Accession Number : AD1011273


Title :   New Approaches for Very Large-Scale Integer Programming


Descriptive Note : Technical Report


Corporate Author : Georgia Institute of Technology Atlanta United States


Personal Author(s) : Nemhauser,George


Full Text : https://apps.dtic.mil/dtic/tr/fulltext/u2/1011273.pdf


Report Date : 24 Jun 2016


Pagination or Media Count : 11


Abstract : The focus of this project is new computational tools for mixed-integer programming (MIP). During the course of this project we have studied and obtained results on the following topics. 1. Dual heuristics for integer programs in order to rapidly improve dual bounds. 2. Choosing good branching variables in branch-and-bound algorithms for MIP. 3. Machine Learning in solving MIPs.4. Parallel Processing in Solving MIPS. The new algorithms are computational tested and, in many cases, outperform existing algorithms. This research has been presented at several conferences and has and will appear in archival journals.


Descriptors :   Integer programming , algorithms , Parallel processing , Learning machines , Heuristic methods , machine learning


Distribution Statement : APPROVED FOR PUBLIC RELEASE