OASIcs.ATMOS.2024.8.pdf
- Filesize: 0.86 MB
- 6 pages
We revisit the Segmented Best Path (sbp) algorithm for online DARP in an offline setting with revenues and a time limit. The goal is to find a subset of the inputted ride requests that can be served within the time limit while maximizing the total revenue earned. sbp divides the time into segments and greedily chooses the highest-revenue path of requests to serve within each time segment. We show that sbp’s performance has an upper bound of 5. Further, while sbp is a tight 4-approximation in the uniform-revenue case, we find that with non-uniform revenues, the approximation ratio of sbp has a lower bound strictly greater than 4; in particular, we provide a lower bound of (√e + 1)/(√e - 1) ≈ 4.08299, which we show can be generalized to instances with ratio greater than 4.278.
Feedback for Dagstuhl Publishing