Jupylet is a Python library for programming 2D and 3D games, graphics, music and sound synthesizers, interactively in a Jupyter notebook. It is intended for three types of audiences:

• Computer scientists, researchers, and students of deep reinforcement learning.
• Musicians interested in sound synthesis and live music coding.
• Kids and their parents interested in learning to program.

A Jupyter notebook is in essence a laboratory for programming. It is the ideal environment for playing around with code, experimenting, and exploring ideas. It is used by professional machine learning scientists who come every day to play at work, so why not by kids?

Jupylet is wonderfully easy to use for creating simple 2D and 3D games and music interactively and experimentally. Change a variable or a function and see how the game is affected immediately while running.

Jupylet makes it is super easy to create and modify environments in which to experiment with deep reinforcement learning algorithms and it includes the API to programmatically control multiple simultaneous games and render thousands of frames per second.

Consider for example the pong game included in this code base. With a few lines of code you can modify the colors of the game to experiment with transfer learning, or turn the game into 4-way pong with agents on all four sides of the game court to experiment with cooperation between multiple agents. And since you can modify the game interactively in Jupyter this process is not only easy but also fun.

Check out the Programming Graphics and the Reinforcement Learning chapters in the Jupylet Programmer's Reference Guide.

Jupylet imports ideas and methods from machine learning into the domain of sound synthesis to easily let you create sound synthesizers as wild as you can dream up - it includes impulse response reverb effects, colored noise generators, resonant filters with cutoff frequency sweeping, oscillators with LFO modulation, multi sampled instruments, and much more... And all of it in pure Python for you to modify and experiment with.

In addition Jupylet draws inspiration from the wonderful Sonic Pi and brings live loops and live music coding to Jupyter and Python. Hook up your MIDI keyboard and take off.

Check out the Programming Sound and Music and the Programming Synthesizers chapters in the Jupylet Programmer's Reference Guide.

Jupylet should run on Python 3.9 and up on Windows, Mac, and Linux.

If you are new to Python, I recommend that you install and use the Miniconda Python distribution.

On Windows – download and run the 64-bit installer for Python 3.11. Once Miniconda is installed press the ⊞ Winkey and then type Miniconda and press the Enter key. This should open a small window that programmers call console or shell in which you can enter commands and run programs.

On macOS with M1 processor – download and run "Miniconda3 macOS Apple M1 64-bit pkg" for Python 3.11. Once installed click the Spotlight icon 🔍 and in the search field type terminal and press the Enter key to open the console. Then you need to run the following command:

pip install --extra-index https://github.com/nir/jupylet/releases/download/v0.9.2/ moderngl glcontext

On macOS with Intel processor – download and run "Miniconda3 macOS Intel x86 64-bit pkg" for Python 3.11. Once installed click the Spotlight icon 🔍 and in the search field type terminal and press the Enter key to open the console.

On Linux – download "Miniconda3 Linux 64-bit". This should download the file Miniconda3-latest-Linux-x86_64.sh. Install it by running the following command in a bash shell (once installed start a new bash shell):

bash Miniconda3-latest-Linux-x86_64.sh

Once Miniconda is installed it is time to install jupylet by typing the following command in the console:

pip install jupylet

Next, to run the example notebooks, download the jupylet source code. If you have Git installed type the following command:

git clone https://github.com/nir/jupylet.git

Alternatively, you can download the source code with the following command:

python -m jupylet download

Next, enter the jupylet/examples/ directory with the change directory command:

cd jupylet/examples/

And start a jupyter notebook with:

jupyter notebook 11-spaceship.ipynb

Run the notebook by following the instructions in the notebook and a game canvas should appear with the spaceship example:

Alternatively, you can run the same game as a Python script from the console with:

python spaceship.py

To get started with Jupylet head over to the Jupylet Programmer's Reference Guide which you can find at jupylet.readthedocs.io.

To complement the online guide check out the growing collection of example notebooks that you can download and run on your computer as explained above.

For questions and feedback send an email to Nir Aides or join the discussion.

Jupylet is a new library and you can help it grow with a few clicks - if you like it let your friends know about it!

• Einar Forselv - The programmer behind ModernGL for his endless help in the trenches of OpenGL programming.
• Alban Fichet - For kindly licensing his sound visualizer Shadertoy as CC BY 4.0 license.

• Support for Python 3.10 and Python 3.11 with MIDI functionality.
• Seamlessly track changes to audio devices on macOS.
• Workaround PIL api change - thanks to @misolietavec.
• Bug fixes.

• Support for Python 3.10 and Python 3.11 - except for MIDI functionality.
• Support for macOS M1.
• Spectrum analyzer.
• Bug fixes.

• Support for Python 3.9.

• Workaround auto-completion bug in Jupyter notebooks.

• Support for rendering Shadertoy OpenGL shaders. Shadertoy is an awesome online platform for
  programming and sharing OpenGL shaders online, and now you can use and render shadertoy shaders in Jupylet!