numpy create a boolean array of False

Question

I try to initialize a Boolean array with False using np.empty,

init_arr = empty(5)
init_arr.fill(False)

but I got

array([0., 0., 0., 0., 0.])

then I create a pd.Series,

dummy_ser = pd.Series(data=init_arr)

which makes it

I tried to initialize a boolean series with various length with False as its initialized values. I am wondering whats the best way to do this.

I am using Python 3.5.2 and Numpy 1.15.4.

Use `np.zeros(5, dtype=bool)`. – jdehesa Jul 30 '19 at 08:45 — jdehesa, Jul 30 '19 at 08:45

score 3 · Accepted Answer · answered Jul 30 '19 at 09:00

This is because the empty() function creates an array of floats:

In [14]: np.empty(5).dtype
Out[14]: dtype('float64')

There are many ways to solve this, supplying dtype=bool to empty() being one of them.

But I would create an array of False like this:

In [17]: np.full(5, False)
Out[17]: array([False, False, False, False, False], dtype=bool)

score 1 · Answer 2 · answered Jul 30 '19 at 08:45

1

you could do :

np.zeros(5).astype(bool)

answered Jul 30 '19 at 08:45

Ayoub ZAROU

This will needlessly create an `int` array first, and cast it to `bool` later, wasting space in the meantime. – Nils Werner Jul 30 '19 at 08:52
1

yeah I agree, I think the efficient way to do so is doing `np.zeros(5, dtype = bool)` as @jdehesa suggested – Ayoub ZAROU Jul 30 '19 at 08:53

score 1 · Answer 3 · answered Jul 30 '19 at 08:49

1

This will work:

>>> import numpy as np
>>> np.broadcast_to(False, (2, 2))
array([[ False,  False],
   [ False,  False]], dtype=bool)

answered Jul 30 '19 at 08:49

Aravind

3 Answers3