Assigning issue with numpy structured arrays

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]]

10 questions - 0 accepted answers. Just saying...

JoshAdel
– JoshAdel

2011-09-27 23:04:18 +00:00
Commented Sep 27, 2011 at 23:04 — JoshAdel
– JoshAdel, Commented Sep 27, 2011 at 23:04

JoshAdel · Accepted Answer · 2011-09-27 23:31:33Z

2

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.

edited Sep 27, 2011 at 23:31

answered Sep 27, 2011 at 23:19

JoshAdel

69.1k27 gold badges146 silver badges146 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

JustInTime Over a year ago

Thanks for the response, I would not have figured that alone, and yes this seems to be a strange behavior as you said.

eumiro Over a year ago

@JustInTime - then accept this answer (and also those in other questions) please.

Chris Over a year ago

I've just run into this problem and discovered this workaround, too. The basic issue seems to be that doing it one way results in a copy, the other a view. It's not clear to me how you are supposed to know, though, other than trying. I found some more info here: stackoverflow.com/questions/3058602/…

Collectives™ on Stack Overflow

Assigning issue with numpy structured arrays

1 Answer 1

3 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

3 Comments

Your Answer

Sign up or log in

Post as a guest

Related