An enhanced interactive environment for the computer algebra system Maxima, based on CL-Jupyter, a Jupyter kernel for Common Lisp, by Frederic Peschanski. Thanks, Frederic!
This file describes the installation and usage of Maxima-Jupyter on a local machine, but you can try out Maxima-Jupyter without installing anything by clicking on the Binder badge above.
MaximaJupyterExample.ipynb — General usage of Maxima from within Jupyter Notebook.
MaximaJupyterTalk.ipynb — My notes for a talk given to the Portland Python User Group.
Plots.ipynb — Usage of plotting facilities from within Jupyter Notebook.
These examples make use of nbviewer. You can submit a link to your own notebook to tell nbviewer to render it.
Note that the Github notebook renderer (i.e., what you see if you click on a notebook file in the Github file browser) is currently (November 2018) somewhat suboptimal (bug report); it renders math as plain text, not as typeset formulas.
Maxima-Jupyter may be installed on a machine using a local installation, a repo2docker installation, or via a Docker image.
To try Maxima-Jupyter you need :
a Maxima executable
built with a Common Lisp implementation which has native threads
SBCL works for sure
Clozure CL works for sure
Other implementations which support the Bordeaux Threads package might work. The Bordeaux Threads project description says "Supports all major Common Lisp implementations: SBCL, CCL, Lispworks, Allegro, ABCL, ECL, Clisp." Aside from SBCL and CCL (i.e. Clozure CL) which are known to work, the others in that list are untested with maxima-jupyter.
Note also that ECL might theoretically work, since it is supported by Bordeaux Threads. However, nobody (neither Maxima-Jupyter developers nor users) has been able to get ECL to work, therefore you should assume ECL does not work with Maxima-Jupyter. SBCL and Clozure CL are known to work, try those instead.
Note specifically that GCL is not supported by Bordeaux Threads, and therefore GCL cannot work with maxima-jupyter.
You might or might not need to build Maxima. (A) If you have available a Maxima binary package compiled with a compatible Lisp implementation (i.e. SBCL, Clozure CL, Lispworks, etc. as enumerated above), then you do not need to build Maxima. (B) Otherwise, you must install a compatible Lisp implementation and compile Maxima yourself.
- When you load Maxima-Jupyter into Maxima for the first time, Quicklisp will download some dependencies automatically. Good luck.
Python 3.2 or above
Jupyter, or IPython 3.x
If the build aborts because the file
zmq.his missing, you may need to install the development files for the high-level C binding for ZeroMQ. On debian-based systems, you can satisfy this requirement by installing the package
The following installation methods are still relatively new. If you experience any issues than please use the "Old Methods" detailed later in the document.
First you must install Jupyter, then you can install Maxima-Jupyter. If you
plan on using JupyterLab then you must install with the
python3 -m pip --user install jupyter jupyterlab
Once Jupyter is installed you can either install from the source files of this repository, or you can install via the AUR if you are using Arch Linux.
Method 1. Source Based Installation
To install from the current source files first download the source files and then start a shell in the source directory. Then start Maxima and load the initialization script.
$ maxima Maxima 5.43.0 http://maxima.sourceforge.net using Lisp SBCL 1.5.5 Distributed under the GNU Public License. See the file COPYING. Dedicated to the memory of William Schelter. The function bug_report() provides bug reporting information. (%i1) load("load-maxima-jupyter.lisp");
After the install script has loaded then install using one of the kernel types.
- User specific Quicklisp kernel:
- User specific binary image kernel:
- System-wide Quicklisp bundled kernel:
After the installation is complete then exit Maxima. For the System-wide
installation copy the files in
pkg to the system root, i.e.
sudo cp -r pkg/* / on Linux.
Method 2. Installation on Arch/Manjaro
The package for Arch Linux is maxima-jupyter-git. Building and installing (including dependencies) can be accomplished with:
yaourt -Sy maxima-jupyter-git
curl -L -O https://aur.archlinux.org/cgit/aur.git/snapshot/maxima-jupyter-git.tar.gz tar -xvf maxima-jupyter-git.tar.gz cd maxima-jupyter-git makepkg -Csri
Please consult the Arch Wiki for more information regarding installing packages from the AUR.
Installing Maxima-Jupyter (Old Method)
First you must install Jupyter, then you can install Maxima-Jupyter.
I installed Jupyter via:
python3 -m pip install jupyter
For Maxima-Jupyter, there are two kernel installation methods.
In both methods, the effect of the installation command is to create a file
kernel.json which tells Jupyter where to find Maxima-Jupyter.
Note that Maxima-Jupyter installation DOES NOT copy any Maxima-Jupyter files;
it only creates
kernel.json which points to the location of Maxima-Jupyter
in your file system.
--user option in Method 1 or Method 2,
kernel.json file is created in a directory somewhere under your home directory.
kernel.json is created in a system directory.
You might need superuser privilege (via
sudo for example) to execute a system installation,
if the directory into which
kernel.json is copied is not user-writable.
jupyter --paths lists file system paths used by Jupyter;
kernels are sought in the paths under
jupyter kernelspec list tells the kernels which are known to Jupyter.
For the record, on my system, a system installation copies
and a user installation copies
Method 1. Maxima-Jupyter binary executable installation (Old Method)
The first installation method is to create a binary executable image, as detailed in make-maxima-jupyter-recipe.txt. After creating that image, execute one of these two commands to tell Jupyter about it.
For a system installation,
python3 ./install-maxima-jupyter.py --exec=path/to/maxima-jupyter-image
For a user installation,
python3 ./install-maxima-jupyter.py --exec=path/to/maxima-jupyter-image --user
Method 2. Maxima-Jupyter loadable source installation (Old Method)
The second installation method executes Maxima and then loads Maxima-Jupyter into Maxima.
The advantange to this method is that the normal initialization behavior of Maxima,
such as loading
maxima-init.mac, is preserved.
Note that in order for this method to work, Quicklisp needs be loaded by default in every Maxima session. See Quicklisp documentation for details.
For a system installation,
python3 ./install-maxima-jupyter.py --root=`pwd`
where the shell command
pwd emits the current working directory
(which must be the Maxima-Jupyter top-level directory,
since it contains
For a user installation,
python3 ./install-maxima-jupyter.py --root=`pwd` --user
--maxima may also be used to specify the location of the Maxima executable.
If not specified, the command which launches Maxima is just
therefore the first instance of
maxima in the PATH environment variable
is the one which is executed.
Code Highlighting Installation
Highlighting Maxima code is handled by CodeMirror in the notebook and Pygments in HTML export.
The CodeMirror mode for Maxima is maxima.js. To install it, find the
CodeMirror mode installation directory, create a directory named
copy maxima.js to the
maxima directory, and update
codemirror/mode/meta.js as shown in codemirror-mode-meta-patch. Yes, this
is pretty painful, sorry about that.
The Pygments lexer for Maxima is maxima_lexer.py. To install it, find the
Pygments installation directory, copy maxima_lexer.py to the
lexers directory, and
lexers/_mapping.py as shown in pygments-mapping-patch. Yes, this is
pretty painful too.
Maxima-Jupyter may be run from a local installation in console mode by the following.
jupyter console --kernel=maxima
Notebook mode is initiated by the following.
When you enter stuff to be evaluated, you must include the usual trailing semicolon or dollar sign:
In : 2*21; Out: 42 In :
Maxima-Jupyter may be run as a Docker image managed by repo2docker which will fetch the current code from GitHub and handle all the details of running the Jupyter Notebook server.
First you need to install repo2docker (
sudo may be required)
pip install jupyter-repo2docker
Once repo2docker is installed then the following will build and start the server. Directions on accessing the server will be displayed once the image is built.
jupyter-repo2docker --user-id=1000 --user-name=mj https://github.com/robert-dodier/maxima-jupyter
A Docker image of Maxima-Jupyter may be built using the following command
sudo may be required). This image is based on the docker image
docker build --tag=maxima-jupyter .
If you'd like to build with a different user than the default (
mj), you may
override it with the following:
docker build --build-arg NB_USER=alice --tag=maxima-jupyter .
After the image is built the container may be run with:
docker run -it maxima-jupyter
Dockerfile makes use of the
ENTRYPOINT command; the default behaviour
jupyter binary with the arguments
If you'd like to run using Juypter's notebook web server, you may do the
following to override the default use of
docker run -it \ -v `pwd`/notebooks:/home/USER/maxima-jupyter/examples \ -p 8888:8888 \ maxima-jupyter \ notebook --ip=0.0.0.0 --port=8888
where the last line is the set of arguments to
jupyter that cause it to run
in the notebook server mode.
To run the Bash shell on the container, just override the entry point:
docker run -it --entrypoint=bash maxima-jupyter
If you cannot build the Docker image, you may use a
by subsituting the Docker image name
maxima-jupyter in the above
calyau/maxima-jupyter. Note that the default user on the
calyau image is not
mj, but is rather
Additional examples of notebooks created using this mode have been created here (taken from the Maxima tutorial): https://github.com/calyau/maxima-tutorial-notebooks.
Have fun and keep me posted. Feel free to send pull requests, comments, etc.
Robert Dodier email@example.com robert-dodier @ github