AI 中文总结
针对智能体无法观察其他匹配对分配的分配博弈变体，提出自稳定集概念，构建稳定收益保证（SPG）集并证明其仅含有效分配，相关收益向量为含两类最优稳定收益的最小区间。
AI 中文摘要
我们考虑Shapley和Shubik（1971）提出的分配博弈的一个变体，其中智能体无法观察到其他匹配对的分配或剩余划分。我们提出了一个集值解概念（自稳定集）并刻画了最大的此类集合，这引出了稳定收益保证（SPG）集的构建，该集合假设智能体至少获得其收益保证且这是共同知识。我们的主要结果是，对于一般剩余矩阵，SPG集仅由有效分配组成，相关收益向量由包含工人最优和企业最优稳定收益的最小区间给出。
英文摘要
We consider a variant of the Assignment Game of Shapley and Shubik (1971), where agents do not observe the assignment or the surplus division of other matched pairs. We propose a set-valued solution concept (Self-Stabilizing Set) and characterize the largest such set. This leads to the formulation of the Stable Payoff Guarantee (SPG) set, which assumes that agents receive at least their payoff guarantees and that this is common knowledge. Our main result is that, for generic surplus matrices, the SPG set consists only of the efficient assignment. The associated payoff vectors are given by the smallest interval that contains the worker-optimal and firm-optimal stable payoffs.
