DISPLIB: Train Dispatching benchmark library
Given a set of trains traveling on a railway, the Train Dispatching Problem is the operational problem that occurs when some trains have become delayed with respect to their prescribed timetable, and we want to make routing and scheduling adjustments to minimize the total delay on the railway. Effective management of dense railway traffic using algorithms has proven to be very hard.
DISPLIB is a problem definition, a file format, and a set of problem instances with the aim of allowing researchers and engineers to better compare and benchmark their algorithms.
The first part of the development of the DISPLIB is the competition described below, which will be an occasion to collect problem instances from a variety of sources and validate the suitability of the problem format in practice. Later, we intend to organize problem instances on this web page, along with more details and classifications, with the intention of creating a standard benchmark suite for train dispatching algorithms.
DISPLIB 2025 Competition
The DISPLIB 2025 competition challenges you to advance the state-of-the-art in real-time train dispatching and contribute to more environmentally friendly transportation!
All the details are found in the files available below. This web page will be updated as the competition progresses, so please check back regularly if you are planning to participate.
No registration is needed — any team may submit new solutions to the organizers at any point before the deadline by sending an email to <displib2025competition@gmail.com>. We encourage teams not to send updates more frequently than once a week. The scoreboard as of the end of Phase 1 (2025-01-31) can be found below. During Phase 2, the scoreboard will not be updated and each team will be required to submit a short report describing their algorithm. Finalists will be selected by a committee of experts in the field, based both on the results of the scoreboard and the quality/novelty of the report. See the competition document below for detailed instructions.
Timeline
- 2024-09-10: Competition announced at ODS conference 2024 and online channels.
- 2024-10-01: First set of problem instances published (Phase 1).
- 2025-01-31: Submission deadline for the Phase 1 scoreboard (optional).
- 2025-02-03: Second set of problem instances published (Phase 2).
- 2025-04-30: Submission deadline for the report and solutions to all instances (Phase 1 and Phase 2).
Files
- DISPLIB 2025 competition and format (updated 2025-02-07)
- Solution verification program (Python) (version 0.3 published 2024-10-08)
- Test instances (published 2024-09-06)
-
Phase 1 instances
(v1.1 updated 2024-10-08)
- (v1.0 published 2024-10-01 had errors in the train references in the objective components of the line1_full instances)
- Phase 2 instances (v1.0 published 2025-02-03)
Phase 1 scoreboard
Seven teams submitted solutions within the Phase 1 deadline. The organizers are very grateful for the high quality of these submissions. Please note that participation to Phase 1 is not required and it is not too late to participate in the competition. The submissions for the final deadline should contain solution to both Phase 1 and Phase 2 instances, and scores will be computed independently of the Phase 1 submissions. This means that it is possible to achieve the highest score in the final scoreboard without having submitted anything for the early deadline. The teams who participated in Phase 1 may also submit improved solutions for Phase 1 instances to count towards the final scoreboard.
team | line1_critical | line1_full | line2_close | line2_headway | line3 | SUM |
---|---|---|---|---|---|---|
openbus
Florian Fuchs, Thomas Dubach, Jan Lordieck, Bernardo Martin-Iradi |
76 | 95 | 80 | 77 | 42 | 370 |
CSLS
Carolin Scholl, Luka Stärk |
76 | 76 | 62 | 61 | 42 | 317 |
The TRAINees
Lina Breuer, Sebastian Brunke, Elias Kaiser, Felix Michel |
43 | 31 | 36 | 64 | 14 | 188 |
WUB (Wien, Udine, Bologna)
Vera Hemmelmayr, Hai Yen Luu, Roberto Maria Rosati, Sara Ceschia, Andrea Schaerf, Alex Fabián Barrales Araneda, Valentina Cacchiani |
76 | 10 | 45 | 4 | 4 | 139 |
Delayed Train
Péter Dobrovoczki, Péter Györgyi, Markó Horváth, Tamás Kis | 8 | 50 | 20 | 30 | 18 | 126 |
TCS Research Railways team
Aakash Swami, Gajendra Malviya | 20 | 23 | 10 | 14 | 26 | 93 |
RailwayRam
Hariharan Subramanian, Sachin Jayaswal | 5 | 0 | 9 | 15 | 1 | 30 |
The A-Team
Andres Espinoza, Alaina Gordon, Aleksandr Kazachkov, Yuyang Sun | 0 | 0 | 0 | 0 | 0 | 0 |
Organization
Scientific committee
An international panel of experts of railway optimization will evaluate the submissions and select a group of finalists (see the competition document for details). The panel consists of:
- Giorgio Sartor, SINTEF Digital, Norway (Chair)
- Marcella Samà, Roma Tre University, Italy
- Paolo Ventura, Siemens Mobility, Italy
- Steven Harrod, Technical University of Denmark, Denmark
- Dennis Huisman, Erasmus University Rotterdam, Netherlands
Organizing committee
The competition is organized by Bjørnar Luteberget, Giorgio Sartor, Oddvar Kloster, and Carlo Mannino, and is funded by SINTEF . Get in touch with us by email if you find any errors or want to contribute new problem instances: <displib2025competition@gmail.com>