Welcome to sched-rl-gym’s documentation!

In this documentation site you will find information about the general design of the environment, a set of tutorials on how to instantiate and use it, along with a reference of the source code.

Please use the following list of contents to select what you’re interested in.

Contents:

sched-rl-gym is an OpenAI Gym environment for job scheduling problems. Currently, it implements the Markov Decision Process defined by DeepRM.

You can use it as any other OpenAI Gym environment, provided the module is registered. Lucky for you, it supports auto registration upon first import.

Therefore, you can get started by importing the environment with import schedgym.envs as schedgym.

As a parallel with the CartPole example in the Gym documentation, the following code will implement a random agent:

import gym
import schedgym.envs as schedgym

env = gym.make('DeepRM-v0', use_raw_state=True)
env.reset()

for _ in range(200):
  env.render()
  observation, reward, done, info = env.step(env.action_space.sample())
env.close()

With the following rendering:

OpenAI Gym Environment rendering

OpenAI Gym Environment rendering

Features

  • OpenAI Gym environment

  • Human rendering

  • Configurable environment

Installation

The easiest/quickest way to install sched-rl-gym is to use pip with the command:

pip install -e git+https://github.com/renatolfc/sched-rl-gym.git#egg=sched-rl-gym

We do recommend you use a virtual environment, to not pollute your python installation with custom packages.

If you want to be able to edit the code, then your best bet is to clone this repository with

git clone https://github.com/renatolfc/sched-rl-gym.git

In this case, you will need to install the dependencies manually.

Dependencies

The dependencies are documented in the requirements.txt file. You can install them with

pip install -r requirements.txt

Contribute

Support

If you’re having issues, please let us know. The easiest way is to open an issue on github.

License

The project is licensed under the MIT license.

Indices and tables