View on GitHub


A simple Python => MATLAB(R) interface and a matlab_magic for ipython

Download this project as a .zip file Download this project as a tar.gz file

Python-MATLAB(R) Bridge and ipython matlab_magic

A python interface to call out to Matlab(R). Original implementation by Max Jaderberg.

This implementation also includes an IPython matlab_magic extension, which provides a simple interface for weaving python and Matlab code together (requires ipython > 0.13).


Pymatbridge communicates with Matlab using zeromq. So before installing pymatbridge you must have zmq library and pyzmq installed on your machine. If you intend to use the Matlab magic extension, you'll also need IPython. To make pymatbridge work properly, please follow the following steps.

Install zmq library

Please refer to the official guide on how to build and install zmq. After zmq is installed, make sure you can find the location where libzmq is installed. The library extension name and default location on different systems are listed below.

Platform library name Default locations
MacOS libzmq.dylib /usr/lib or /usr/local/lib
Linux /usr/lib or /usr/local/lib
Windows libzmq.dll C:\Program Files\ZeroMQ 3.2.4\bin

If you specified a prefix when installing zmq, the library file should be located at the same prefix location.

After the library file is located, you need to add it to dynamic loader path on your machine. On MacOS, you can do this by adding the following line to your .bash_profile:

export DYLD_LIBRARY_PATH=$DYLD_LIBRARY_PATH:<Path to your zmq lib directory>

On Linux, add the following line to your .bash_profile:

export LD_LIBRARY_PATH=$DYLD_LIBRARY_PATH:<Path to your zmq lib directory>

On Windows, add the install location of libzmq.dll to the PATH environment variable. On Windows 7+, typing "environment variables" into the start menu will bring up the apporpriate Control Panel links.

Install pyzmq

After step 1 is finished, please grab the latest version of pyzmq and follow the instructions on the official page. Note that pymatbridge is developed with pyzmq 14.0.0 and older versions might not be supported. If you have an old version of pyzmq, please update it.

Install pymatbridge

After the steps above are done, you can install pymatbridge. Download the zip file of the latest release. Unzip it somewhere on your machine and then issue:

python install

This should make the python-matlab-bridge import-able.


To use the pymatbridge you need to connect your python interperter to a Matlab session. This is done in the following manner:

from pymatbridge import Matlab
mlab = Matlab()

This creates a matlab session class instance, into which you will be able to inject code and variables, and query for results. By default, when you use start, this will open whatever gets called when you type matlab in your Terminal, but you can also specify the location of your Matlab application when initialzing your matlab session class:

mlab = Matlab(matlab='/Applications/')

You can then start the Matlab server, which will kick off your matlab session, and create the connection between your Python interperter and this session:


which will return True once connected.

results = mlab.run_code('a=1;')

Should now run that line of code and return a results dict into your Python namespace. The results dict contains the following fields:

{u'content': {u'code': u'a=1',
 u'datadir': u'/private/tmp/MatlabData/',
 u'figures': [],
 u'stdout': u'\na =\n\n     1\n\n'},
 u'success': u'true'}

In this case, the variable a is available on the Python side, by using the get_variable method:


You can run any MATLAB functions contained within a .m file of the same name. For example, to call the function jk in jk.m:

function lol = jk(args)
    arg1 = args.arg1;
    arg2 = args.arg2;
    lol = arg1 + arg2;

you would call:

res = mlab.run_func('path/to/jk.m', {'arg1': 3, 'arg2': 5})
print res['result']

This would print 8.

You can shut down the MATLAB server by calling:


Tip: you can execute MATLAB code at the beginning of each of your matlab sessions by adding code to the ~/startup.m file.

Matlab magic:

The Matlab magic allows you to use pymatbridge in the context of the IPython notebook format.

%load_ext pymatbridge

These lines will automatically start the matlab session for you. Then, you can simply decorate a line/cell with the '%matlab' or '%%matlab' decorator and write matlab code:

a = linspace(0.01,6*pi,100);
grid on
hold on

More examples are provided in the examples directory


Python communicates with Matlab via an ad-hoc zmq messenger. This is inherently insecure, as the Matlab instance may be directed to perform arbitrary system calls. There is no sandboxing of any kind. Use this code at your own risk.