Skip to content

Julia

Julia is a high-level, high-performance programming language designed for technical and numerical computing, known for its speed and ease of use. Because of its popularity, we have pre-installed versions of Julia on each of our clusters. You can begin using Julia right away by running module load julia or by specifying the module specifically:

On Rocky8 nodes:

module load julia/1.10.1

On Centos7 nodes:

module load julia/1.8.5

module load julia/1.5.3-rls2opu

module load julia/1.10.1

Installing Packages

Julia organizes packages by the version of Julia you're running. When you install packages, a .julia folder automatically gets created in your home directory that holds all package installations. Julia will automatically create an environment for that version, which will be saved in ~/.julia/environments.

You can change the default package install location by setting the $JULIA_DEPOT_PATH environment variable from the command line before you start Julia. For example:

export JULIA_DEPOT_PATH=/home/$USER/orcd/r8/pool

Juliaup

Juliaup provides a convenient way to manage different versions of Julia and different package installations. It can be installed by running the following command:

curl -fsSL https://install.julialang.org | sh

You will be asked if you want to proceed with default settings or to customize your installation. We recommend customizing your installation. The default settings are as follows:

  1. Save the .juliaup folder to your home directory. This folder contains all installations of Julia and their associated packages that are managed by Juliaup.

    • On Satori, home directories are limited to 20 GB, so you may want to change this location to /nobackup/users/$USER.

  2. Edit your .bashrc and .bash_profile files to automatically add the Juliaup-managed version of Julia to your $PATH environment variable.

    • We generally discourage editing your .bashrc file because it can cause issues when trying to use other software. For example, if you want to use a pre-installed Julia module, you would have to manually remove Juliaup from your $PATH any time you connect to the cluster.

    • To add Juliaup to your $PATH manually, run: 

      export PATH=/path/to/.juliaup/bin${PATH:+:${PATH}}

Click here for more information on installing and using Juliaup.

Note

Currently, Juliaup is not compatible with Satori or SuperCloud.

Using Different Julia Versions

Juliaup Versions

If you're using Juliaup, installing different versions of Julia is straightforward:

# Install Julia 1.9.0:
juliaup add 1.9.0
# Use Julia 1.9.0:
julia +1.9.0

Note that Juliaup installs versions and packages to your .julia folder.

Manual Installation

If you are unable to use Juliaup and you need a version of Julia that is not pre-installed on the cluster, you can manually download it.

A complete list of previous Julia versions can be found here. From this site, copy the link to the .tar.gz file that corresponds to the version you need. Be sure to select the version for a Linux operating system and x86_64 architecture.

Download the .tar.gz file: 

wget [link to file]

Extract the .tar.gz file: 

tar -xvzf [file name]

Add the downloaded version to your path: 

export PATH="~/path/to/julia/bin:$PATH"

The following example is for Julia 1.9.0: 

wget https://julialang-s3.julialang.org/bin/linux/x64/1.9/julia-1.9.0-linux-x86_64.tar.gz
tar -xvzf julia-1.9.0-linux-x86_64.tar.gz
export PATH="~/julia-1.9.0/bin:$PATH"

Jupyter Notebooks

While Jupyter is heavily integrated with Python, it supports compatibility with Julia. You can run Jupyter notebooks on the web portals of SuperCloud and Engaging.

On the Engaging OnDemand web portal, you can specify one of the pre-installed Julia modules under the "Additional Modules" section. You can see which Julia modules are available by running module avail julia from the command line.

For Jupyter to recognize Julia, you need to have the IJulia package installed and built in your Julia environment:

using Pkg
Pkg.add("IJulia")
Pkg.build("IJulia")

Note that the Julia environment you are using needs to match the version of Julia that you are loading as a module.

To use a Jupyter notebook on the SuperCloud web portal, navigate to /jupyter/ and launch a notebook. When the session is running, open the notebook and select a Julia kernel.

For more information on running Jupyter notebooks on SuperCloud, check the SuperCloud documentation.

Port Forwarding

If you would like to use a different version of Julia that is not offered as a module, we suggest running a Jupyter notebook manually via port forwarding. This involves running the notebook on a compute node, and then accessing the notebook on your local machine by SSH tunnelling through a login node.

To do this, first request a compute node with your desired resources:

salloc -N 1 -n 4 -p mit_normal

Make a note of the node that your job is running on. In this example, we're running on node1600.

Then, start your desired version of Julia add the IJulia package to your environment:

using Pkg
Pkg.add("IJulia")
Pkg.build("IJulia")
quit()

Because of Jupyter's interaction with Python, you need to create and activate a Conda environment with jupyter installed:

module load miniforge
conda create -n jupyter_env jupyter
conda activate jupyter_env

Note

For more information on Conda, see our Python documentation.

Now, we can run the notebook. To be able to access it on our local machine, we need to add a few arguments:

jupyter-lab --ip=0.0.0.0 --port=8888

The port can be any number between 1024 and 9999. When you run the notebook, the output will contain a link with a token that allows you to access the notebook:

http://127.0.0.1:<remote port>/lab?token=<token>

For example:

http://127.0.0.1:8888/lab?token=7e97d59f9a17c91c11289bc5bec35ad3921725c6db55fe33

In a second terminal window, set up an SSH tunnel to your Jupyter notebook that's running on the compute node:

ssh -L <local port>:<node>:<remote port> <USER>@orcd-login001.mit.edu

In general, it's easier if you keep the local port and the remote port as the same number:

ssh -L 8888:node1600:8888 <USER>@orcd-login001.mit.edu

Now you can access Jupyter in an internet browser:

http://127.0.0.1:<local port>/lab?token=<token>

If you kept the local and remote ports as the same number, then you can directly copy the link that was given to you earlier:

http://127.0.0.1:8888/lab?token=7e97d59f9a17c91c11289bc5bec35ad3921725c6db55fe33

VS Code

Please refer to the VS Code page for using VS Code on the cluster.

VS Code supports compatibility with Jupyter notebooks. If you have installed and built IJulia in your Julia environment, then you should be able to find the correct Julia kernel by navigating to Select Kernel > Select Another Kernel > Jupyter Kernel.

FAQs

I have loaded/installed a specific version of Julia, but it is not being recognized. What do I do?

There may be another/no version of Julia in your PATH environment variable. You can check this by running echo $PATH.

If you have loaded a Julia module but do not want to use it, you can run module purge.