Chained list comprehension in python

Question

this = rand(100,3,4)
for i in range(0,100):
    for j in range(0,3):
        for k in range(0,4):
            if rand()<0.5:
                this[i,j,k]=0

Where rand is numpy.random.rand

Can the above be written in chained list comprehension?

Aim: to assign value 0 to each term in "this" (3D matrix) with certain probability(0.5)

The output is not a list, so not sure why you are asking for a list comprehension? Are you asking is there a more pythonic way to refactor your nested for loops? — Burhan Khalid, Jun 02 '15 at 04:49
@Burhan : yes. Can the same thing be done using a chained list comprehension? Without nested loops? — Abhishek Tripathi, Jun 02 '15 at 05:01
The result of a list comprehension is a _list_; your code is not generating a list (it is instead modifying the matrix) so not sure why you are looking for a list comprehension. — Burhan Khalid, Jun 02 '15 at 05:11
matrix can be thought of as lists in list, right? But I am more comfortable with your solution. Just that comprehensions run faster. That's why I was into list comprehension. using `numpy.where` should be fast enough. :) — Abhishek Tripathi, Jun 02 '15 at 05:14
I should have been more explicit - a list comprehension expression evaluates to a _new list_; it is not designed for modifying lists in-place (you should avoid doing that anyway - its a great source of bugs). `numpy.where` would probably be faster if not equivalent to a comprehension - noting that you would be building a list, and then discarding it if you were to implement your loop as a comprehension. — Burhan Khalid, Jun 02 '15 at 05:28

score 3 · Accepted Answer · edited Jun 02 '15 at 04:55

3

It seems to me that you're better off using builtin numpy routines such as numpy.where...

>>> import numpy
>>> import numpy.random
>>> x = numpy.random.rand(100, 3, 4)
>>> mask = numpy.random.rand(*x.shape)
>>> result = numpy.where(mask < 0.5, 0, x)

edited Jun 02 '15 at 04:55

Burhan Khalid

169,990
18
245
284

answered Jun 02 '15 at 04:52

mgilson

300,191
65
633
696

Could you explain how does that `rand(*x.shape)` works? What does that * mean? – Abhishek Tripathi Jun 02 '15 at 05:07
@AbhishekTripathi See [Expanding tuples into arguments](http://stackoverflow.com/questions/1993727/expanding-tuples-into-arguments) – Burhan Khalid Jun 02 '15 at 05:12

Chained list comprehension in python

1 Answers1