We aim to make the source code of most of our algorithms available. Some of these are available from within our group's GitHub AlgTUDelft organization, others are shared through other repositories. An overview of the currently available toolboxes and the corresponding papers is provided below.

• SolvePOMDP - A toolbox containing exact and approximate algorithms for Partially Observable Markov Decision Processes. In particular, it contains the software implementation of the accelerated vector pruning algorithm described in the following paper: Erwin Walraven and Matthijs T. J. Spaan. Accelerated Vector Pruning for Optimal POMDP Solvers. Proceedings of the 31st AAAI Conference on Artificial Intelligence, 2017. For feature requests or any other inquiries, please contact Erwin Walraven.
• MADP - The MultiAgent Decision Process toolbox contains several algorithms and tools for planning and learning in multiagent systems. For more details we refer to the dedicated webpage. Contact: Matthijs Spaan.
• PowerTAC - A simulator for trading in wholesale electricity markets. Most of this work is done by John Collins and Wolf Ketter, but Mathijs de Weerdt has contributed to some of the design questions. See also the report describing this tool.
• STP and all-pairs shortest path algorithms and benchmark data - This page collects various benchmark problem sets to compute shortest paths on, for example to solve all-pairs short path problems such as useful for solving Simple Temporal Planning (STP) problems. In particular, it contains the code and data described in the following paper: L. R. Planken, M. M. de Weerdt and R. P.J. van der Krogt (2012) "Computing All-Pairs Shortest Paths by Leveraging Low Treewidth", Volume 43, pages 353-388.
• The data used for the simulations described in "Intention-Aware Routing of Electric Vehicles"
• Multi-Machine Scheduling Lower Bounds Using Decision Diagrams - This is an implementation of decision diagrams in order to obtain lower bounds for a multi-machine scheduling problem, as described in van den Bogaerdt, P., de Weerdt, M.M.: Multi-machine scheduling lower bounds using decision diagrams. Operations Research Letters 46(6), 616–621 (2018).
  • Implementation
  • Datasets
  • Accepted manuscript
  • (Preprint)
• Lower Bounds for Uniform Machine Scheduling Using Decision Diagrams - This is an implementation of decision diagrams for lower bounds on uniform multi-machine scheduling problem instances, based on the implementation  above. This is described in van den Bogaerdt, de Weerdt (2019), Lower Bounds for Uniform Machine Scheduling Using Decision Diagrams, In Louis-Martin Rousseau, Kostas Stergiou (Eds.), Integration of Constraint Programming, Artificial Intelligence, and Operations Research, pp. 565-580.
  • Implementation
  • Datasets
  • Accepted manuscript
• ConstrainedPlanningToolbox - The toolbox provides a collection of models, algorithms and tools that can be used for multi-agent planning under uncertainty in environments with resource constraints. The toolbox relies on Markov Decision Processes and Partially Observable Markov Decision Processes to model planning problems, which are augmented with additional constraints on resource consumption. This model can be used for a variety of constrained planning problems. Examples include robot navigation with limited battery capacity and planning for electric vehicle charging in congested distribution grids. This is described in: Frits de Nijs, Erwin Walraven, Mathijs M. de Weerdt, and Matthijs T. J. Spaan. "Constrained Multiagent Markov Decision Processes: a Taxonomy of Problems and Algorithms". Journal of Artificial Intelligence Research 70 (2021): 955-1001.
• Order acceptance and scheduling with sequence-dependent setup times - Implementations of different algorithms and benchmark instances for a single machine scheduling problem with rejection can be found here. The respective paper can be found here. More up to date source code with an improved algorithm can be found here. Updated benchmark instances can be found here. If you use these, please cite our work: He L , De Weerdt M , Yorke-Smith N. Time/sequence-dependent scheduling: the design and evaluation of a general purpose tabu-based adaptive large neighbourhood search algorithm. Journal of Intelligent Manufacturing, 2019
• B-FELSA - A framework for benchmarking flexible electric load scheduling algorithms. The toolbox includes a number of optimization algorithms and a scenario generator. See here for a bit more explanation. The toolbox is described in Van der Linden K, Romero N, de Weerdt MM. Benchmarking Flexible Electric Loads Scheduling Algorithms. Energies. 2021; 14(5):1269. https://doi.org/10.3390/en14051269.
• TORS - A simulator for train unit shunting and servicing. A first publication on this appears in the demo track of AAMAS 2021.
• Deep reinforcement learning for wake control of a wind farm - published at AAMAS 2022.​​​​​​​
• GridPenguin -  A district heating network simulator - published at the New Energy for Industry Conference 2022​​​​​​​