Restricted Existence and Approximation Algorithms for PMMS

This paper studies the problem of dividing m indivisible items among n agents fairly and efficiently. Specifically, this research concentrates on pairwise maximin share (PMMS), which is defined to be the maximum value that an agent can guarantee for herself if she were to repartition the items with another agent and receive the bundle with the minimum value. PMMS is an important concept in the fair division. However, whether PMMS for indivisible items exists is still open. This work concentrates on PMMS by proving the existence of PMMS on linear graphs with binary valuation functions. Besides, this paper designs an algorithm to approximate PMMS in the case where different agents have identical valuations among the same items, achieving a ratio strictly greater than 0.8, which outperforms the state-of-the-art ratio of 0.781 from Kurokawa [22]. The time complexity of our FFD-based algorithm is O(mn + m log m).