The INFORMS Annual Meeting 2014 returns with a rich and varied program, bridging data and decisions. Each year, the INFORMS meeting brings together experts from academia, industry and government to consider a broad range of ORMS and analytics research and applications.
M2ACS Presentations at INFORMS Annual Meeting
Sven Leyffer: Exploiting Nonlinear Structure in Global Optimization
We present a new approach to generate tight and tractable convex relaxations of nonconvex functions based on exploiting group-partial separability. We demonstrate this approach in the context of two classes of powerful relaxation techniques: semi-definite relaxations and the reformulation-linearization technique. In both cases, we derive tight relaxations that can be solved orders of magnitude faster than traditional techniques that do not exploit the nonlinear structure.
Fu Lin: A Multi-Scale Optimization Approach to Co-Generation in Buildings
Jeffrey Linderoth: Cloud Computing for Optimization
Jim Luedtke: Valid Inequalities for the Pooling Problem
Jean-Paul Watson: Mixed-Integer Lower Bounds for Progressive Hedging
Nonconvexities and time limits may cause the progressive hedging (PH) algorithm to terminate in the mixed-integer case with a feasible solution, but no corresponding lower bound. We describe new methods to obtain lower bounds using both the information prices generated by PH and bundling methods. We additionally discuss experimental results and supporting open-source software implemenations.