This is the artifact for the FM2024 publication "Stochastic Games for User Journeys" by Kobialka, Pferscher, Bergersen, Johnsen, and Tapia Tarifa. We provide the artifact as a Docker container that includes all required data, tools and source code.
Docker is a virtualization software, distributed here. Please follow the installation instructions for your machine.
Ensure that docker is running by executing docker --version in the terminal.
To load the docker container, execute the following command:
docker load --input fm-artefact.tar
You can check with docker images if the docker image is loaded.
Docker-Desktop limits the available RAM automatically to 8GB, which is not sufficient. Please increase the limit to the maximum, at least 25 GB. In our experiments, MacOS based systems seemed to require more memory than Linux based systems.
To run the container replace path/to/your/output/repository by any directory on your local machine to which the generated plots, figures and tables should be exported.
docker run -v 'path/to/your/output/repository:/home/out' --platform linux/amd64 fm-artefact:latest
To run the fast execution, taking 90 min, append the short execution flag to your call.
--short_execution
You might need root rights to run the docker image.
The artefact contains all necessary data and scripts to reproduce the findings in our paper. Specifically, the artefact reproduces:
- Table 2
- Figures 3, 4, 5, 6 and 7.
Under the short execution, fig4c.png and bpic_bounded.png differ from the presented ones due to increased step-sizes and reduced model-sizes.
Additional plots shown on the github page are also included.
After running the image, the linked out folder contains an new bpic17 and greps folder.
The png's for Figures 3, 4 and 5 are contained in out/greps, for Figures 6 and 7 in out/bpic17, and are called fig3.png, fig4a.png, fig4b.png etc.
A markdown-file for Table 2 is generated in out/table2.md.
Important: Please note that the produced png's for Fig. 5 and 7 are not equal to those in the paper. The used Sankey library allows adjusting the nodes in the plot before saving the image to improve readability. Therefore, *.html files are additionally generated which allow adjusting the nodes to the layout seen in the paper.
To ensure the successful installation and correct execution, please run the image under the short execution setting and confirm that no errors where thrown.