How to Create Your Own Universal Function (ufunc) in NumPy
In NumPy, ufuncs (universal functions) are used to perform element-wise
operations on arrays efficiently.
But did you know you can create your own ufunc using Python
functions?
Here’s how you can do it:
Steps:
Define a regular Python function (just like any other function).
Use np.frompyfunc() to convert it into a ufunc.
np.frompyfunc() Syntax:
numpy.frompyfunc(function, inputs, outputs)
- function: Your custom function name.
- inputs: Number of input arguments (usually the number of arrays you want to process).
- outputs: Number of outputs your function returns.
Program:
Create a Custom ufunc for Addition
import numpy as np
# Step 1: Define a simple function
def myadd(x, y):
return x + y
# Step 2: Convert it into a ufunc
myadd = np.frompyfunc(myadd, 2, 1)
# Step 3: Use it with NumPy arrays
result = myadd([1, 2, 3, 4], [5, 6, 7, 8])
print(result)
Output:
[6 8 10 12]
How to Check if a Function is a NumPy ufunc
In NumPy, ufuncs (universal functions) are functions that operate
element-wise on arrays, such as np.add or np.subtract.
To check whether a function is a ufunc, you can inspect its type.
Program:
Check if a function is a ufunc:
import numpy as np
print(type(np.add))
Output:
<class 'numpy.ufunc'>
This confirms that np.add is a ufunc.
Program:
Check a function that is not a ufunc:
import numpy as np
print(type(np.concatenate))
Output:
<class 'function'>
np.concatenate is a regular function, not a ufunc.
Program:
Check an invalid function (this will raise an error):
import numpy as np
print(type(np.blahblah))
Output:
AttributeError: module 'numpy' has no attribute 'blahblah'
Program:
You can use an if statement to check whether a function is a ufunc:
import numpy as np
if type(np.add) == np.ufunc:
print("add is a
ufunc")
else:
print("add is not a
ufunc")
Output:
add is a ufunc
More topic in Numpy
