Blog post for CP ecosystem

_posts/2021-09-18-cp-0-6.md

@@ -0,0 +1,74 @@
+---
+layout: post
+title: "An update on constraint programming in JuMP"
+date: 2021-09-18
+categories: [releases]
+permalink: /blog/constraint-programming-update/
+---
+
+_Author: Thibaut Cuvelier (@dourouc05)_
+
+JuMP and MathOptInterface are oriented towards traditional mathematical
+optimization, encompassing problem classes such as mixed-integer linear programs
+and conic optimization. However, the MathOptInterface API is amenable to other
+kinds of formalism, including [constraint programming](https://en.wikipedia.org/wiki/Constraint_programming).
+
+In constrast to linear or conic programs, constraint programs typically have no
+objective function, but a much wider variety of constraints that are supported.
+The best-known constraint is probably `all_different`, which models the fact
+that several integer-valued variables should take different values (i.e., no two
+variables are allowed to have the same value). However, there are a range of
+other constraint types, with a focus on describing combinatorial structures.
+Efficient solvers have been written for constraint programs.
+
+There are a few tools for constraint programming outside the Julia ecosystem.
+The most popular are [MiniZinc](https://www.minizinc.org) and
+[OR-Tools](https://developers.google.com/optimization).
+
+The goal of the [ConstraintProgrammingExtensions.jl](https://github.com/dourouc05/ConstraintProgrammingExtensions.jl)
+package is to bring first-class support for constraint programming to JuMP and
+MathOptInterface.
+
+Discussions about constraint programming support in the JuMP ecosystem started
+[in 2019](https://github.com/jump-dev/JuMP.jl/issues/2014). The original scope
+was quite limited, as it only planned to propose some modeling capabilities to
+JuMP. However, quite soon, it morphed into a larger proposal of a full
+[constraint programming environment](https://github.com/jump-dev/JuMP.jl/issues/2227),
+accessible from the same API as JuMP, including wrapping various specialized
+constraint programming solvers.
+
+In the last few months, the constraint programming ecosystem in JuMP has made
+tremendous progress. Highlights include:
+
+- The v0.6 release of [ConstraintProgrammingExtensions.jl](https://github.com/dourouc05/ConstraintProgrammingExtensions.jl).
+  Highlights of the ConstraintProgrammingExtensions.jl package include:
+  * set definitions for [dozens of constraint types](http://tcuvelier.be/ConstraintProgrammingExtensions.jl/dev/reference/sets/)
+  * a number of [bridges](http://tcuvelier.be/ConstraintProgrammingExtensions.jl/dev/reference/bridges_sets/),
+    including many bridges to convert constraint programming models into
+    standard mixed-integer programming models
+  * Support for the [FlatZinc file format](https://www.minizinc.org/doc-2.5.5/en/fzn-spec.html), a standard in constraint programming
+    to exchange models (similar to the LP and MPS formats for linear programming)
+
+- Interfacing several solvers through ConstraintProgrammingExtensions.jl. These
+  include:
+  * The pure-Julia [ConstraintSolver.jl](https://github.com/Wikunia/ConstraintSolver.jl/)
+  * The industry standard [IBM CPLEX CP Optimizer](https://github.com/dourouc05/CPLEXCP.jl/)
+  * The open-source [Chuffed](https://github.com/dourouc05/Chuffed.jl/)
+
+  Other pure-Julia constraint solvers have been released, but they do not yet
+  use ConstraintProgrammingExtensions API:
+  * [JuliaConstraints](https://github.com/JuliaConstraints)
+  * [SeaPearl](https://github.com/corail-research/SeaPearl.jl)
+
+- A session on constraint programming talks at JuMP-dev 2021
+  * [The design of MiniZinc](https://www.youtube.com/watch?v=tHi0Jd7vgVA)
+  * [ConstraintProgrammingExtensions.jl](https://www.youtube.com/watch?v=B0lO6HdlFAw&list=PLP8iPy9hna6Q343_8sSq4f306VGLW4TLK&index=104)
+  * [ConstraintSolver.jl](https://www.youtube.com/watch?v=FXqQc8bi_vw&list=PLP8iPy9hna6Q343_8sSq4f306VGLW4TLK&index=101)
+  * [JuliaConstraints](https://www.youtube.com/watch?v=G4siuvNMj0c&list=PLP8iPy9hna6Q343_8sSq4f306VGLW4TLK&index=4)
+
+This is the start of our journey towards constraint programming in JuMP. Although MathOptInterface and JuMP are nearing
+their 1.0 release, the constraint programming ecosystem has still a long way to go.
+In the near term, more modelling features should be added as extensions to JuMP and
+more solvers should be made accessible. In the longer term, constraint
+programming modelling will also benefit from the [next generation of nonlinear
+support](https://github.com/jump-dev/MathOptInterface.jl/issues/846).