Using Distributed Computing to Improve the Efficiency of a Genetic Algorithm
J. Andrew Holey, Computer Science
This project will show that the traveling salesperson problem (TSP) will become a much more manageable problem when these two techniques are used: 1) A genetic algorithm (GA) is used to greatly speed up finding usable solutions and 2) The algorithm will be distributed to several computers to further improve the efficiency of the algorithm. Although GA's have been shown numerous times to be an extremely effective manner in finding possible solutions to NP-complete problems, there are some limitations. The major drawback to GA is the inability to guarantee that the optimal solution will be found. GA's employ a great deal of randomness which makes it impossible to say exactly how quickly the best solution will be found or even if it will be found. It is inaccurate to say that a GA solves because finding the best solution is never guaranteed. A solution in a GA is simply one possible way of solving the problem, not the best solution. Keeping in mind the limitations of GA's, the interesting part of this project is to show how distributing the algorithm can really speed up the process.
Athman, Joseph, "Using Distributed Computing to Improve the Efficiency of a Genetic Algorithm" (2004). Honors Theses, 1963-2015. 420.