Here are detailed instructions on how to get SnapPy working on a variety of platforms. The current version is 3.0.3 which was released on December 2021. If you encounter problems installing SnapPy, please let us know.
Simply download SnapPy.dmg and copy SnapPy.app to the Applications folder. Double-click to start it, just like any other application. Works with macOS/OS X 10.9 and newer. Earlier releases can be found here.
Simply download and run InstallSnapPy.exe. Earlier releases can be found here.
Here are short recipes which work on most Linux systems, specifically
those that run a 64-bit kernel and have Python 3.6 or newer. These
instructions assume you have system administrator (superuser)
privileges to install software packages from your Linux distribution
but want to install SnapPy (and its various Python dependencies) just
in your own user directory, specifically
~/.local. For other
Linux systems, try the one closest to yours below, and if that fails,
follow the instructions for generic Unix.
Ubuntu/Debian/Mint: Tested on Ubuntu 20.04:
sudo apt-get install python3-tk python3-pip # Note no "sudo" on the next one! python3 -m pip install --upgrade --user snappy
Users of Ubuntu 18.04 or older should do:
sudo apt-get install python3-tk python3-pip # Note no "sudo" on the next two python3 -m pip install --upgrade --user pip wheel python3 -m pip install --upgrade --user snappy
Fedora: Tested on Fedora 30:
sudo yum install python3-tkinter python3-pip # Note no "sudo" on the next one! python3 -m pip install --upgrade --user snappy
Red Hat Enterprise Linux/CentOS/SciLinux: These instructions are for version 7 or later, and you need to have the EPEL packages available. For CentOS and SciLinux, you can access EPEL packages by doing:
sudo yum install epel-release
Now install via:
sudo yum install python36-tkinter python36-pip # Note no "sudo" on the next one! python36 -m pip install --upgrade --user snappy
Arch/Manjaro: Install via:
sudo pacman -Sy python-pip tk # Note no "sudo" on the next one! python -m pip install --upgrade --user snappy
openSUSE: Install via:
sudo zypper install -y python3-tk python3-pip # Note no "sudo" on the next one! python3 -m pip install --upgrade --user snappy
If you want the larger version of HTLinkExteriors that includes the 15
crossing knots (uses 110M of disk space), also install the Python
snappy_15_knots, e.g. on Ubuntu do:
python3 -m pip install --upgrade --user snappy_15_knots
Once you have installed SnapPy, do the following to start it:
You may get a message about creating a “.ipython” directory; this is
normal, just hit return to continue. There should also now be a
command “SnapPy” which does the same thing. To make it so that you
can start SnapPy with just the command
SnapPy, make sure
~/.local/bin is in in your path.
Python Modules for Macintosh or Windows¶
If you write Python programs on a Macintosh or Windows system, you may wish to install SnapPy as a Python module into your own copy of Python. We support Python 3.6 and up. (On macOS, use a Python downloaded from Python.org and not the one provided by Apple.) After installing Python, you may install a SnapPy module from your Terminal application or Command Prompt with the commands:
python3 -m pip install --upgrade --user snappy
If you want the larger version of HTLinkExteriors that includes the 15 crossing knots (uses 110M of disk space), do:
python3 -m pip install --upgrade --user snappy_15_knots
If you use Python 2 rather than Python 3, replace
python in the above. If your Python lacks the pip module, get it
SnapPy has some special features when used within SageMath, the universal mathematics software based on Python. This section describes how to install SnapPy into your existing copy of SageMath, but you may find it easier to use the kitchen sink approach instead. You can install it as a Sage optional package via the following if using Sage 6.4 or newer:
sage -pip install snappy sage -pip install snappy_15_knots # Larger version of HTLinkExteriors
If you are on macOS, we recommend use this binary.
Alternatively, SageMath on CoCalc (formerly the SageMathCloud) also has SnapPy preinstalled, and the graphics features even work via the X11 interface, see the bottom of that page for more.
If you previously installed SnapPy into SageMath and want to upgrade SnapPy to the latest version, do:
sage -pip install --upgrade snappy
If it has trouble when compiling CyOpenGL, you are probably missing the “gl.h” headers <openglmesa>. The graphical features may or may not work, depending on how Tkinter was configured within Sage, and may seem to “hang” when you try to start them. To deal with the latter issue type “%gui tk” at the Sage prompt; please note that doing so may break Sage’s “attach” feature.
SnapPy gains extra features when used in SageMath and one can use Sage’s Python to interact not just with SnapPy but a range of other computational tools in low-dimensional topology including Regina, snap, heegaard, gridlink, and flipper. We offer a prepackaged Docker image with all of the above tools and many more; using this is frequently the easiest way to get a working setup for such multifaceted computations. For more, watch this demonstration.
We also offer conda environments with SnapPy and optionally Sage (only on Mac OS and Linux). While it has none of the other aforementioned tools, it has the advantage that the GUI elements such as the link editor and the browser can be used directly.
If you use a Unix other than OS X or Linux, or if the prebuilt packages don’t work for you, you’ll need to build SnapPy from source. Here are some detailed instructions.
Things you’ll need:
Python 3 with Tkinter: You’ll need to have Python (version 3.6 or newer) and Tk (at least version 8.5) with Tkinter to connect them, including the header files. For instance, on Debian or Ubuntu, install the packages “python3-tk”, “python3-pip”, and “python3-dev”. On Fedora, you’ll want e.g. “python3-tkinter”, “python3-pip”, and “python3-devel”, and “python3-wheel”.
Test that Python is in order by installing PLink from source:
python3 -m pip install --user plink python3 -m plink.app # Should start the link editor!
Support for OpenGL (3D graphics): This is built in on OS X and the most installations of Fedora and Ubuntu. But you’ll need the MESA header files “gl.h” and “glu.h” to compile SnapPy. On Debian and Ubuntu, install “libglu1-mesa-dev”; On Fedora install “mesa-libGLU-devel”.
Cython, which you can install via:
python3 -m pip install --user cython
The gcc C++ compiler, g++.
CyPari: a stand-alone version of Sage’s Python interface to the PARI number theory library. Usually, you can install this with:
python3 -m pip install --user cypari
Now download the source code listed below, for instance
wget https://pypi.python.org/packages/source/s/snappy/snappy-3.0.3.tar.gz tar xfz snappy-3.0.3.tar.gz; cd snappy-*
There is one more dependency that may need to be dealt with:
Togl: a 3d widget for Tk. For OS X and Linux, there are pre-built binaries of this in the snappy subdirectory, e.g. snappy/linux2-tk8.4. For Linux these are built for 64-bit kernels, and should work on most systems. If they don’t, you’ll need to edit or follow “build_togl.sh” to build Togl directly.
Finally, compile and install the SnapPy module (which will install certain other dependencies) and test:
python3 setup.py build python3 -m pip install --user . python3 -m snappy.test python3 -m snappy.app
The complete source code for all platforms: snappy-3.0.3.tar.gz
You can also browse our source code repository or clone it using git via:
git clone https://github.com/3-manifolds/SnapPy.git