On Randomized Online Scheduling - Review
On Randomized Online Scheduling deals with the most common problem found in the study of online algorithms: minimize the makespan of a sequence of tasks or jobs when scheduled in different parallel machines. The Rand algorithm proposed by Alberts performs better - 1.916 competitive ratio - than known deterministic solutions for general m (being m the number of machines available). The approach is a combination of two different scheduling alternatives which are chosen randomly with 1/2 probability to serve the entire sequence of tasks. In this paper I present a review of the Rand algorithm itself, analyzing the lower bound theory presented, technical approach, the algorithm, some proof and concepts and particular examples of the findings.
Comments