Scheduling Search Procedures

Paper in proceeding, 2002

We analyze preemptive on-line scheduling against randomized adversaries, where the goal is only to finish an unknown distinguished target job. The problem is motivated by a clinical gene search project, but it leads to theoretical questions of broader interest, including some natural but unusual probabilistic models. For some versions of our problem we get optimal competitive ratios, expressed by given parameters of instances.