compare x ** p vs np.exp(p * np.log(x))

comp_pow.py

#!/usr/bin/env python3

import numpy as np
import timeit

def setup(N):
    x = np.random.rand(N)
    return x

def pow1(x, p):
    return x ** p

def pow2(x, p):
    return np.exp(p * np.log(x))

N = 10000
loop = 10000
p = 0.3

x = setup(N)
assert np.allclose(pow1(x, p), pow2(x, p))

result = timeit.timeit("y1 = pow1(x, p)", setup="x = setup(N)", globals=globals(), number=loop)

print(f"pow1(): {result}")

result = timeit.timeit("y2 = pow2(x, p)", setup="x = setup(N)", globals=globals(), number=loop)

print(f"pow2(): {result}")