Installation Guide

Requirements

• NVIDIA CUDA GPU: Compute Capability of the GPU must be at least 3.0.

• CUDA Toolkit: Supported Versions: 8.0, 9.0, 9.1, 9.2, 10.0, and 10.1.

  (Note: Qualia2.0 can be used without GPU)

• Python 3.6+

Recommended

Ubuntu is recommended for the installation environment. Some packages fail to build on Windows.

• Ubuntu Desktop

Installation

Upgrade of setuptools and pip is recommended before the installation:

$ pip install -U setuptools pip

CUDA Toolkit version can be found by:

$ nvcc --version

Depending on the CUDA version you have installed on your host, choose the best option from following.

(For CUDA 8.0)
$ python setup.py install --cuda 80
(For CUDA 9.0)
$ python setup.py install --cuda 90
(For CUDA 9.1)
$ python setup.py install --cuda 91
(For CUDA 9.2)
$ python setup.py install --cuda 92
(For CUDA 10.0)
$ python setup.py install --cuda 100
(For CUDA 10.1)
% python setup.py install --cuda 101
(For without CUDA)
$ python setup.py install

Supplemental Information

Installation on Windows

For the installation on windows, there might be some error during the installation.

Here are some wheel files that might help:

• fastrlock
• atari-py
• box2d

The way to install from wheel is:

Note that you do need to install the wheel package first:

$ pip install wheel

Then,

$ pip install /path/to/the/wheel_file.whl

RL environments

For several challenging continuous control environments, OpenAI gym requires the user to install MuJoCo, a commercial physics engine which requires a license to run for longer than 30 days. Here are two options:

1. Install MuJoCo

1. Obtain a 30-day free trial on the MuJoCo website or free license if you are a student. The license key will arrive in an email with your username and password.

2. Download the MuJoCo version 2.0 binaries.

3. Unzip the downloaded mujoco200 directory into ~/.mujoco/mujoco200, and place your license key (the mjkey.txt file from your email) at ~/.mujoco/mjkey.txt.

4. Install mujoco-py

$ git clone https://github.com/openai/mujoco-py.git
$ cd mujoco-py
$ pip install -r requirements.txt
$ pip install -r requirements.dev.txt
$ python setup.py install

• To run mujoco-py, MuJoCo needs to be installed on the machine.
• For windows, install Microsoft Visual C++ Build Tools 2015 with Windows 8.1 SDK and Windows 10 SDK options

2. Install PyBullet Gymperium

I highly recommend PyBullet as a free open source alternative to MuJoCo for continuous control tasks.

PyBullet Gymperium is an open-source implementation of the OpenAI Gym MuJoCo environments for use with the OpenAI Gym Reinforcement Learning Research Platform in support of open research.

git clone https://github.com/benelot/pybullet-gym.git
cd pybullet-gym
pip install -e .