CODEWITHSUNDEEP
pythonnumpymethodsproperties
emehex
emehex

Reputation: 10548

Numpy functions/methods without brackets explanation

I'm just trying to wrap my head around the syntax here:

# some data
import numpy as np
x = linspace(0, 1, 100)

x.sum()    # works with brackets
> 50.0

x.shape    # works without brackets
> (100,)

x.shape()  # fails with brackets
> TypeError

Why is it that some methods/functions have brackets while others don't?

Upvotes: 0

Views: 803

Answers (1)

MSeifert
MSeifert

Reputation: 152725

When you access an attribute of your instance you really access a descriptor. There are three common cases:

  • The descriptor returns a function like x.sum, that returns a bound function. Bound because the first argument to that function will be x. You obviously need to call () that function to give you an result.

  • There are class and instance attributes, that simply return a value either saved in x.__dict__ or x.__class__.__dict__. You don't need to call anything to get these. However x.shape isn't one of those!

  • There are properties, when you access these it will implicitly call a method of your class. In general properties, say x.something, are equivalent to x.__class__.something.fget(x). Seems weird but works ... sort of. The important thing here is that you normally don't need to call it directly () because you can just do x.something.

But that's technical, maybe the simpler explanation is better to remember:

  • sum is a method. Methods should be called. To call a function or method you need the ().
  • shape is a property (but could also be an attribute) and you don't need to call those! It could return a function as well but in this case it returns a tuple and you can't call a tuple instance.

Upvotes: 3

Related Questions