Multi-Agent Control
	Challenge: agents should choose local inputs that are globally optimal
				How?