Exercises in reinforcement learning.

1. Clone Repository.

   git clone https://github.com/djjh/reinforcement-learning-labs

2. Go to working directory. All steps assume you are working from /path/to/repo-directory.

   cd reinforcement-learning-labs

brew install pyenv

1. Install required headers on Mac OSX Mojave. Without this step, you may see this error while installing python versions: 'zipimport.ZipImportError: can't decompress data; zlib not available.

   open /Library/Developer/CommandLineTools/Packages/ macOS_SDK_headers_for_macOS_10.14.pkg

   Sources:
   • pyenv/pyenv#1219 (comment)
2. Install python 3.6.3. On Mac OSX you must also enable a framework build to use matplotlib.

   env PYTHON_CONFIGURE_OPTS="--enable-framework" pyenv install 3.6.3

   Sources:
   • https://matplotlib.org/faq/osx_framework.html
   • https://github.com/pyenv/pyenv/wiki#how-to-build-cpython-with-framework-support-on-os-x

1. Enable the new version of python you just installed.

   pyenv local 3.6.3

   If you want to use this new version by default, then:

   pyenv global 3.6.3

2. Create the virtual environment. Note that you must use venv to be able to use the framework build of python you just installed.

   cd /path/to/repo-directory
   mkdir virtual-environment
   cd virtual-environment
   python -m venv --prompt reinforcement-learning-labs .

3. Enable the virtual environment.

   source bin/activate

4. Install dependencies using pip.

   pip install -e ..

All steps assume you are in working directory /path/to/repo-directory.

cd reinforcement-learning-labs

pyenv local 3.6.3

source virtual-environment/bin/activate

To run an experiment:

python -m experiments.<experiment_name>  # e.g. python -m experiments.03_basic

• python setup scripts: https://docs.python.org/3.6/distutils/setupscript.html
• pytest good practices: https://docs.pytest.org/en/latest/goodpractices.html
• python import traps: http://python-notes.curiousefficiency.org/en/latest/python_concepts/import_traps.html