"Unsupported datatype returned from MATLAB" with use of MATLAB Engine for Python

Hello, It seems the "Engine for Python" is fairly new or not often used since there is very little discussion about it here or anywhere else. I'm running a MATLAB code that generates very large sparse matrices (binary entries ~30,000 by 30,000, stored as a sparse matrix). I need to run a python code on this data, and I need to do this a lot of times. The problem is that since the MATLAB code generates a matrix, Python doesn't know what to do. The tutorial on this subject demonstrates how to manually create python variables from MATLAB arrays, but there's nothing on matrix outputs from functions. Currently, when I issue the commands in Python:

import matlab.engine eng = matlab.engine.start_matlab() eng.mainf(3.0,0,10)

I get: TypeError: unsupported datatype returned from MATLAB

Any thoughts?

[SAMUEL HEROY]

thanks all

Answer 1

[Robert Snoeberger]

Does your function mainf return a sparse matrix? The documentation lists sparse array in unsupported MATLAB types:

http://www.mathworks.com/help/matlab/matlab_external/handle-data-returned-from-matlab-to-python.html

[Sarah Palfreyman]

True, as of R2014b there is not support for sparse storage types, and you can use issparse to check your data type. To use sparse data, recast the data to a matrix using full prior to passing it to the Python Engine. Generally speaking a MATLAB matrix of double values is returned to Python as matlab.double.