Formally, the challenge has several elements the details of which may vary by VRP variant:
- An effort geared toward collecting and improving sets of benchmark instances.
- An algorithm evaluation effort in which participants develop and evaluate their algorithms and implementations on the test instances provided. The detailed specifications for algorithm evaluation are tailored to each variant and will be provided on the associated webpage for each variant. Algorithm evaluation will consist of one or two phases, depending on the variant. Key dates are listed in this timeline. (There may be additional deadlines associated with individual variants.)
- A workshop in which participants present their work and where competition results are announced.
- Special sections or issues of refereed journals dedicated to new research presented at the workshop. Papers submitted for publication will go through the journal's normal review process.
This challenge is open to anyone who wishes to participate. Participation can take a variety of forms, and may include any or all of:
- Contribute new instances. The instances submitted should be natural and interesting. Instances stemming from real-world applications are especially welcome, as are synthetic instances that are particularly challenging for one or more algorithms.
- Evaluate algorithm(s). Participants in the algorithm evaluation will submit timing results and ultimately a paper describing the implementation and its performance on designated test instances. Given the Challenge's overarching goal of sharing information, participants should be willing to share details about their algorithm and its implementation with other participants. ("Black box" implementations are strongly discouraged.) In general, algorithm performance is multi-faceted, and there are several dimensions for improving the state-of-the-art. We hope to encourage progress along dimensions that include: 1) speed of solution; 2) quality of solution; 3) quality of solution within a given time; 4) versatility across variants; and 5) novelty of approach. Evaluation criteria will be posted for each variant (on its associated webpage) as the competition proceeds. To enter an algorithm for evaluation, the participant/team must be willing to submit & present a paper describing their solution approach and results at the workshop. (Depending on the status of the pandemic, there may be an option to present remotely.)
Registration for algorithm evaluation is now closed.Registered Competitors: View registered teams.
- Participate in the workshop. The workshop is open to all to attend, regardless of participation in algorithm evaluation. You will need to register separately to attend the workshop. Registration for the workshop will open in early 2022 and be accessible from the workshop webpage.
- Present in the workshop: To present in the workshop, you must participate in the algorithm evaluation portion of the Challenge for at least one variant. You must also submit a short paper (i.e. extended abstract) describing your algorithm. The organizing committee will invite speakers based on their review of your paper and the performance of your algorithm. Several of the top performers for each variant will be invited to present, provided that their algorithms are adequately described in the associated paper.
- Submit a journal paper. You may submit a paper to include in a journal special issue/section devoted to the Implementation Challenge. Selection will be based on the journal's normal review process.
We aim for participation to be inclusive and flexible. In particular:
- Anyone may contribute instances without presenting a related paper at (or even attending) the workshop, although attendance is encouraged.
Participants may report results for as many algorithms (and variants) as they wish.
- Approaches may be heuristic or exact.
- Participants are welcome to submit papers heavily based on work they have previously published elsewhere, as long as they are clearly marked as such. Such submissions are eligible for presentation at the workshop and for participation in the competition. The journal special issue, however, is restricted to papers with substantially new content and subject to the standard journal reviewing process (with its editors).
We have set up a mailing list to facilitate communication among participants and organizers.