matman9
matman9

Reputation: 490

find tuple in ndarray of tuples and return index of searched tuple

I have a dictionary with a coordinates entry as an ndarray of tuples,

import numpy as np

data = np.arange(0, 18)
coord = [(i, i, i) for i in data]
arr = np.empty(18, dtype=object)
arr[:] = coord
arr = arr.reshape(3, 6)

d = dict()
d.update({'coord': arr}) 

I'd like to query the dictionary with a coordinate and be returned it's index within the array.

When I try to find the index with np.where it returns no match.

np.where(d['coord'] == (0, 0, 0))
(array([], dtype=int64),)

This would ideally return the index (0, 0).

When providing an index to the dictionary entry and tuple values it returns True, so the tuple exists at an index.

d['coord'][0,0] == (0, 0, 0)
True

Can I get the index this way??

Thanks.

Upvotes: 2

Views: 428

Answers (1)

hpaulj
hpaulj

Reputation: 231385

The problem is with the == test of an object array with a tuple.

In [346]: d['coord']                                                            
Out[346]: 
array([[(0, 0, 0), (1, 1, 1), (2, 2, 2), (3, 3, 3), (4, 4, 4), (5, 5, 5)],
       [(6, 6, 6), (7, 7, 7), (8, 8, 8), (9, 9, 9), (10, 10, 10),
        (11, 11, 11)],
       [(12, 12, 12), (13, 13, 13), (14, 14, 14), (15, 15, 15),
        (16, 16, 16), (17, 17, 17)]], dtype=object)
In [347]: d['coord']==(0, 0, 0)                                                 
/usr/local/bin/ipython3:1: DeprecationWarning: elementwise == comparison failed; this will raise an error in the future.
  #!/usr/bin/python3
Out[347]: False

The solution is to compare one object array with another object array:

In [348]: x=np.array(None); x[()]=(0,0,0)                                       
In [349]: x                                                                     
Out[349]: array((0, 0, 0), dtype=object)
In [350]: d['coord']==x                                                         
Out[350]: 
array([[ True, False, False, False, False, False],
       [False, False, False, False, False, False],
       [False, False, False, False, False, False]])

Upvotes: 2

Related Questions