Reshape arbitrary length vector into square matrix with padding in numpy

Question

I have an arbitrary length vector, and I want to reshape it into a square matrix, something like:

np.arange(6).reshape((3, 3))

[1,2,x]   [1,2,3]    
[3,4,x]   [4,5,6]
[5,6,x]   [x,x,x] 

The x can be horizontally and/or vertically placed.

Obviously reshape function will only allow parameter like (3,2) in the above example. Is there a way to produce effect of squared shaped matrix.Thanks.

Answer 1

You'll have to pad the array, either before or after reshape.

For example, using the resize method to add the needed 0s:

In [409]: x=np.arange(6)
In [410]: x.resize(3*3)
In [411]: x.shape
Out[411]: (9,)
In [412]: x.reshape(3,3)
Out[412]: 
array([[0, 1, 2],
       [3, 4, 5],
       [0, 0, 0]])

The np.resize replicates values. np.pad is also handy for adding 0s, though it might overkill.

With np.arange(6) we can pad before or after reshape. With np.arange(5) we have to stick with before, because the padding will be irregular.

In [409]: x=np.arange(6)
In [410]: x.resize(3*3)
In [411]: x.shape
Out[411]: (9,)
In [412]: x.reshape(3,3)
Out[412]: 
array([[0, 1, 2],
       [3, 4, 5],
       [0, 0, 0]])

In any case there isn't one function that does all of this in one call - at least not that I know of. This isn't a common enough operation.