Optimal-in-expectation redistribution mechanisms

Many important problems in multiagent systems involve the allocation of multiple resources among the agents. If agents are self-interested, they will lie about their valuations for the resources if they perceive this to be in their interest. The well-known VCG mechanism allocates the items efficiently, is strategy-proof (agents have no incentive to lie), and never runs a deficit. Nevertheless, the agents may have to make large payments to a party outside the system of agents, leading to decreased utility for the agents. Recent work has investigated the possibility of redistributing some of the payments back to the agents, without violating the other desirable properties of the VCG mechanism. Previous research on redistribution mechanisms has resulted in a worst-case optimal redistribution mechanism, that is, a mechanism that maximizes the fraction of VCG payments redistributed in the worst case. In contrast, in this paper, we assume that a prior distribution over the agents' valuations is available, and our goal is to maximize the expected total redistribution. In the first part of this paper, we study multi-unit auctions with unit demand. We analytically solve for a mechanism that is optimal among linear redistribution mechanisms. We also propose discretized redistribution mechanisms. We show how to automatically solve for the optimal discretized redistribution mechanism for a given discretization step size, and show that the resulting mechanisms converge to optimality as the step size goes to zero. We present experimental results showing that for auctions with many bidders, the optimal linear redistribution mechanism redistributes almost everything, whereas for auctions with few bidders, we can solve for the optimal discretized redistribution mechanism with a very small step size. In the second part of this paper, we study multi-unit auctions with nonincreasing marginal values. We extend the notion of linear redistribution mechanisms, previously defined only in the unit demand setting, to this more general setting. We introduce a linear program for finding the optimal linear redistribution mechanism. This linear program is unwieldy, so we also introduce one simplified linear program that produces relatively good linear redistribution mechanisms. We conjecture an analytical solution for the simplified linear program.