Converting a scipy.sparse matrix into an equivalent MATLAB sparse matrix

Question

I have a scipy.sparse.lil_matrix that I want to feed into a MATLAB method (that is not written by me) using the MATLAB Engine API for Python . The posts I've seen so far are either about how to convert a MATLAB sparse matrix into a python equivalent or they require modifying the matlab code which I'd rather circumvent.

Answer 1

Internally I believe MATLAB use the csc like format. But construction is (at least when I used it years ago) with coo style inputs - data, rows, cols.

I'd suggest making a sparse matrix in MATLAB, and saving it (in the pre-HDF5 mode) to a .mat. Then load that with scipy.io.loadmat . Then use that result as guide when writing a scipy.sparse matrix back to a .mat .

scipy.sparse has a save function, but it uses the np.savez to write the respective attribute arrays. If you had MATLAB code that could handle .npy files, you probably could load such a save (again using the coo format).

===

A test.

Create and save a sparse matrix:

In [263]: from scipy import io, sparse                                                                          
In [264]: M = sparse.random(10,10,.2,'coo')                                                                     
In [265]: io.savemat('sparse.mat', {'M':M})       

test load on Python side:

In [268]: io.loadmat('sparse.mat')                                                                              
Out[268]: 
{'__header__': b'MATLAB 5.0 MAT-file Platform: posix, Created on: Wed Jul  3 11:41:23 2019',
 '__version__': '1.0',
 '__globals__': [],
 'M': <10x10 sparse matrix of type '<class 'numpy.float64'>'
    with 20 stored elements in Compressed Sparse Column format>}

So savemat converted the coo format to csc before saving.

In an Octave session:

>> load sparse.mat
>> M
M =

Compressed Column Sparse (rows = 10, cols = 10, nnz = 20 [20%])

  (4, 1) ->  0.41855
  (6, 1) ->  0.33456
  (7, 1) ->  0.47791
  (4, 3) ->  0.27464
  (2, 4) ->  0.96700
  (3, 4) ->  0.60283
  (10, 4) ->  0.41400
  (1, 5) ->  0.57004
  (2, 5) ->  0.44211
  (1, 6) ->  0.63884
  (3, 7) ->  0.012127
  (8, 7) ->  0.77328
  (8, 8) ->  0.25287
  (10, 8) ->  0.46280
  (1, 9) ->  0.0022617
  (6, 9) ->  0.70874
  (1, 10) ->  0.79101
  (3, 10) ->  0.81999
  (6, 10) ->  0.12515
  (9, 10) ->  0.60660

So it looks like the savemat/loadmat code handles sparse matrices in a MATLAB compatible way.