Abstract:We study the problem of fairly allocating $m$ indivisible goods to $n$ agents, where agents may have different preferences over the goods. In the traditional setting, agents' valuations are provided as inputs to the algorithm. In this paper, we study a new comparison-based query model where the algorithm presents two bundles of goods to an agent and the agent responds by telling the algorithm which bundle she prefers. We investigate the query complexity for computing allocations with several fairness notions including proportionality up to one good (PROP1), envy-freeness up to one good (EF1), and maximin share (MMS). Our main result is an algorithm that computes an allocation satisfying both PROP1 and $\frac12$-MMS within $O(\log m)$ queries with a constant number of $n$ agents. For identical and additive valuation, we present an algorithm for computing an EF1 allocation within $O(\log m)$ queries with a constant number of $n$ agents. To complement the positive results, we show that the lower bound of the query complexity for any of the three fairness notions is $\Omega(\log m)$ even with two agents.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is: when fairly allocating indivisible items, how to effectively find an allocation scheme that meets certain fairness criteria through the comparison - based query model. Specifically, the research focuses on optimizing the query complexity to calculate the allocations that meet the following fairness concepts:
1. **Proportionality up to one good (PROP1)**: After each agent obtains an additional item, the value of its allocation is at least its average share of the total value of all items.
2. **Envy - freeness up to one good (EF1)**: For any two agents \(i\) and \(j\), if one item is removed from \(j\)'s allocation, agent \(i\) will no longer envy \(j\).
3. **Maximin share (MMS)**: The value of the items obtained by each agent is at least the maximin share value.
### Research Background and Motivation
In traditional fair allocation problems, it is usually assumed that the valuation function of each agent is known and serves as the input of the algorithm. However, in many practical scenarios, it may be difficult or impractical to directly obtain these valuation functions. Therefore, the author proposes a new comparison - based query model, in which the algorithm obtains information only by asking agents which bundle of items they prefer. This model not only simplifies the information - gathering process but also protects the agents' privacy.
### Main Contributions
1. **Algorithm Design**:
- An algorithm is proposed that can find an allocation scheme that simultaneously satisfies PROP1 and \(\frac{1}{2}\)-MMS for a constant number of agents within \(O(\log m)\) query times.
- For the same and additive valuation functions, another algorithm is proposed to find an allocation scheme that satisfies EF1 within \(O(\log m)\) query times.
2. **Theoretical Results**:
- It is proved that for any of the above three fairness criteria, even if there are only two agents, the lower bound of the query complexity is \(\Omega(\log m)\).
3. **Technical Means**:
- Three key techniques are used: the IterativeMinimumBundleKeeping sub - program, the PROP1 - PROP sub - program, and the Hall matching method. These techniques work together to ensure the effectiveness and efficiency of the algorithm.
### Definitions of Fairness Concepts
To better understand the research content of the paper, the following are the definitions of several key fairness concepts:
- **Envy - Free (EF)**: If for any two agents \(i\) and \(j\), \(u_i(A_i)\geq u_i(A_j)\), then the allocation \(A\) is envy - free.
- **Proportional (PROP)**: If for each agent \(i\), \(u_i(A_i)\geq\frac{1}{n}u_i(M)\), then the allocation \(A\) is proportional.
- **Envy - freeness up to one good (EF1)**: If for any two agents \(i\) and \(j\), either \(u_i(A_i)\geq u_i(A_j)\) or there exists \(g\in A_j\) such that \(u_i(A_i)\geq u_i(A_j\setminus\{g\})\), then the allocation \(A\) is EF1.
- **Proportionality up to one good (PROP1)**: If for each agent \(i\), there exists \(g\in M\setminus A_i\) such that \(u_i(A_i\cup\{g\})\geq\frac{1}{n}u_i(M)\), then the allocation \(A\) is PROP1.
- **Maximin share (MMS)**