Hexaly.jl

Warning

This package is still a work in progress in early stage of development.

Hexaly.jl is a wrapper for the Hexaly Optimizer.

It provides two layers of access:

1. A thin wrapper over the Hexaly Python API (via PythonCall.jl), exposed as Hexaly.raw_optimizer().
2. A MathOptInterface (MOI) wrapper exposed as Hexaly.Optimizer, which makes Hexaly usable from JuMP.

Installation

import Pkg
Pkg.add(url = "https://github.com/NexOR-Optimization/Hexaly.jl")

Hexaly requires a license. See Hexaly's documentation for instructions.

Affiliation

This wrapper is maintained by NexOR Optimization and is not officially supported by Hexaly.

Use with JuMP

To use Hexaly with JuMP, use Hexaly.Optimizer:

using JuMP, Hexaly

model = Model(Hexaly.Optimizer)
set_attribute(model, "time_limit", 10)  # in seconds
set_silent(model)

@variable(model, 0 <= x <= 3, Int)
@variable(model, 0 <= y <= 3, Int)
@constraint(model, x + y <= 4)
@objective(model, Max, 3x + 2y)
optimize!(model)

@show value(x), value(y), objective_value(model)

Raw Python API

using Hexaly
using Hexaly.PythonCall

optimizer = Hexaly.raw_optimizer()
m = optimizer.model
x = m.int(0, 10)
m.constraint(x >= 3)
m.minimize(x)
m.close()

optimizer.param.time_limit = 5
optimizer.solve()
@show pyconvert(Int, x.value)

Supported MOI features

• Variables:
  • Unconstrained (float)
  • MOI.Integer, MOI.ZeroOne
  • MOI.EqualTo, MOI.LessThan, MOI.GreaterThan, MOI.Interval (integer or float)
• Constraints:
  • MOI.VariableIndex in bound sets (integer or float)
  • MOI.ScalarAffineFunction in MOI.{EqualTo, LessThan, GreaterThan}
  • MOI.VectorOfVariables in MOI.AllDifferent
  • MOI.VectorOfVariables in MOI.Circuit
  • MOI.VectorOfVariables in MOI.BinPacking
• Objectives:
  • MOI.VariableIndex
  • MOI.ScalarAffineFunction
• Options:
  • MOI.Silent
  • MOI.TimeLimitSec
  • MOI.RawOptimizerAttribute("<hexaly-param>")