AI 中文总结
该研究提出用于不可分割对象分配的罗尔斯式公平概念，证明其分配集合与辅助市场稳定分配集合一致，主体提议的延迟接受算法可计算其主体最优元素，但无策略防卫的罗尔斯式公平规则。
AI 中文摘要
我们提出罗尔斯式公平（Rawlsian equity），这是一种在无对象特定授权的主体间分配不可分割对象的公平概念。罗尔斯式公平要求：除非另一主体获得了更偏好的对象，否则不得将某一对象分配给对其评价更高的主体。我们证明，罗尔斯式公平分配的集合与辅助市场中的稳定分配集合一致，该市场中对象的优先级取决于主体的偏好报告。基于主体提议的延迟接受算法可计算该集合中主体最优的元素，但总体而言，不存在策略防卫的罗尔斯式公平规则。
英文摘要
We introduce Rawlsian equity, a notion of fairness for allocating indivisible objects among agents without object-specific entitlements. Rawlsian equity requires that an object not be assigned to an agent who ranks it more highly than another agent unless the latter receives a more preferred object. We show that the set of Rawlsian equitable allocations coincides with the set of stable allocations in an auxiliary market where object priorities depend on agents' preference reports. The agent-proposing deferred acceptance algorithm computes the agent-optimal element of this set, but no Rawlsian equitable rule is strategy-proof in general.
