Say I have
>>> arr = np.array([True, True, False], dtype=bool)
Is it possible to call something like
>>> arr.flip_boolean_array_by_index(2)
[True, True, True]
Add a comment
|
2 Answers
You can use the bitwise-negation operator ~, or bitwise-xor (^) with 1.
arr[idx] = ~arr[idx]
or
arr[idx] ^= 1
idx can be an index, a slice, a "fancy" index, etc.
7 Comments
user2357112
^= True might be cleaner.
user3557216
I know how to do it in multiple steps. I want to know if there's a utility function that returns it in-place, it would read very nicely.
shx2
@user2357112 I indeed contemplated that. On one hand it is cleaner because we are doing a boolean operation. on the other hand, it is way more common to think about xor'ing bits than xor'ing (logical) boolean values...
user2357112
@user3557216: There is such a function. I dunno if it reads very nicely, though, and it's a fairly new feature:
numpy.logical_not.at(arr, 2).
Joran Beasley
or you could just make it nice and easy
arr[idx] = not arr[idx] ... no need for binary negation ... |
I found that ^= 1 on a dytpe=numpy.bool_ array gave a UFuncTypeError (can't cast from dtype int64 to dtype bool). However, ^= True works, e.g. arr[idx] ^= True. This was in numpy 1.26.3.
