AI 中文总结
本文在弱优先序下比较SIC与EADA机制，证明严格偏序下两者结果等价，而在无环优先关系下分别出现改进不足与过度改进的相反失败形式。
AI 中文摘要
本文比较了Erdil和Ergin（2008）的稳定改进循环（SIC）机制与Kesten（2010）的效率调整递延接受（EADA）机制在不完全优先序的择校问题中的表现。当学校优先序为严格偏序时，我们证明EADA机制产生约束有效匹配。结合先前关于SIC机制的结果，这确立了两机制的结果等价性：它们可达到的匹配集合一致。随后，我们超越传递性，考虑无环优先关系。在此更广的域上，两种机制都不能保证产生约束有效匹配，但它们的失败形式相反。SIC机制可能改进不足：其结果总是稳定的，但可能被另一个稳定匹配帕累托支配。相比之下，EADA机制可能过度改进：其结果从不被任何稳定匹配帕累托支配，但本身可能不稳定。
英文摘要
This paper compares the stable improvement cycles (SIC) mechanism of Erdil and Ergin (2008) with the efficiency-adjusted deferred acceptance (EADA) mechanism of Kesten (2010) in school choice problems with incomplete priorities. When school priorities are strict partial orders, we show that the EADA mechanism produces a constrained efficient matching. Combined with earlier results for the SIC mechanism, this establishes the outcome equivalence of the two mechanisms: the sets of matchings attainable under them coincide. We then move beyond transitivity and consider acyclic priority relations. On this broader domain, neither mechanism is guaranteed to produce a constrained efficient matching, but their failures take opposite forms. The SIC mechanism may under-improve: its outcome is always stable but may be Pareto dominated by another stable matching. By contrast, the EADA mechanism may over-improve: its outcome is never Pareto dominated by any stable matching but may itself be unstable.
