These are the presentations that were made by the IBM Decision Optimization team at the INFORMS Annual Meeting, Seattle, October 2019.
Andrea Tramontani, Recent Progress In CPLEX Benders Decomposition
In this talk we present the Benders decomposition branch-and-cut that is implemented in CPLEX for Mixed Integer Linear Programming (MILP). We illustrate the main algorithmic components behind our implementation and discuss the latest improvements that are currently work in progress. Finally, we present an extensive computational analysis on some classes of decomposable MILP problems, to assess the performance of Benders branch-and-cut in comparison with the default branch-and-cut of CPLEX. The results show that some models that are out of reach for a “standard” branch-and-cut can instead be solved by Benders decomposition.
2019-11 – INFORMS Seattle – Benders
Daniel Junglas, The Generic Callback in 2019
With version 12.10 CPLEX introduces the BRANCHING context into the generic callback framework. Using that context, users can now control the branches that are created at a search tree node that could not be fathomed.
In this presentation we review the generic callback framework, explain how it replaces the old CPLEX callbacks and what advantages is has over those old callbacks. We also explain some technical details that users should keep in mind in order to get the best out of the new framework.
2019-11 – INFORMS Seattle – Generic Callback 2019
Ed Klotz, Strategies for Solving Infeasible Mixed Integer Programs
Infeasible Mixed Integer Programs often pose a distinct set of challenges for branch and cut algorithms. The absence of a cutoff value established by an integer feasible solution essentially disables certain features of the branch and cut algorithm, making it less effective. However, for MIPs that are suspected to be infeasible, other tactics are particularly effective at proving infeasibility. We will examine some of these and also show how to prove infeasibility for two open MIPLIB 2017 models that, as of this writing, had an unknown feasibility status.
2019-11 – INFORMS Seattle – Solving Infeasible MIPs
Pierre Bonami, Applying a Classifier to Solve Mixed Integer Quadratic Problems in CPLEX
Within state-of-the-art optimization solvers such as IBM-CPLEX the ability to solve both convex and nonconvex Mixed-Integer Quadratic Programming (MIQP) problems to proven optimality goes back few years, but still presents unclear aspects. We are interested in understanding whether for solving an MIQP problem it is favorable to linearize its quadratic part or not. Our approach employs Machine Learning techniques to learn a classifier that predicts, for a given MIQP instance, the most suitable resolution method within IBM-CPLEX’s algorithmic framework.
2019-11 – INFORMS Seattle – Learning-MIQP-Classifier
Roland Wunderling, What’s new in CPLEX 12.10
We will present the performance improvements realized in the latest release of CPLEX and describe the ideas that helped achieve them.
2019-11 – INFORMS Seattle – Whats new in CPLEX 12.10
Ryan Kersh, Penalty Variables and Multiobjective Optimization
With the CPLEX 12.9 release it is possible to define multiple objectives for a single problem. In this talk we will introduce a simplified definition of a penalty variable, and present a method for taking advantage of penalty variables with a multiobjective formulation.
Finally, we will compare the performance of this method against default CPLEX.
2019-11 – INFORMS Seattle – Penalty Variables Multiobjective