This is another toolbox providing access to Python from MATLAB, with the goals of building on MATLAB's new support for Python.
Download! and put the folder on your MATLAB path.
Now, setup the Python installation:
>>> pybox install_conda % setup conda installation (dl if needed)
...
and install some packages
>>> pybox install numpy scipy pyqt % install some packages
...
Finally, to tell MATLAB to use this Python
>>> pybox use % tell MATLAB to use our python install
>>> py.sys.executable
ans =
Python str with no properties.
/home/duke/MATLAB/pybox/conda/bin/python
Make array of random numbers
>>> py.numpy.random.rand(3)
ans =
Python ndarray with properties:
T: [1x1 py.numpy.ndarray]
...
nbytes: 24
ndim: 1
real: [1x1 py.numpy.ndarray]
shape: [1x1 py.tuple]
size: 3
strides: [1x1 py.tuple]
[ 0.32473666 0.59561982 0.77702851]
MATLAB automatically exposes most stuff on Python objects in the console and the variable inspector, but it's useful to have the results as a MATLAB array:
>>> np2m(py.numpy.arange(8).reshape(2, 4))
ans =
0 4
1 5
2 6
3 7
Shapes are handled correctly.
Usuability is not bad, see how MATLAB even can make intelligent corrections
(here I misspelled QApplication
)
Miniconda from Continuum is used for setting up environments with prebuilt packages, in cross platform fashion (or at least as much as conda is) especially onerous for scientific libraries.