Bregman convolutional barycenter 2d return an error for non-square images

[ThibautLG]

When ever I call
ot.bregman.convolutional_barycenter2d(A, reg, weights)
with A whose shape is not of the form (k,n,n), the function returns an error.

In my case, using the same code as in the example in the doc, I get the following error for (600,500) images.

ValueError                                Traceback (most recent call last)
<ipython-input-13-f723cd8affb9> in <module>
     55         else:
     56             # call to barycenter computation
---> 57             pl.imshow(ot.bregman.convolutional_barycenter2d(A, reg, weights), cmap=cm)
     58             pl.axis('off')
     59 pl.show()

~/.local/lib/python3.6/site-packages/POT-0.6.0-py3.6-linux-x86_64.egg/ot/bregman.py in convolutional_barycenter2d(A, reg, weights, numItermax, stopThr, stabThr, verbose, log)
   1361         b = np.zeros_like(A[0, :, :])
   1362         for r in range(A.shape[0]):
-> 1363             KV[r, :, :] = K(A[r, :, :] / np.maximum(stabThr, K(U[r, :, :])))
   1364             b += weights[r] * np.log(np.maximum(stabThr, U[r, :, :] * KV[r, :, :]))
   1365         b = np.exp(b)

~/.local/lib/python3.6/site-packages/POT-0.6.0-py3.6-linux-x86_64.egg/ot/bregman.py in K(x)
   1352 
   1353     def K(x):
-> 1354         return np.dot(np.dot(xi1, x), xi1)
   1355 
   1356     while (err > stopThr and cpt < numItermax):

<__array_function__ internals> in dot(*args, **kwargs)

ValueError: shapes (600,500) and (600,600) not aligned: 500 (dim 1) != 600 (dim 0)

My system information:

Linux-5.3.0-28-generic-x86_64-with-Ubuntu-18.04-bionic
Python 3.6.9 (default, Nov 7 2019, 10:44:02)
[GCC 8.3.0]
NumPy 1.17.4
SciPy 1.3.3
POT 0.6.0

[rflamary]

This issue has been clsed in PR #132