Nov. 21, 2017, 10:15 a.m.

Title: "Scheduling jobs to minimize energy, electricity, and other resources
Speaker: Prudence Wong (University of Liverpool)
Time and place: Tuesday, 21st of November, 2017, 10:15 am, room 119


Energy is a big concern these days both in terms of computation and household usage. There are many different scheduling problems that attempt to address this issue. In this talk we will discuss a few scheduling problems including processor scheduling by using dynamic speed scaling or processor sharing. We also consider a simple model of a household scenario where we want to minimize the electricity cost. In these problems, we are given a set of jobs arriving with an arrival time, a deadline, a processing size (which may be given as a quantity or as a rectangle of fixed shape). Depending on the problem, we may process the job by stretching/shrinking a job as long as its "area" stays the same or we may have to process the job as a rigid rectangle. The cost of processing the jobs at any time is a function of the load at that time and the function is typically convex, e.g., (load(t))^alpha for alpha > 1. We will discuss online algorithms for different variants of the problem. We will also consider some variant in which other resources are the objective function.