numpy.frompyfunc() in Python

numpy.frompyfunc(func, nin, nout) function allows to create an arbitrary Python function as Numpy ufunc (universal function).

Parameters:
func: [A python function object ] An arbitrary python function 
nin: [int] Number of input arguments to that function. 
nout: [int] Number of objects returned by that function.
Return: A Numpy universal function object.

For example, abs_value = numpy.frompyfunc(abs, 1, 1) will create a ufunc that will return the absolute values of array elements.

Code #1:

Python3

# Python code to demonstrate the
# use of numpy.frompyfunc
import numpy as np

# create an array of numbers
a = np.array([34, 67, 89, 15, 33, 27])

# python str function as ufunc
string_generator = np.frompyfunc(str, 1, 1)

print("Original array-", a)
print("After conversion to string-", string_generator(a))

Output:

Original array- [34 67 89 15 33 27]
After conversion to string- ['34' '67' '89' '15' '33' '27']

Code #2:

Python3

# Python code to demonstrate
# user-defined function as ufunc
import numpy as np

# create an array of numbers
a = np.array([345, 122, 454, 232, 334, 56, 66])

# user-defined function to check
# whether a no. is palindrome or not
def fun(x):
    s = str(x)
    return s[::-1]== s

# 'check_palindrome' as universal function
check_palindrome = np.frompyfunc(fun, 1, 1)
print("Original array-", a)
print("Checking of number as palindrome-",
                      check_palindrome(a))

Output:

Original array- [345 122 454 232 334  56  66]
Checking of number as palindrome- [False False True True False False True]

Note: This custom ufunc created using frompyfunc always accept a ndarray as an input argument and also return a ndarray object as output.