Skip to content

rgu01/MoCReL-Experiments

Repository files navigation

Experiments for MoCReL

Experiment requirement

  • The experiments were conducted on a server with 48 CPU (Intel Xeon CPU E5-2678 v3 @ 2.50GHz), 258G memory, running Ubuntu 20.04.
  • Put the folder "MoCReL-Experiments" under "/home/username".

Experiment design

  • The models are seperated into 6 groups.
  • Groups A-E are for showing the performance of MoCReL dealing with different numbers of agents, tasks, milestones, and fixed/flexible task execution time.
  • Group Runs is for showing how the number of runs used by the learning algorithm can influence the quality of the synthesized strategies, i.e., the probabilities of satisfying the liveness property.

Run the experiments

  • If necessary, give permission to files by using: "chmod +x *.sh", "chmod +x stratego/*.*", "chmod +x stratego/verifyta", "chmod +x stratego/server".
  • Repeat the experiments by running the *.sh files. You must run the *.sh files under the "MoCReL-Experiments" folder.
  • See the models by opening them with the latest version of UPPAAL STRATEGO: https://people.cs.aau.dk/~marius/stratego/

A tool for model generation

  • The models used in the experiments are generated by a tool that is designed based on TG templates.
  • In the folder "model_generator", one can find the tool.

Additional statement

  • If the hardware of the laptops does not match the requirement, some models may exhaust the memory or Linux may kill the processes.
  • One can run "demo.sh" to see how strategies are synthesized, verified, and compacted for some simple models that are selected from groups A-E. These examples should be relatively easier and feasible on personal laptop.
  • The printed strategies of game4-A are stored under the diretory: ./strategies/A.
  • To see the full experiment results, please visit: https://docs.google.com/spreadsheets/d/1k1j6LebfhJNeM_RfCoSL5OBv3b5RAIhx31ypivd7PQQ/edit?usp=sharing
  • The binaries in the stratego folder are a beta version. Please do not distribute them.

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages