Skip to content

Latest commit

 

History

History
46 lines (30 loc) · 1.8 KB

README.md

File metadata and controls

46 lines (30 loc) · 1.8 KB

Investigating the Impact of Direct Punishment on the Emergence of Cooperation in Multi-Agent Reinforcement Learning Systems

Code to replicate results of "Investigating the Impact of Direct Punishment on the Emergence of Cooperation in Multi-Agent Reinforcement Learning Systems"

Set Up

Use the package manager pip to install the requirements from the requirements.txt file.

source ./venv/bin/activate
pip install -r requirements.txt
export PYTHONPATH="$PWD:$PYTHONPATH"

Every Simulation folder e.g. Simulations/DP_S/Centipede should contain a Logs and a Results folder before running the simulation.

e.g.

├── Simulations
│   ├── DP
│   │   ├── Centipede
│   │   │   ├── Logs
│   │   │   ├── Results
│   │   │   ├── MultipleRunner_DPOnly_punishOnlyRep_repInPlayState_centipede.py
│   │   │   └── play_DPOnly_punishOnlyRep_repInPlayState_centipede.py

Simulations

Run simulations from the root folder as so:

python Simulations/TPP_S/Centipede/MultipleRunner_TPPSelect_playPunishRep_repInPlayState_centipede.py

Analysis

Use the metric and plotting functions in the analysis.py file to analyse the results.

Naming Conventions

play_MECHANISM_REPINFO_REPSTATE_CONTEXT denotes a single simulation of the social mechanism MECHANISM when reputation is calculated using the information defined inREPINFO and reputational information is added to REPSTATE. This simulation is performed with the following CONTEXT e.g. social dilemma.

MultipleRunner_MECHANISM_REPINFO_REPSTATE_CONTEXT runs a single simulation multiple times to allow for rolling mean and a 95% confidence interval to be calculated.