Question

I was trying this simple line of assigning codes to a structured array in numpy, I am not quiet sure, but something wrong happens when I assign a matrix to a sub_array in a structured array I created as follows:

new_type = np.dtype('a3,(2,2)u2')
x = np.zeros(5,dtype=new_type)
x[1]['f1'] = np.array([[1,1],[1,1]])
print x
Out[143]: 
array([('', [[0, 0], [0, 0]]), ('', [[1, 0], [0, 0]]),
   ('', [[0, 0], [0, 0]]), ('', [[0, 0], [0, 0]]),
   ('', [[0, 0], [0, 0]])], 
  dtype=[('f0', '|S3'), ('f1', '<u2', (2, 2))])

Shouldn’t the second field of the subarray equals at this stage

[[1,1],[1,1]]

Answer 1

I think you want to set things slightly differently. Try:

x['f1'][1] = np.array([[1,1],[1,1]])

which results in:

In [43]: x = np.zeros(5,dtype=new_type)

In [44]: x['f1'][1] = np.array([[1,1],[1,1]])

In [45]: x
Out[45]: 
array([('', [[0, 0], [0, 0]]), ('', [[1, 1], [1, 1]]),
       ('', [[0, 0], [0, 0]]), ('', [[0, 0], [0, 0]]),
       ('', [[0, 0], [0, 0]])], 
      dtype=[('f0', '|S3'), ('f1', '<u2', (2, 2))])

This is not to say that this isn’t strange behavior though since both x['f1'][1] and x[1]['f1'] print the same results, but clearly are different:

In [51]: x['f1'][1]
Out[51]: 
array([[1, 1],
       [1, 1]], dtype=uint16)

In [52]: x[1]['f1'] 
Out[52]: 
array([[1, 1],
       [1, 1]], dtype=uint16)

In [53]: x[1]['f1'] = 2

In [54]: x
Out[54]: 
array([('', [[0, 0], [0, 0]]), ('', [[2, 1], [1, 1]]),
       ('', [[0, 0], [0, 0]]), ('', [[0, 0], [0, 0]]),
       ('', [[0, 0], [0, 0]])], 
      dtype=[('f0', '|S3'), ('f1', '<u2', (2, 2))])

In [55]: x['f1'][1] = 3

In [56]: x
Out[56]: 
array([('', [[0, 0], [0, 0]]), ('', [[3, 3], [3, 3]]),
       ('', [[0, 0], [0, 0]]), ('', [[0, 0], [0, 0]]),
       ('', [[0, 0], [0, 0]])], 
      dtype=[('f0', '|S3'), ('f1', '<u2', (2, 2))])

I’d have to think about it a bit more to figure out exactly what is going on.