Automatic programming methods applied to logistic problems

Costs associated with software development and maintenance are usually high, especially when the software is designed for parallel MIMD machines or requires complex algorithmic structures. Thus, there is a strong motivation from the economic sector in seeing the emergence of methods for the generation of standard (possibly not optimal) code in order to solve numerical or combinatorial problems. We first present a general description of the possible approaches. Then we focus our attention upon the case relating to the parallel asynchronous programming. Finally, we propose a simultaneous rooting/scheduling heuristic that allows a gain of up to 50% of the total duration time with regard to the MCP heuristic followed by rooting