Abstract
Suppose that two players, P1 and P2, must divide a set of indivisible items that each strictly ranks from best to worst. Assuming that the number of items is even, suppose also that the players desire that the allocations be balanced (each player gets half the items), item-wise envy-free (EF), and Pareto-optimal (PO). Meeting this ideal is frequently impossible. If so, we find a balanced maximal partial allocation of items to the players that is EF, though it may not be PO. Then, we show how to augment it so that it becomes a complete allocation (all items are allocated) that is EF for one player (Pi) and almost-EF for the other player (Pj) in the sense that it is EF for Pj except for one item — it would be EF for Pj if a specific item assigned to Pi were removed. Moreover, we show how low-ranked (for Pj) that exceptional item may be, thereby finding an almost-EF allocation that is as close as possible to EF — as well as complete, balanced, and PO. We provide algorithms to find such almost-EF allocations, adapted from algorithms that apply when complete balanced EF-PO allocations are possible.
Original language | English (US) |
---|---|
Article number | 24 |
Journal | Operations Research Forum |
Volume | 3 |
Issue number | 2 |
DOIs | |
State | Published - Jun 2022 |
Keywords
- 2-Person fair division
- Envy-freeness up to one item
- Indivisible items
- Pareto-optimal
ASJC Scopus subject areas
- Economics, Econometrics and Finance (miscellaneous)
- Computer Science Applications
- Control and Optimization
- Applied Mathematics