Install information:

A few debian packages will need to be installed for the wrapper to work.

To do this, run: sudo ./mf-install/install.sh

You will need to set the MODIFLY_BIN_PATH environment variable to /usr/bin, and your append /usr/lib to your Python path.

to run the interactive experiment, use this command:

python main.py -algo ppo

this will run PPO on the environment we imported in the main.py and utils.py files. Right now the environment is imported with the name as env.py. If you made a new environemnt and save it with a different name, make sure to change the import statement in main.py and utils.py files.

if you want to change the algorithm, you can change the algo parameter in the command above. The available algorithms are:

python main.py -algo a2c python main.py -algo ddpg python main.py -algo sac python main.py -algo td3

How to run experiments in passive mode:

make sure you have fold in the exp/modify folder. Once the folder is there, you can run the following command to run the experiment in passive mode:

cd lhpo

then run this command to parse the rules:

./parsing_rules.bash ../exp/modifly/

if you want to count the number of experiments, you can run this command:

./count.bash ../exp/modifly/

if you want to run the experiments, you can run this command:

./optimizer.bash ../exp/modifly/ 2. here 2 is the number of experiments you want to run. You can change this number to run more or less experiments.

let me know if you have any questions or issues.