LGO for AMPL
LGO is capable of determining high-quality solutions to global optimization problems that have (possibly many) locally optimal solutions. Because LGO makes only limited assumptions as to problem structure and differentiability, it can be applied in a broad range of design and operational applications.
Getting started with LGO
LGO downloads are available from the My Downloads page of your account at the AMPL Portal, and are included in the bundles that are used for free trials.
Summary
Developer: Pintér Consulting Services, Inc.
Current version: 2014-12-04
Problem types supported: Continuous nonlinear objectives and constraints without restriction as to smoothness or convexity.
Algorithms available: A suite of methods including heuristic global presolvers, partition and search (continuous branch-and-bound), global random search (single-start and multi-start), and constrained local (reduced gradient) optimization.
Special features: An extension to support integer-valued variables is in preparation.
Further Information
LGO for AMPL User’s Guide.
János D. Pintér’s home page at Rutgers Business School