Scheduling stochastic jobs with increasing hazard rate on identical parallel machines (Q1206392)

Several identical parallel machines are available to process a collection of jobs whose processing requirements are i.i.d. random variables with increasing hazard rate. A job completion at time \(t\) earns a reward \(r(t)\) where \(r\) is a non-increasing convex function such that \(r(t)\to 0\), \(t\to\infty\). New jobs may arrive in an arbitrary manner over time provided that the new jobs have received less prior processing than the job currently under execution. The SEPT policy which at every decision epoch chooses jobs for processing whose residual processing is smallest in expectation is shown to maximize the total expected reward earned in the class of preemptive policies.