There is a set J of h jobs to be processed. For every i, job J(i) dema
nds n(i) units of resources and returns a(i) units after completion. T
he problem is to find the best K schedules whose resource requirements
are minimum among all h! ones. In this paper, we present some importa
nt structural properties and then propose an O(h log h + hK log K) tim
e algorithm, which runs in polynomial time if K is fixed.