Closed
Description
Taken from this StackOverflow question: Why is (2^31) >> 32 not 0?
A 32-bit right shift applied to a uint32 array with the value 231 in it produces an array with the value 2147483648 instead of 0. Interestingly, when the operation is applied to a scalar with the same value the result is 0.
As discussed in the linked question, this probably has to do with 32-bit shifts of 32-bit values not being well-defined in C. However, this shouldn't condition the behavior of NumPy. Most importantly, the behavior should be consistent between arrays and scalars.
Reproducing code example:
import numpy as np
print(np.array([2 ** 31], dtype=np.uint32) >> 32)
# [2147483648]
print(np.array(2 ** 31, dtype=np.uint32) >> 32)
# 0Error message:
NA
Numpy/Python version information:
1.15.4 3.6.7 |Anaconda, Inc.| (default, Oct 28 2018, 19:44:12) [MSC v.1915 64 bit (AMD64)]
Metadata
Metadata
Assignees
Labels
No labels