Extracting parts of array repeatedly

Question

I have an array:

[a1, b1, c1, d1, a2, b2, c2, d2, a3, b3, etc]

How do I select every b and c without a for loop? Can I use slicing or anything else?

The resulting array should look like this:

[b1, c1, b2, c2, b3, c3, etc]

Answer 1

You can use masking. Provided your input array has always a length in a multiple of 4, you can create a mask of pattern False, True, True, False. I am taking an input of strings for example.

arr = np.array(['a1', 'b1', 'c1', 'd1', 'a2', 'b2', 'c2', 'd2', 'a3', 'b3', 'c3', 'd3'], dtype='str')
mask = [False, True, True, False]*int(len(arr)/4)

print (arr[mask])
# array(['b1', 'c1', 'b2', 'c2', 'b3', 'c3'])

Answer 2

You can select the data you want with numpy.lib.stride_tricks.as_strided:

import numpy as np
from numpy.lib.stride_tricks import as_strided

data = np.array(['a0', 'b0', 'c0', 'd0', 'a1', 'b1', 'c1', 'd1', 'a2', 'b2', 'c2', 'd2'])
s = data.strides[0]
# No data is copied
data2 = as_strided(data[1:], shape=(data.size // 4, 2), strides=(4 * s, s), writeable=False)
print(data2)
# [['b0' 'c0']
#  ['b1' 'c1']
#  ['b2' 'c2']]
data3 = data2.ravel()  # This causes a copy
print(data3)
#['b0' 'c0' 'b1' 'c1' 'b2' 'c2']