如何使用NumPy将数字数组求幂？

Question

NumPy has log , log2 , and log10 methods which can perform vectorized log base e / 2 / 10 (respectively). However, for doing the inverse operation (exponentiation), I only see exp . Why is there no exp2 / exp10 / etc?

I've tried using np.power(10, nums) , but it won't let me raise to a negative power. 10 ** nums does not work either.

Answer 1

It should work fine with 10 ** nums provided you use the float dtype. Otherwise it will create an integer array:

>>> a = numpy.array([-1, 0, 1, 2, 3], dtype=int)
>>> 2 ** a
array([0, 1, 2, 4, 8])
>>> 10 ** a
array([   0,    1,   10,  100, 1000])
>>> a = numpy.array([-1, 0, 1, 2, 3], dtype=float)
>>> 10 ** a
array([  1.00000000e-01,   1.00000000e+00,   1.00000000e+01,
         1.00000000e+02,   1.00000000e+03])

You can also coerce to float by using 10.0 instead of 10 :

>>> a = numpy.array([-1, 0, 1, 2, 3], dtype=int)
>>> 10.0 ** a
array([  1.00000000e-01,   1.00000000e+00,   1.00000000e+01,
         1.00000000e+02,   1.00000000e+03])