Sum of list of lists; returns sum list
You could try this:
In [9]: l = [[3,7,2],[1,4,5],[9,8,7]]
In [10]: [sum(i) for i in zip(*l)]
Out[10]: [13, 19, 14]
This uses a combination of zip
and *
to unpack the list and then zip the items according to their index. You then use a list comprehension to iterate through the groups of similar indices, summing them and returning in their 'original' position.
To hopefully make it a bit more clear, here is what happens when you iterate through zip(*l)
:
In [13]: for i in zip(*l):
....: print i
....:
....:
(3, 1, 9)
(7, 4, 8)
(2, 5, 7)
In the case of lists that are of unequal length, you can use itertools.izip_longest
with a fillvalue
of 0
- this basically fills missing indices with 0
, allowing you to sum all 'columns':
In [1]: import itertools
In [2]: l = [[3,7,2],[1,4],[9,8,7,10]]
In [3]: [sum(i) for i in itertools.izip_longest(*l, fillvalue=0)]
Out[3]: [13, 19, 9, 10]
In this case, here is what iterating over izip_longest
would look like:
In [4]: for i in itertools.izip_longest(*l, fillvalue=0):
...: print i
...:
(3, 1, 9)
(7, 4, 8)
(2, 0, 7)
(0, 0, 10)
For any matrix (or other ambitious numerical) operations I would recommend looking into NumPy.
The sample for solving the sum of an array along the axis shown in your question would be:
>>> from numpy import array
>>> data = array([[3,7,2],
... [1,4,5],
... [9,8,7]])
>>> from numpy import sum
>>> sum(data, 0)
array([13, 19, 14])
Here's numpy's documentation for its sum function: http://docs.scipy.org/doc/numpy/reference/generated/numpy.sum.html#numpy.sum
Especially the second argument is interesting as it allows easily specify what should be summed up: all elements or only a specific axis of a potentially n-dimensional array(like).