Rewrite the package to use Symbolics and ReversePropagation (#212)

* Add source files for rewrite

* Add basic examples

* Add ConstraintProblem

* Bump versions

* Add model.jl

* update Project.toml and workflow

* Update rewrite branch. Working with Symbolics 5,  IntervalArithmetic 0.20

* Pass nanmath=false to build_function

* Fix action of separator on box

* Require Julia 1.10

* Working with new versions of packages

* Add basic usage and image to readme

* Add boundary to plot

* Fix

* Replace PNG with SVG

* Get tests passing with new API

* Clean up by removing unneeded code and MacroTools dep

* Add back missing files after name change

---------

Co-authored-by: lucaferranti <[email protected]>

Author: dpsanders

.DS_Store

@@ -1,10 +1,7 @@
 name: CI
 on:
-  push:
-    branches: "master"
-    tags: ["*"]
-  pull_request:
-  release:
+  - push
+  - pull_request
 jobs:
   test:
     name: Julia ${{ matrix.version }} - ${{ matrix.os }} - ${{ matrix.arch }} - ${{ github.event_name }}
@@ -14,9 +11,7 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.3'  # oldest Julia version that works
-          - '1.8'  # newest Julia version that works with ModelingToolkit
-          - '1'
+          - '1.10'
           - 'nightly'
         os:
           - ubuntu-latest

.github/workflows/CI.yml

@@ -12,5 +12,4 @@ jobs:
       - name: CompatHelper.main()
         env:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-          COMPATHELPER_PRIV: ${{ secrets.COMPATHELPER_PRIV }}
         run: julia -e 'using CompatHelper; CompatHelper.main()'

.github/workflows/CompatHelper.yml

@@ -2,11 +2,4 @@
 *.jl.*.cov
 *.jl.mem
 docs/build/
-
-*.DS_Store
-
-coverage/
-
-.vscode
-
-Manifest.toml
+Manifest.toml

.gitignore

@@ -1,14 +1,14 @@
 # IntervalConstraintProgramming.jl
 
-# v0.11
+#v0.11
 ## Minimum Julia version
 - The minimum Julia version supported is now Julia 1.1
 
-## Functionality's are added
+##Functionality's are added
 - Contractor can be make by just function name only
 - New type of Contractor named as `BasicContractor` can be construct which only contain fields of useful data.
 
-
+ 
 # v0.10
 ## Minimum Julia version
 - The minimum Julia version supported is now Julia 1.0.
@@ -17,7 +17,6 @@
 
 - By the help of `ModelingToolkit.jl` we can construct contractors and separators without the use of macros.
 
-
 # v0.9
 ## Minimum Julia version
 - The minimum Julia version supported is now Julia 0.7. The package is fully compatible with Julia 1.0.
@@ -30,14 +29,12 @@
 ## Minimum Julia version
 - The minimum Julia version required has been bumped to 0.6; this will be the last release to support 0.6.
 
-
 # v0.7
 
 ## New dependency: `IntervalContractors.jl`
 
 The reverse functions used for constraint propagation have been factored out into the `IntervalContractors.jl` package.
 
-
 # v0.6
 ## Minimum Julia version
 - The minimum Julia version required has been bumped to 0.5
@@ -48,7 +45,6 @@ The reverse functions used for constraint propagation have been factored out int
 ## Dependency change
 - The dependency on `ValidatedNumerics.jl` has been replaced by `IntervalArithmetic.jl` and `IntervalRootFinding.jl`
 
-
 # v0.5
 - API change: Contractors now have their dimension as a type parameter
 - Refactoring for type stability
@@ -57,7 +53,6 @@ The reverse functions used for constraint propagation have been factored out int
 - Generated code uses simpler symbols
 - Example notebooks have been split out into a separate repository: https://github.com/dpsanders/IntervalConstraintProgrammingNotebooks
 
-
 # v0.4
 - `@function f(x) = 4x` defines a function
 - Functions may be used inside constraints
@@ -89,7 +84,6 @@ C = @constraint (x-$a)^2 + (y-$b)^2
 The constraint will *not* change if the constants are changed, but may be
 updated (changed) by calling the same `@constraint` command again.  
 
-
 # v0.2
 - `setinverse` now returns an object of type `Paving`  [#17](https://github.com/dpsanders/IntervalConstraintProgramming.jl/pull/17)
 
@@ -101,7 +95,7 @@ updated (changed) by calling the same `@constraint` command again.
 to eliminate collisions with user-defined variables [#20](https://github.com/dpsanders/IntervalConstraintProgramming.jl/pull/20)
 
 
-# v0.1.1
+## v0.1.1
 - Add `sqrtRev` reverse-mode function
 
 - Add solid torus example, including 3D visualization with GLVisualize

NEWS.md

@@ -1,16 +1,25 @@
 name = "IntervalConstraintProgramming"
 uuid = "138f1668-1576-5ad7-91b9-7425abbf3153"
-version = "0.13"
+version = "0.14.0"
 
 [deps]
 IntervalArithmetic = "d1acc4aa-44c8-5952-acd4-ba5d80a2a253"
+IntervalBoxes = "43d83c95-ebbb-40ec-8188-24586a1458ed"
 IntervalContractors = "15111844-de3b-5229-b4ba-526f2f385dc9"
-MacroTools = "1914dd2f-81c6-5fcd-8719-6d5c9610ff09"
-Requires = "ae029012-a4dd-5104-9daa-d747884805df"
+ReversePropagation = "527681c1-8309-4d3f-8790-caf822a419ba"
+StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
+Symbolics = "0c5d862f-8b57-4792-8d23-62f2024744c7"
 
 [compat]
-IntervalArithmetic = "0.16, 0.17, 0.18, 0.19, 0.20"
-IntervalContractors = "0.4"
-MacroTools = "0.4, 0.5"
-Requires = "0.5, 1"
-julia = "1.3"
+IntervalArithmetic = "0.22.12"
+IntervalContractors = "0.5"
+ReversePropagation = "0.3"
+Symbolics = "5, 6"
+julia = "1"
+
+[extras]
+Symbolics = "0c5d862f-8b57-4792-8d23-62f2024744c7"
+Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[targets]
+test = ["Test", "Symbolics"]

Project.toml

@@ -2,7 +2,6 @@
 
 [![Build Status](https://github.com/JuliaIntervals/IntervalConstraintProgramming.jl/workflows/CI/badge.svg)](https://github.com/JuliaIntervals/IntervalConstraintProgramming.jl/actions/workflows/CI.yml)
 [![Docs](https://img.shields.io/badge/docs-stable-blue.svg)](https://juliaintervals.github.io/pages/packages/intervalconstraintprogramming/)
-[![coverage](https://codecov.io/gh/JuliaIntervals/IntervalConstraintProgramming.jl/branch/master/graph/badge.svg)](https://codecov.io/gh/JuliaIntervals/IntervalConstraintProgramming.jl)
 
 This Julia package allows us to specify a set of constraints on real-valued variables,
 given by inequalities, and
@@ -13,14 +12,44 @@ The package is based on interval arithmetic using the
 [`IntervalArithmetic.jl`](https://github.com/JuliaIntervals/IntervalArithmetic.jl) package (co-written by the author),
 in particular multi-dimensional `IntervalBox`es (i.e. Cartesian products of one-dimensional intervals).
 
-## Documentation
+<!-- ## Documentation
 Documentation for the package is available [here](https://juliaintervals.github.io/pages/packages/intervalconstraintprogramming/).
 
-The best way to learn how to use the package is to look at the tutorial, available in the organisation webpage [here](https://juliaintervals.github.io/pages/tutorials/tutorialConstraintProgramming/).
+The best way to learn how to use the package is to look at the tutorial, available in the organisation webpage [here](https://juliaintervals.github.io/pages/tutorials/tutorialConstraintProgramming/). -->
+
+## Basic usage
+
+```jl
+using IntervalArithmetic, IntervalArithmetic.Symbols
+using IntervalConstraintProgramming
+using IntervalBoxes
+using Symbolics
+
+vars = @variables x, y
+
+C1 = constraint(x^2 + 2y^2 ≥ 1, vars)
+C2 = constraint(x^2 + y^2 + x * y ≤ 3, vars)
+C = C1 ⊓ C2
+
+X = IntervalBox(-5..5, 2)
+
+tolerance = 0.05
+inner, boundary = pave(X, C, tolerance)
+
+# plot the result:
+using Plots
+
+plot(collect.(inner), aspectratio=1, lw=0, label="inner");
+plot!(collect.(boundary), aspectratio=1, lw=0, label="boundary")
+```
+
+- The inner, blue, region is guaranteed to lie *inside* the constraint set.
+- The outer, white, region is guaranteed to lie *outside* the constraint set.
+- The in-between, red, region is not known at this tolerance.
+
+![Inner and outer ellipse](ellipses.svg?)
 
-#### ℹ️ Note on ModelingToolkit
 
-ModelingToolkit can be used to conveniently define constraints, as described in the tutorial. However, that package is only compatible up to Julia version 1.9, and it needs to be installed manually and restricted to version 3. Later versions of ModelingToolkit or Julia are not compatible yet.
 
 
 ## Author

README.md

@@ -0,0 +1,6 @@
+julia 1.0
+ModelingToolkit
+IntervalArithmetic 0.15
+IntervalRootFinding 0.4
+IntervalContractors 0.3
+MacroTools 0.4

REQUIRE

@@ -26,7 +26,7 @@ julia> using IntervalConstraintProgramming, IntervalArithmetic
 julia> S = @constraint x^2 + y^2 <= 1
 Separator:
   - variables: x, y
-  - expression: x ^ 2 + y ^ 2 ∈ [-∞, 1]
+  - expression: x ^ 2 + y ^ 2 ∈ [-Inf, 1]
 ```
 It works out automatically that `x` and `y` are variables.
 The macro creates a `Separator` object, in this case a `ConstraintSeparator`.
@@ -73,7 +73,7 @@ julia> @variables x y
 julia> S = Separator(x+y<1)
 Separator:
   - variables: x, y
-  - expression: x() + y() == [-∞, 1]
+  - expression: x() + y() == [-Inf, 1]
 
 julia> C = Contractor(x+y)
  Contractor in 2 dimensions:
@@ -91,7 +91,7 @@ julia> vars = @variables x y z
 julia> S = Separator(vars, x+y<1)
 Separator:
   - variables: x, y, z
-  - expression: x() + y() == [-∞, 1]
+  - expression: x() + y() == [-Inf, 1]
 
 julia> C = Contractor(vars, y+z)
 Contractor in 3 dimensions:
@@ -120,7 +120,7 @@ f (generic function with 1 method)
 julia> S=Separator(vars, f)
 Separator:
   - variables: x, y
-  - expression: x() + y() == [-∞, 1]  
+  - expression: x() + y() == [-Inf, 1]  
 
 julia> using DynamicPolynomials       #using polynomial functions
 
@@ -133,7 +133,7 @@ p (generic function with 1 method)
 julia> S=Separator(pvars, f)
 Separator:
   - variables: x, y
-  - expression: x() + y() == [-∞, 1]
+  - expression: x() + y() == [-Inf, 1]
 ```
 #### BasicContractor
 Objects of type `Contractor` have four fields (variables, forward, backward and expression), among them data of two fields (forward, backward) are useful (i.e forward and backward functions) for further usage of that object, thats why it is preferred to use an object of type `BasicContractor` in place of `Contractor` which only contain these two fields for less usage of memory by unloading all the extra stuff.(Note: Like object of `Contractor` type,`BasicContractor`'s object will also have all the properties which are discussed above).
@@ -196,7 +196,7 @@ There are sample 3D calculations in the `examples` directory, in particular in t
 
 
 ## Set operations
-Separators may be combined using the operators `!` (complement), `∩` and `∪` to make
+Separators may be combined using the operators `!` (complement), `⊓` and `⊔` to make
 more complicated sets; see the [notebook](https://github.com/JuliaIntervals/IntervalConstraintProgrammingNotebooks/blob/master/Basic%20examples%20of%20separators.ipynb) for several examples. Further examples can be found in the repository [IntervalConstraintProgrammingNotebooks](https://github.com/JuliaIntervals/IntervalConstraintProgrammingNotebooks).
 
 ## Author