| Cooperative and hybrid optimization |
|
With ILOG
Concert Technology as the foundation, ILOG CPLEX can be used in cooperation
with ILOG Solver to create
hybrid and cooperative applications that combine the power of constraint programming
and mathematical programming to solve problems that would be unsolvable
using just a single methodology. The common representation paradigm provided
with ILOG Concert Technology (within ILOG Solver and ILOG CPLEX) makes it easy
to create these kinds of applications. Examples of these applications include:
Airline crew scheduling
Typically, airline crew scheduling problems are solved using custom
code to generate pairings that represent tours of duties for crews. An
alternative methodology is to use constraint programming with ILOG Solver
to represent the complex constraints that tours of duties must satisfy,
and to let the constraint programming framework enumerate feasible pairings.
Each pairing corresponds to a column of a set partitioning problem that
is then solved with ILOG CPLEX. Using ILOG Concert Technology as the foundation,
this type of application becomes much easier to develop.
Credit card manufacturing
A company needs to sequence the order of credit cards to be manufactured
through a machine. When the color of the card changes, the printing cartridge
must be changed as well. If constraint programming is used alone to solve this
problem, there will be too many cards to sequence. The problem can be solved by using mathematical programming to determine the required number of cartridge changes, followed by constraint
programming to sequence the cards between changes in the cartridges. ILOG Concert Technology's common representation
layer makes this algorithmic solution easier to develop.
Complete beverage supply chain
A company wanted to model their entire beverage supply chain: deciding
when and where to buy raw materials, blending them to produce different beverage
brands, and bottling them, followed by distributing cases to warehouses and
then to stores. The entire problem was too difficult to solve using
a large mixed integer programming model. The entire
problem became tractable by using ILOG Solver to solve
the distribution problem, and then using these solutions as input for
the production-planning problem solved with ILOG CPLEX. ILOG Concert Technology makes these kinds of problem
decompositions easier to implement.
Scheduling
Jain
and Grossmann have demonstrated the effectiveness
of combining constraint programming and mathematical programming
on a class of scheduling problems that processes a set
of orders on a set of dissimilar parallel machines. For
this research, ILOG
OPL Studio was used. The same application is also
easy to implement using ILOG Concert Technology.
Cooperative solvers allow the ILOG CPLEX simplex optimizers
to be treated as a global constraint by ILOG Solver. Using ILOG Concert Technology,
a problem can contain a mix of logical and linear constraints. A search procedure
developed with ILOG Solver can treat the set of linear constraints as a single
constraint. The domain reduction and constraint propagation procedures cause
communication to automatically occur between the different kinds of constraints.
|