4.3 Functions

4.3 Functions#

Functions are essential in programming as they allow you to encapsulate and reuse code efficiently. By breaking down complex problems into smaller, manageable parts, functions enhance code organization and readability. As we already saw you can use built-in functions or import them from specialized modules. Quite often it is also useful for you to define your own functions. In Python, this can be done by using the def keyword followed by a function name:

def my_function():
  print("I'm a custom function!")

my_function() # use the function

I'm a custom function!

Function arguments and returns#

Functions can accept arguments (or parameters) that provide inputs or modify their behaviour. For example, if we want to add two numbers, we can pass both of them as arguments and further return the result to the user:

def add(a, b):
  return a + b

result = add(1, 2)
print(result)

3

Arguments#

Python functions can have two different kinds of arguments: positional arguments, and keyword arguments. Positional arguments on the one hand are defined by their position whithin the round brackets and must be provided for the function to run without giving an error. In the previous example, both a and b are positional arguments and the function would not know what to do if any of them would be missing. Keyword arguments on the other hand are assigned a default value. This means that the function knows what to do even if the user does not explicitly provide a value for that argument.

As an example, let’s create a function that adds random gaussian noise to a given input x:

import random

def add_noise(x, mu=0, sd=1):
  """Adds gaussian noise to the input.

  Parameters
  ----------
  x : number
    The number to add noise to.
  mu : float, optional
    The mean of the gaussian noise distribution.
    Default: 0
  sd : float, optional
    The standard deviation of the noise distribution.
    Default: 1

  Returns
  -------
  float
  """

  noise = random.normalvariate(mu, sd)
  return x + noise

Note: So far, you have mostly seen comments within the code using the # sign. Another way to provide comments is by using triple double quotes """comment""", as shown in the previous code snippet. This approach is useful for providing larger portions of text, such as when documenting your functions. In the add_noise() function, you can see an example where the input and output parameters are clearly described, making it easier to understand how to use the function. This is also the official way of documenting functions and classes in Python, so it is recommended to follow this convention.

If we now call this function by just providing a value for xit will still work as expected by using a mean of 0 and a standard deviation of 1 to calculate and add the noise.

add_noise(5)

4.7464003884233525

If you decide you need different noise with a standard deviation of 5, you can simply add this new value. As positional arguments are optional, their order does not matter. You can provide any keyword argument in any order you like, as long as you provide its name and all positional arguments have been correctly provided before.

add_noise(5, sd=3)

4.81395643684874

You can also specify all arguments (including the positional one) by name (in this case the order of sd and mu doesnt matter)

add_noise(x=5, sd=3, mu=2)

6.789090548546618

or not use any name at all

add_noise(5, 3, 2)

10.043656687530023

However, the last example is differnet compared to the the previous one! This is because if no keywords are provided, Python will only be able to rely on the default order of arguments, which would result in the arguments being interpreted as x=5, mu=2 and sd=3 as defined in the function.

Argument unpacking#

Another feature of Python is the ability to provide a function with an unspecified number of arguments by using *args and **kwargs. While we will not cover this topic here, I want to mention it for completeness. If you are interested or encounter it in practice, feel free to explore this topic on your own (e.g., here).

Contents