Envy-free division of multi-layered cakes

TL;DR

This paper addresses envy-free division of multi-layered cakes, providing existence results under certain conditions and designing an FPTAS for approximate solutions with monotone preferences.

Contribution

It establishes the existence of envy-free, non-overlapping, and contiguous multi-layered cake divisions for prime power number of agents and layers, and develops an FPTAS for approximate solutions.

Findings

Existence of envy-free divisions for prime power agents and layers.

Development of an FPTAS for two-layered cake division among three agents.

Extension of results to groups of nearly equal size and arbitrary group sizes with monotone preferences.

Abstract

We study the problem of dividing a multi-layered cake under non-overlapping constraints. This problem, recently proposed by Hosseini et al. (IJCAI, 2020), captures several natural scenarios such as the allocation of multiple facilities over time where each agent can utilize at most one facility simultaneously, and the allocation of tasks over time where each agent can perform at most one task simultaneously. We establish the existence of an envy-free multi-division that is both non-overlapping and contiguous within each layer when the number of agents is a prime power and the number of layers is at most the number of agents, providing a positive partial answer to an open question by Hosseini et al. To achieve this, we employ a new approach based on a general fixed point theorem, originally proven by Volovikov (1996), and recently applied by Joji\'{c}, Panina, and \v{Z}ivaljevi\'{c} (2020) to the envy-free division problem of a cake. We further design a fully polynomial-time approximation scheme (FPTAS) to find an approximate envy-free solution that is both non-overlapping and contiguous for a two-layered cake division among three agents with monotone preferences. More generally, we establish all the results for divisions among groups of almost the same size. When there are three groups and one layer, our FPTAS is actually able to deal with groups of any predetermined sizes, still under the monotone preference assumptions. For three groups, this provides thus an algorithmic version of a recent theorem by Segal-Halevi and Suksompong (2021). A classical lemma from topology ensures that every self-map of a simplex mapping each facet to itself is surjective. This lemma has often been used in economics theory and our algorithm can be interpreted as an FPTAS counterpart of its two-dimensional version.