Truthful Facility Location with Candidate Locations and Limited Resources
Panagiotis Kanellopoulos, Alexandros A. Voudouris
TL;DR
The paper addresses truthful facility location with $k$ facilities built at candidate locations within $[0,1]$, where agents have private positions and approval preferences. It establishes an impossibility for deterministic strategyproof mechanisms to bound the approximation ratio, and introduces a randomized mechanism that achieves a tight ratio of $k$, with a case-based design dependent on the leftmost/rightmost candidates. In the restricted known-positions setting, it presents a deterministic mechanism with an approximation around $2.325$ (improved to $2$ for $k=2$) and proves lower bounds of $3/2$ (deterministic) and $6/5$ (randomized). These results delineate the fundamental limits of strategyproof facility location under discrete candidate locations and offer concrete mechanisms with provable guarantees, while suggesting avenues for multiwinner and participatory budgeting extensions.
Abstract
We study a truthful facility location problem where one out of $k\geq2$ available facilities must be built at a location chosen from a set of candidate ones in the interval $[0,1]$. This decision aims to accommodate a set of agents with private positions in $[0,1]$ and approval preferences over the facilities; the agents act strategically and may misreport their private information to maximize their utility, which depends on the chosen facility and their distance from it. We focus on strategyproof mechanisms that incentivize the agents to act truthfully and bound the best possible approximation of the optimal social welfare (the total utility of the agents) they can achieve. We first show that deterministic mechanisms have unbounded approximation ratio, and then present a randomized mechanism with approximation ratio $k$, which is tight even when agents may only misreport their positions. For the restricted setting where agents may only misreport their approval preferences, we design a deterministic mechanism with approximation ratio of roughly $2.325$, and establish lower bounds of $3/2$ and $6/5$ for deterministic and randomized mechanisms, respectively.