This repository contains the instances and the code used in the tests of A.Saken et al., *Computational evaluation of cut-strengthening techniques in logic-based Benders' decomposition*
If you use the instances or code from this repository, please cite
For questions please contact elina.ronnberg@liu.se.
## Vehicle routing problem with location congestion
The instances used for the vehicle routing problem with location congestion was introduced in E. Lam, G. Gange, P. J. Stuckey, P. Van Hentenryck, and J. J. Dekker. *Nutmeg: A MIP and CP Hybrid Solver Using Branch-and-Check*. SN Operations Research Forum. These instances are available at https://github.com/ed-lam/nutmeg/tree/master/examples/vrplc/Instances.
The instances can also be found in
## Cumulative facility scheduling with fixed costs
The primary reference for instances used for the cumulative facility scheduling with fixed costs is J. N. Hooker. *Planning and Scheduling by Logic-Based Benders Decomposition*, Operations research 55:33, pp. 588-602. Information about the instances is available at http://public.tepper.cmu.edu/jnh/instances.htm and the instances themselves can be found at http://public.tepper.cmu.edu/jnh/planningandscheduling.zip
## Single-facility scheduling with a segmented timeline
For the single-facility scheduling problem with a segmented timeline the instance is generated according to description in E. Coban and J. N. Hooker. *Single-facility scheduling by logic-based Benders decomposition*, Annals of Operations Research 210, pp. 245-272, 2013, but with different parameter settings to make them more challenging.
The instance generator is available in [instances.py](instances.py) and we have used Python 3.8 to generate them. To generate all instances used in our article perform the following command.
```bash
>>># source virtual environment with python3.8
>>> sh segmented_instances.sh
```
The generated instances can be found in [sms_sdspmtw__instance_set_ii](sms_sdspmtw__instance_set_ii).
The detailed parameters used to create the instances are described in the following table
Snapshots (zip archives) of the code that was used to perform the test in the paper can also be found in this repository. For the code used to test the vehicle routing problem with location congestion, see [lbbd_vrplc-master.zip](lbbd_vrplc-master.zip), to test the cumulative facility scheduling problem with fixed costs, see [lbbd_cfs-master.zip](lbbd_cfs-master.zip), and to test the single machine scheduling problem with multiple time windows, see [lbbd_smsmtwsdsp-master.zip](lbbd_smsmtwsdsp-master.zip). Note that some of the instances used have been removed from the archive due to copyright restrictions and must be added into a *instances* folder for the code to run as expected.
Here is some information about the zip-files that could be useful.
- Every zip-file corresponds to a problem formulation in our paper, i.e, the implementation is done differently for each problem.
- The entry point for running an instance of any problem is always the main.py file and an example call (for each problem) can be found at the bottom of each main.py file.
- The code is tested/the results were run using the versions described in our paper, but it would probably work well with other versions of the tools/programming language as well.
## Paper
An PDF-version of the preprint can be found at [strengthening_of_feasibility_cuts_in_logic-based_benders_decomposition.pdf](strengthening_of_feasibility_cuts_in_logic-based_benders_decomposition.pdf).
## License
The content residing in this repository is licensed under "Creative Commons: CC BY 4.0 license", see [LICENSE](LICENSE).
## Acknowledgement
Computational experiments were performed on resources provided by the Swedish National Infrastructure for Computing (SNIC) at National Supercomputer Centre (NSC).
