A local search optimization algorithm based on natural principles of gravitation
Bernhard, Philip J.
MetadataShow full item record
This paper discusses the concept of an algorithm designed to locate the optimal solution to a problem in a (presumably) very large solution space. The algorithm attempts to locate the optimal solution to the problem by beginning a search at an arbitrary point in the solution space and then searching in the "local" area around the start point to find better solutions. The algorithm completes either when it locates what it thinks is the optimal solution or when predefined halt conditions have been met. The algorithm is repair-based, that is, it begins with a given solution and attempts to "repair" that solution by changing one or more of the components of the solution to bring the solution closer to the optimal. The algorithm uses the natural principles of gravity that act on a body in motion through space and simulates those principles to take a given solution to the problem at hand and repair it to locate an optimal solution. In this document two versions of the algorithm, called GLSA1 (based on simple gravitational force calculation) and GLSA2 (based on gravitational field calculation), are presented and the manner in which an initial evaluation of the algorithm was conducted. Then, by way of example a particular problem is given that the algorithm can be used to solve, along with a description of how the algorithm would be used to solve that problem. Finally, some conclusions and opportunities for future direction are presented.