Product UpdatesCPLEX 11: Faster and smarter than everAlgorithmic advances and enhanced multicore support give the new ILOG CPLEX 11 more than twice the solving speed for mixed integer programming (MIP) problems of any previous version of the product. ILOG CPLEX’s new dynamic search algorithm takes branch-and-cut algorithms to the next level. With innovations in branching, node and cut strategies, the algorithm reaches optimal solutions faster and helps broaden ILOG CPLEX’s problem range. To work better on multicore computers, ILOG CPLEX now offers two modes for parallel optimization: deterministic and opportunistic. Users get repeatable solution paths with the first, while the second allows independence among threads that often results in even better performance. Giving users a choiceNot every aspect of a problem can be captured with a MIP model. ILOG CPLEX’s new solution pool feature assembles a selection of alternative solutions for the user to choose from. For example, a production planner can request five solutions within 2 percent of optimality, and then select the best one for balancing production lines based on information that could not be captured in the MIP model. 
ILOG CPLEX 11 also introduces a performance tuning utility for applications. It analyzes a model or a set of models to identify parameter settings that provide better performance than default settings. “We have broken new ground with ILOG CPLEX 11,” says Rosemary Berger, ILOG CPLEX product manager. “The algorithmic performance improvements alone are greater than ever – and combined with utilization of multicore computers, you have a vastly more powerful computational engine for planning and scheduling applications. For particularly difficult models, we’re talking about a speed up of an order of magnitude compared to CPLEX 10.” Learn more in our technical article ILOG CPLEX 11: Performance, Productivity, and Usability like nothing before and at http://cplex.ilog.com. |
|---|