This page is for an old version of Hexaly Optimizer. We recommend that you update your version and read the documentation for the latest stable release.

Hexaly Optimizer

Object-oriented APIs are provided for Python, allowing a full integration of Hexaly Optimizer in your Python business applications. Hexaly’s APIs are lightweight, with only a few classes to manipulate. Note that Hexaly Optimizer is a model & run math programming solver: having instantiated the model, no additional code has to be written in order to run the solver.

Build your model

First, you have to create a LocalSolver environment. It is the main class of the Hexaly Optimizer library. Then, use the methods of the class LSModel to build your model with expressions. Expressions are a particularly important concept in Hexaly Optimizer. In fact, every aspect of a model is an expression: variables, objectives and even constraints are LSExpression. There are 3 different ways to create these LSExpressions with the class LSModel:

  1. You can use the available shortcut methods like LSModel.sum(), LSModel.eq(), LSModel.bool() or LSModel.sqrt().
  2. You can also use the more generic version of these operators with the method LSModel.create_expression(). It takes the type of the expression to add as first argument, then the list of the operands of the expression. It is also possible to add operands one-by-one with the method LSExpression.add_operand(). See LSOperator for the complete list of available operators.
  3. Finally, you can use the overloaded operators for common operations: +, -, *, /, <=, >=, ==, !=, >, <, **, [], &, ^, ~, |, ().

Most of these methods accept LSExpressions as arguments but also integers or double constants. If you prefer, you can also create constants explicitly with LSModel.create_constant().

Solve your model

Once you have created your model, you have to close it with LSModel.close() and call LocalSolver.solve() to launch the resolution. By default, the search will continue until an optimal solution is found. To set a time limit or an iteration limit, create a LSPhase, with create_phase(), then set the according attributes.

Retrieve the solution

The solution is available through the attribute LocalSolver.solution and carries the values of all expressions in the model and the status of the solution. There are 4 different statuses:

  • INCONSISTENT: The solver was able to prove that the model admits no feasible solution.
  • INFEASIBLE: The solution is infeasible. Some constraints or expressions are violated.
  • FEASIBLE: The solution is feasible but the optimality was not proven.
  • OPTIMAL: The solution is optimal. All objective bounds are reached.

You can also directly use the value attribute available on LSExpression to get the value of the expression in the solution.

Consult statistics

You can retrieve statistics of the search (number of iterations, % of feasible moves, etc.) with the LSStatistics object. Statistics are provided for the global search or for each phase.

Error handling

All classes and methods of the LocalSolver API can throw exceptions. The exception type related to LocalSolver errors is LSError.