pd.Series.rpow
¶
pandas.Series.rpow(other, level=None, fill_value=None, axis=0)
Supported Arguments¶
argument | datatypes |
---|---|
other |
|
fill_value |
numeric scalar |
Note
Series.rpow
is only supported on Series of numeric data.
Example Usage¶
>>> @bodo.jit
... def f(S, other):
... return S.rpow(other)
>>> S = pd.Series(np.arange(1, 1001))
>>> other = pd.Series(reversed(np.arange(1, 1001)))
>>> f(S, other)
0 1000
1 998001
2 994011992
3 988053892081
4 980159361278976
...
995 3767675092665006833
996 0
997 -5459658280481875879
998 0
999 1
Length: 1000, dtype: int64