:
+
+### IntVarView
+
+```@docs
+SeaPearl.IntVarViewMul
+SeaPearl.IntVarViewMul(x::AbstractIntVar, a::Int, id::String)
+SeaPearl.IntVarViewOpposite
+SeaPearl.IntVarViewOpposite(x::AbstractIntVar, id::String)
+SeaPearl.IntVarViewOffset
+SeaPearl.IntVarViewOffset(x::AbstractIntVar, id::String)
+```

docs/src/index.md

@@ -3,11 +3,11 @@
 SeaPearl was created as a way for researchers to have a constraint programming solver that can
 integrate seamlessly with Reinforcement Learning technologies, using them as heuristics for value selection during branching.
 
+The paper accompanying this solver can be found on the [arXiv](https://arxiv.org/abs/2102.09193v1). If you use SeaPearl in your research, please cite our work.
+
 The Julia language was chosen for this project as we believe it is one of the few languages that can be used for Constraint Programming as well as Machine/Deep Learning.
 
 The constraint programming part, whose architecture is heavily inspired from [Mini-CP framework](https://minicp.readthedocs.io/en/latest/intro.html), is focused on readability. The code was meant to be clear and modulable so that researchers could easily get access to CP data and use it as input for their ML model.
 
 SeaPearl comes with a set of examples that can be found in the [SeaPearlZoo](https://github.com/corail-reseach/SeaPearlZoo) repository.
 
-The new thing in SeaPearl is that the Reinforcement Learning part does *not* directly solve the CP problem, it is only used as a value-selection heuristic and therefore lets you use constraint propagation and different search strategies to, for example, prove optimality. The RL agent is only called when the CP solver has to branch and assign an arbitrary value to a variable.
-Therefore, the RL agent can be trained to find good solutions as well as proving their optimality and is completely integrated into the CP framework.

src/CP/variables/IntVarView.jl

@@ -11,7 +11,7 @@ struct IntVarViewMul <: IntVarView
     id              ::String
 
     """
-        IntVarViewMul(x::IntVar, a::Int, id::String)
+        IntVarViewMul(x::AbstractIntVar, a::Int, id::String)
 
     Create a *fake* variable `y`, such that `y == a*x`. This variable behaves like an usual one.
     """
@@ -32,7 +32,7 @@ struct IntVarViewOpposite <: IntVarView
     id              ::String
 
     """
-    IntVarViewOpposite(x::IntVar, id::String)
+    IntVarViewOpposite(x::AbstractIntVar, id::String)
 
     Create a *fake* variable `y`, such that `y = -x`. This variable behaves like an usual one.
     """
@@ -54,7 +54,7 @@ struct IntVarViewOffset <: IntVarView
     id              ::String
 
     """
-    IntDomainViewOffset(x::IntVar, id::String)
+    IntVarViewOffset(x::AbstractIntVar, id::String)
 
     Create a *fake* variable `y`, such that `y = x + c`. This variable behaves like an usual one.
     """