How To Repeatedly Concatenate Numpy Arrays?

I am trying to see if I can concatenate an empty array with fixed size with other arrays with the same size: import numpy as np final_array = np.empty([3, 5]) >>>final_ar

Solution 1:

np.vstack may suffice for your purpose:

array_1 = np.array([1, 1, 1])
array_2 = np.array([2, 2, 2])
array_3 = np.array([3, 3, 3])

lst = [array_1, array_2, array_3]

np.vstack(lst)

# array([[1, 1, 1],
#        [2, 2, 2],
#        [3, 3, 3]])

An alternative is itertools.chain:

from itertools import chain

np.fromiter(chain(*lst), dtype=np.int8).reshape((len(lst), len(lst[0])))

# array([[1, 1, 1],
#        [2, 2, 2],
#        [3, 3, 3]], dtype=int8)

Solution 2:

In [1]: final_array = np.empty([3, 5])
In [2]: final_array
Out[2]: 
array([[-1.52691690e-041,  6.36598743e-314,  2.01589600e-312,
         2.41907520e-312,  1.90979622e-313],
       [ 6.79038654e-313,  6.79038653e-313,  3.18299369e-313,
         2.14321575e-312,  2.37663529e-312],
       [ 4.45619117e-313,  1.93101617e-312,  8.70018274e-313,
        -1.52680304e-041,  8.70018275e-313]])

This is a (3,5) shaped array with random, 'unfilled' values.

In [3]: np.concatenate([final_array, np.array([1,1,1])])    
ValueError: all the input arrays must have same number of dimensions

This doesn't work because final_array is 2d, and the other is 1d.

In [4]: np.concatenate([final_array, np.array([[1,1,1]])])

ValueError: all the input array dimensions except for the concatenation axis must match exactly

Now the 2nd is (1,3), which doesn't match on the 2nd dimension with (3,5).

In [5]: np.concatenate([final_array, np.array([[1,1,1,1,1]])])
Out[5]: 
array([[-1.52691690e-041,  6.36598743e-314,  2.01589600e-312,
         2.41907520e-312,  1.90979622e-313],
       [ 6.79038654e-313,  6.79038653e-313,  3.18299369e-313,
         2.14321575e-312,  2.37663529e-312],
       [ 4.45619117e-313,  1.93101617e-312,  8.70018274e-313,
        -1.52680304e-041,  8.70018275e-313],
       [ 1.00000000e+000,  1.00000000e+000,  1.00000000e+000,
         1.00000000e+000,  1.00000000e+000]])
In [6]: _.shape
Out[6]: (4, 5)

This works, adding a new row to the original (3,5). But the original random values are still there.

It is better to build a list of arrays, and do one concatenate

In [7]: alist = []      # not at all like `np.empty`
In [8]: for i in range(3):
   ...:     alist.append(np.ones((3,),int)*(i+1))
   ...:     
In [9]: alist
Out[9]: [array([1, 1, 1]), array([2, 2, 2]), array([3, 3, 3])]

In [10]: np.array(alist)
Out[10]: 
array([[1, 1, 1],
       [2, 2, 2],
       [3, 3, 3]])

In [11]: np.stack(alist)   # equivalent
Out[11]: 
array([[1, 1, 1],
       [2, 2, 2],
       [3, 3, 3]])

In [12]: np.vstack(alist)
Out[12]: 
array([[1, 1, 1],
       [2, 2, 2],
       [3, 3, 3]])

But concatenate joins them on the 1 existing dimension:

In [13]: np.concatenate(alist)
Out[13]: array([1, 1, 1, 2, 2, 2, 3, 3, 3])