The First Automatic Constraint Programming (CP) Optimizer

ILOG CP Optimizer is the newest constraint programming (CP) engine from ILOG. Its combination of a constraint propagation engine with an optimization engine lets you more easily solve problems that are too irregular for mathematical optimization.

ILOG CP Optimizer is also the CP engine embedded in the ILOG OPL-CPLEX Development System, allowing it to take full advantage of the development, debugging and tuning features of ILOG’s modeling environment. Our older CP engine, ILOG CP, remains available for complex scheduling and routing problems that are beyond the current capabilities of ILOG CP Optimizer.

Mathematical programming (MP) doesn’t always work
Important optimization problems, such as the sequencing of activities, resource allocation and timetabling are too irregular for MP.

The most common reasons optimization problems can't be solved with MP include:

• Constraints that are non-linear in nature, or that are linearized into an exponential number of linear constraints
• A non-convex solution space that contains many local optima
• Many disjunctions that result in a highly fractional relaxation with poor lower bounds

Although these problems can sometimes be modeled with an MIP solver, MP doesn’t always give a good result or a fast convergence to optimality.

The efficiency of constraint propagation
To solve these problems, optimization research (OR) specialists often develop complex neighborhood search algorithms that quickly explore the solution space. These efforts are often rewarded by faster and better solutions than those given by an MP model, but in practice, they can be difficult to build and maintain.

ILOG CP Optimizer takes advantage of the inherent efficiency of constraint propagation and automatically explores a solution space based on a mathematical formulation of the constraints. ILOG CP Optimizer automatically generates the search algorithms that would normally require weeks or months of difficult development.