Description
This software aids in the selection of good parameters for optimization solvers. The user selects a solver interface and provides a set of representative instances along with a maximum number of trials. Given this limited number of trials, the software designs a series of tests to identify a good set of parameters. The software uses the ideas of parameter interactions and pairwise coverage as well as machine learning through regression trees or artificial neural networks. The software includes interfaces for several commonly used mixed-integer linear programming solvers. Interfaces to additional solvers can be created by the user.IP Status
Software
