DistOpt: A Software Framework for Modeling and Evaluating Optimization Problem Solutions in Distributed Environments
✍ Scribed by Javier Contreras; Arturo Losi; Mario Russo; Felix F. Wu
- Publisher
- Elsevier Science
- Year
- 2000
- Tongue
- English
- Weight
- 447 KB
- Volume
- 60
- Category
- Article
- ISSN
- 0743-7315
No coin nor oath required. For personal study only.
✦ Synopsis
We present a flexible software environment, named DistOpt, which is useful in building coarse-grain algorithms for solving optimization problems and simulating the solution of the resulting subproblems in multicomputer systems. It is based on a decomposition-coordination approach, by which large optimization problems can be split into subproblems, which are then easier to solve and can be solved in parallel. The object-oriented methodology on which DistOpt is based, and its graphical user interface, are well suited to modify, interface, and extend existing software modules with minimal impact. DistOpt is a flexible and innovative software environment that can be customized by the user in a short development time.
2000 Academic Press
1. Introduction
Thanks to the relevant developments in communication technology, parallel distributed computing can be today efficiently developed on multicomputer systems, that is, on sets of ``state-of-the-art'' processors connected by small-scale high-speed localÂwide-area networks. Future developments may even lead to Internet-wide metacomputing [23]. Consequently, there has been a growing interest in coarsegrain computational models and algorithms, mainly because the algorithms for fine-grain models rarely translate to fast code on multicomputer systems [5].