A Scheme kernel for Jupyter that can use Python libraries





GitHub Stars



Last Commit

17d ago









Calysto Scheme

codecov CircleCI

You can try Calysto Scheme without installing anything by clicking on the following button:


Calysto Scheme is a real Scheme programming language, with full support for continuations, including call/cc. It can also use all Python libraries. Also has some extensions that make it more useful (stepper-debugger, choose/fail, stack traces), or make it better integrated with Python. For more details on using Calysto Scheme, see:

In Jupyter notebooks, because Calysto Scheme uses MetaKernel, it has a fully-supported set of "magics"---meta-commands for additional functionality. This includes running Scheme in parallel. See all of the MetaKernel Magics.

Calysto Scheme is written in Scheme, and then translated into Python (and other backends). The entire functionality lies in a single Python file: However, you can easily install it (see below).

Calysto Scheme in use:

Parallel Processing

To use Calysto Scheme in parallel, do the following:

  1. Make sure that the Python module ipyparallel is installed. In the shell, type:
pip install ipyparallel
  1. To enable the extension in the notebook, in the shell, type:
ipcluster nbextension enable
  1. To start up a cluster, with 10 nodes, on a local IP address, in the shell, type:
ipcluster start --n=10 --ip=
  1. Initialize the code to use the 10 nodes, inside the notebook from a host kernel (can be any metakernel kernel), type:
%parallel calysto_scheme CalystoScheme
  1. Run code in parallel, inside the notebook, type:

Execute a single line, in parallel:

%px (+ 1 1)

Or execute the entire cell, in parallel:

(* cluster_rank cluster_rank)

Results come back in a Scheme vector, in cluster_rank order. Therefore, the above would produce the result:

#10(0 1 4 9 16 25 36 49 64 81)

You can get the results back in the host Scheme by accessing the variable _ (single underscore).

Notice that you can use the variable cluster_rank to partition parts of a problem so that each node is working on something different.

In the examples above, use -e to evaluate the code in the host Scheme as well. Note that cluster_rank is not defined on the host machine, and that this assumes the host kernel is the same as the parallel machines.

A full notebook example can be found here: Mandelbrot.ipynb


You can install Calysto Scheme with Python3:

pip3 install --upgrade calysto-scheme --user
python3 -m calysto_scheme install --user

or in the system kernel folder with:

sudo pip3 install --upgrade calysto-scheme
sudo python3 -m calysto_scheme install

You can also use the --sys-prefix to install into your virtualenv.

Change pip3/python3 to use a different pip or Python. The version of Python used will determine how Calysto Scheme is run.

Use it in the Jupyter console, qtconsole, or notebook:

jupyter console --kernel calysto_scheme
jupyter qtconsole --kernel calysto_scheme
jupyter notebook

You can also just use the Python program, but it doesn't have a fancy Read-Eval-Print Loop. Just run:

python calysto_scheme/


  • Python3
  • metakernel (installed automatically)

Calysto Scheme can also be un under PyPy for increased performance.


Calysto Scheme supports:

  • continuations
  • use of all Python libraries
  • choose/fail - built in fail and try again
  • produces stack trace (with line numbers), like Python
  • test suite


  • Object-oriented class definitions and instance creation
  • complete Scheme functions (one can fall back to Python for now)


  • Runs slow on CPython; try PyPy

Rate & Review

Great Documentation0
Easy to Use0
Highly Customizable0
Bleeding Edge0
Responsive Maintainers0
Poor Documentation0
Hard to Use0
Unwelcoming Community0