BUG: Integers raised to negative powers gives wrong results

In current master:

np.array(1, dtype=int) ** -2
Out[5]: 0

In 1.11 and earlier:

In [3]: np.array(1, dtype=int) ** -2
Out[3]: 1

(Found in astropy testing; astropy/astropy#4950)

Duplicate of #7648 – this can probably be closed.

Oops, sorry. Will close (though my example is even worse...)

Actually, you're right – this looks like a regression in master that might be unrelated.

Navigation Menu