Maximum-Size Envy-free Matchings

10/16/2019 ∙ by Prem Krishnaa, et al. ∙ 0

We consider the problem of assigning residents to hospitals when hospitals have upper and lower quotas. Apart from this, both residents and hospitals have a preference list which is a strict ordering on a subset of the other side. Stability is a well-known notion of optimality in this setting. Every Hospital-Residents (HR) instance without lower quotas admits at least one stable matching. When hospitals have lower quotas (HRLQ), there exist instances for which no matching that is simultaneously stable and feasible exists. We investigate envy-freeness which is a relaxation of stability for such instances. Yokoi (ISAAC 2017) gave a characterization for HRLQ instances that admit a feasible and envy-free matching. Yokoi's algorithm gives a minimum size feasible envy-free matching, if there exists one. We investigate the complexity of computing a maximum size envy-free matching in an HRLQ instance (MAXEFM problem) which is equivalent to computing an envy-free matching with minimum number of unmatched residents (MIN-UR-EFM problem). We show that both the MAXEFM and MIN-UR-EFM problems for an HRLQ instance with arbitrary incomplete preference lists are NP-hard. We show that MAXEFM cannot be approximated within a factor of 21/19. On the other hand we show that the MIN-UR-EFM problem cannot be approximated for any alpha > 0. We present 1/(L+1) approximation algorithm for MAXEFM when quotas are at most 1 where L is the length of longest preference list of a resident. We also show that both the problems become tractable with additional restrictions on preference lists and quotas. We also investigate the parameterized complexity of these problems and prove that they are W[1]-hard when deficiency is the parameter. On the positive side, we show that the problems are fixed parameter tractable for several interesting parameters.



There are no comments yet.


page 1

page 2

page 3

page 4

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.