There is a continuity of scheduling models describing shared supercomputing systems. Classic approaches model the problem as single-objective optimization; the objective corresponds to optimizing the performance of the system (the makespan, or the sum of jobs' completion times). Multi-objective methods assign an objective to each agent in the system---be it the end user, or an organization. Finally, game-theoretic methods model possible actions of the agents, be it a strategic change in the job parameters to try to get to the beginning of the queue, or a coalition of organizations deciding to isolate themselves from the rest of the system.

game-theoretic resource management and scheduling

multi-objective scheduling