低配额下的无嫉妒与放松稳定
Envy-freeness and Relaxed Stability under lower quotas
摘要:
我们考虑的问题下,双边偏好的最大以及最低配额的代理商的要求。这种设置,研究了低qualities(HRLQ)在文献中的医院居民,模型重要的现实世界的问题,如分配医疗实习生(居民)的医院,和教学助理的导师,其中最低保证是必不可少的。当没有最小配额时,稳定性是最优性的事实概念。然而,在最低配额存在的情况下,在许多情况下,既要确保稳定,又要满足较低的配额,并不是一个可以实现的目标。为了解决这个问题,文献中提出了一种称为无嫉妒的稳定性松弛。在我们的工作中,我们从计算的角度彻底研究了无嫉妒。我们的研究结果表明,计算匹配最大数量的代理的无嫉妒匹配是计算困难的,也很难近似到一个常数因子。此外,已知满足较低配额的无嫉妒匹配可能不存在。为了克服这些缺点,我们提出了一个新的概念,称为松弛稳定性。我们表明,放松稳定的匹配是保证存在的,即使在存在较低的配额。尽管找到一个最大的匹配,这是可行的和放松稳定的计算棘手,我们给出了有效的算法,计算一个常数因子近似这个匹配的大小。
英文摘要:
We consider the problem of matchings under two-sided preferences in the presence of maximum as well as minimum quota requirements for the agents. This setting, studied as the Hospital Residents with Lower Quotas (HRLQ) in literature, models important real world problems like assigning medical interns (residents) to hospitals, and teaching assistants to instructors where a minimum guarantee is essential. When there are no minimum quotas, stability is the de-facto notion of optimality. However, in the presence of minimum quotas, ensuring stability and simultaneously satisfying lower quotas is not an attainable goal in many instances. To address this, a relaxation of stability known as envy-freeness, is proposed in literature. In our work, we thoroughly investigate envy-freeness from a computational view point. Our results show that computing envy-free matchings that match maximum number of agents is computationally hard and also hard to approximate up to a constant factor. Additionally, it is known that envy-free matchings satisfying lower-quotas may not exist. To circumvent these drawbacks, we propose a new notion called relaxed stability. We show that relaxed stable matchings are guaranteed to exist even in the presence of lower-quotas. Despite the computational intractability of finding a largest matching that is feasible and relaxed stable, we give efficient algorithms that compute a constant factor approximation to this matching in terms of size.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。