Abstract:We study the problem of fairly allocating a set of chores to a group of agents. The existence of envy-free up to any item (EFX) allocations is a long-standing open question for both goods and chores. We resolve this question by providing a negative answer for the latter, presenting a simple construction that admits no EFX solutions for allocating six items to three agents equipped with superadditive cost functions, thus proving a separation result between goods and bads. In fact, we uncover a deeper insight, showing that the instance has unbounded approximation ratio. Moreover, we show that deciding whether an EFX allocation exists is NP-complete. On the positive side, we establish the existence of EFX allocations under general monotone cost functions when the number of items is at most $n+2$. We then shift our attention to additive cost functions. We employ a general framework in order to improve the approximation guarantees in the well-studied case of three additive agents, and provide several conditional approximation bounds that leverage ordinal information.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is about the fairness in chore assignment, specifically the existence and approximation of "envy - freeness up to any item" (EFX) assignment. For chore assignment, whether EFX assignment exists is a long - unsolved problem. This paper proves by constructing counter - examples that in some cases, EFX assignment indeed does not exist, and explores its computational complexity and the possibility of approximate solutions.
### Specific Problem Description
1. **Existence of EFX Assignment**:
- For the assignment of chores, is there an assignment method such that each agent will not envy the assignment of other agents after removing any chore?
- The author proves by constructing a specific example that in the case of three agents and six chores, when using a super - additive cost function, EFX assignment does not exist.
2. **Computational Complexity**:
- Determining whether an instance has an EFX assignment is an NP - complete problem. This means that finding an EFX assignment is computationally very difficult.
3. **Approximate Solutions**:
- When EFX assignment does not exist, can a good approximate solution be found? The author studies the approximation guarantees under different conditions and provides several improved approximation ratios.
4. **Existence of EFX in Special Cases**:
- In the case of a small number of chores (for example, the number of chores is no more than the number of agents plus 2), does EFX assignment exist? The author proves that in this case, EFX assignment indeed exists.
### Main Contributions
- **Non - Existence and Non - Approximability**: By constructing counter - examples, it is proved that in some cases EFX assignment does not exist and there is no finite approximate solution.
- **NP - Completeness**: It is proved that determining whether an instance has an EFX assignment is an NP - complete problem.
- **Existence in the Case of a Small Number of Chores**: It is proved that when the number of chores is no more than the number of agents plus 2, EFX assignment always exists.
- **Improved Approximation Algorithms**: Several improved approximation algorithms are proposed, especially in the case of three agents with an additive utility function.
### Related Formulas
- **Super - Additive Cost Function**:
\[
c(S \cup T) \geq c(S)+c(T)
\]
- **EFX Definition**:
\[
c_i(X_i\setminus e) \leq c_i(X_j),\quad \forall i, j\in N,\forall e\in X_i
\]
- **α - EFX Definition**:
\[
c_i(X_i\setminus e) \leq \alpha\cdot c_i(X_j),\quad \forall i, j\in N,\forall e\in X_i
\]
Through these studies, the paper provides new insights into the fairness problem in chore assignment and points out the directions for future research.