Vectorized conversion of decimal integer array to binary array in numpy

Question

I'm trying to convert an array of integers into their binary representations in python. I know that native python has a function called bin that does this. Numpy also has a similar function: numpy.binary_repr.

The problem is that none of these are vectorized approaches, as in, they only take one single value at a time. So, in order for me to convert a whole array of inputs, I have to use a for-loop and call these functions multiple times, which isn't very efficient.

Is there any way to perform this conversion without for-loops? Are there any vectorized forms of these functions? I've tried numpy.apply_along_axis but no luck. I've also tried using np.fromiter and map and it was also a no go.

I know similar questions have been asked a few other times (like here), but none of the answers given are actually vectorized.

Pointing me into any direction would be greatly appreciated!

Thanks =)

Answer 1

The easiest way is to use binary_repr with vectorize, it will preserve the original array shape, e.g.:

binary_repr_v = np.vectorize(np.binary_repr)
x = np.arange(-9, 21).reshape(3, 2, 5)
print(x)
print()
print(binary_repr_v(x, 8))

The output:

[[[-9 -8 -7 -6 -5]
  [-4 -3 -2 -1  0]]

 [[ 1  2  3  4  5]
  [ 6  7  8  9 10]]

 [[11 12 13 14 15]
  [16 17 18 19 20]]]

[[['11110111' '11111000' '11111001' '11111010' '11111011']
  ['11111100' '11111101' '11111110' '11111111' '00000000']]

 [['00000001' '00000010' '00000011' '00000100' '00000101']
  ['00000110' '00000111' '00001000' '00001001' '00001010']]

 [['00001011' '00001100' '00001101' '00001110' '00001111']
  ['00010000' '00010001' '00010010' '00010011' '00010100']]]