Working with nested lists in Python

Question

list_ = [(1, 2), (3, 4)]

What is the Pythonic way of taking sum of ordered pairs from inner tuples and multiplying the sums? For the above example:

(1 + 3) * (2 + 4) = 24

• erip: What happens if there are 3 tuples: list_ = [(1,2), (3,4), (5,6)]?
• blackened: @erip (1+3+5) * (2+4+6) = 108

Answer 1

For example:

import operator as op
import functools
functools.reduce(op.mul, (sum(x) for x in zip(*list_)))

works for any length of the initial array as well as of the inner tuples.

Another solution using numpy:

import numpy as np
np.array(list_).sum(0).prod()

• erip: @PeterWood Sure, why not?
• MB-F: @erip because reduce was removed from python 3: docs.python.org/3.0/whatsnew/3.0.html#builtins
• Nelewout: @eumiro just to add to the available options: rather than importing operator.mul, lambda x, y: x * y works fine also. Although really there is nothing wrong with op.mul to begin with :).
• eumiro: @kazemakase: thank you for the reduce info - I have updated the answer.
• eumiro: @blackened: how about functools.reduce(op.mul, (sum(x) for x in list(zip(*list_))[:-1])) ?

Answer 2

If the lists are small as is implied, I feel that using operator and itertools for something like this is applying a sledgehammer to a nut. Likewise numpy. What is wrong with pure Python?

result = 1
for s in [ sum(x) for x in zip( *list_) ]: 
 result *= s

(although it would be a lot nicer if pure Python had a product built-in as well as sum ). Also if you are specifically dealing only with pairs of 2-tuples then any form of iteration is a sledgehammer. Just code

result = (list_[0][0]+list_[1][0] )*( list_[0][1]+list_[1][1])

• jamylak: You could also argue that using zip(*) is a sledgehammer when you could just do it with pure python. Using reduce simply does the same thing your code does but more concise
• nigel222: Yes, reduce is a built-in as well as zip so its just that I personally find this way easier to read.
• nigel222: Actually my previous is wrong. zip is a Python 3 built-in but reduce is not, so my way is Python 2/3 agnostic.