The objective is to convert from float64 to double.
However, using the astype(np.double) does not changed the array type.
numpy.random.seed(0)
arr=np.random.rand(32,30,30)
arr1=arr.astype(np.double)
print(f'arr:{arr.dtype} and arr1: {arr1.dtype}')
arr:float64 and arr1: float64
>Solution :
import numpy as np
np.double is np.float64 # returns True
As you can see they are virtually the same
You can see that in the docs here where it is mentioned that np.float64 is just an alias for np.double