Get Started

In Few Easy steps!

QuickStart on the Cloud

The simplest way to get started with NetKet is to use Google’s Colab, an online notebook environment with access to GPUs. If you have a google, account, simply head to NetKet’s documentation, select the first tutorial, and click on the small rocket icon on the top navigation bar. This will launch the tutorial on an online environment that is ready to use.

For example, you can simply click on this button and get started! The notebook will automatically install netket.

Open in Colab

Local Installation

You can also install NetKet on your local computer following those instructions:

Make sure you have Python 3.7 or higher installed on Mac or Linux

If Python 3.7 is not available on your computer, don’t despair! You can use pyenv (the easiest way to install it is with the pyenv installer to install any Python version, or you can use Anaconda, even though the latter is not recomended if you plan on using MPI.

Windows is not natively supported because Jax does not yet support it. However, if you use WSL (Windows Subsystem for Linux) NetKet will run smoothly.

If you plan to use MPI, make sure you have an up to date version of the mpicc compilers available on your path.

When using MPI, we recommend not to use Anaconda unless it’s for small experimentation on a laptop. This is due to a dependency of netket, mpi4jax. You can read more about the limitations on the mpi4jax documentation.

Use this widget below to find the best command to install NetKet.

Quick Start Locally


Select your preferences and run the install command. More text here as needed.

Your OS
Linux
MacOS
MacOS ARM
Pkg Manager
Pip
Conda
Git Master
Accellerator
CPU
CPU/GPU
Extras
MPI
QuTiP
OpenFermion
Run this command
conda install windows 3.7

If you have issues, check the full documentation.

Explore our Tutorials and check out our Examples. If you still have questions, get in touch with us!

  • Tutorials are commented python notebooks.
  • Examples are sample python files that can be run.
  • If you still have questions, feel free to get in touch through our GitHub Issues or to start a discussion thread
  • If you want to experiment with mpi, try running some examples with
mpirun -np2 python path/to/example.py