A Generator Expression is doing basically the same thing as a List Comprehension does, but the GE does it lazily. The difference is quite similar to the difference between range and xrange.

A List Comprehension, just like the plain range function, executes immediately and returns a list.

A Generator Expression, just like xrange returns an object that can be iterated over.

The comparision is not perfect though, because in an object returned by the generator expression, we cannot access an element by index.

The difference between the two kinds of expressions is that the List comprehension is enclosed in square brackets [] while the Generator expression is enclosed in plain parentheses (). 

l = [n*2 for n in range(1000)] # List comprehension
g = (n*2 for n in range(1000))  # Generator expression

Type

The types of the resulting values are list and generator respectively: 

print(type(l))  # 'list'
print(type(g))  # 'generator'

Size in memory

The size of the objects is 9032 bytes (the list), and 80 bytes (the generator): 

print(sys.getsizeof(l))  # 9032
print(sys.getsizeof(g))  # 80

Access by Index

We can access the elements of the list, but if we try to access the elements of the generator we get a TypeError: 

print(l[4])   # 8
print(g[4])   # TypeError: 'generator' object has no attribute '__getitem__'

Loop over

Finally, but most importantly, we can iterate over either of them: 

for v in l:
    pass
for v in g:
    pass

The full example

examples/python/generator_expression.py

#!/usr/bin/env python
from __future__ import print_function
import sys

l = [n*2 for n in range(1000)] # List comprehension
g = (n*2 for n in range(1000)) # Generator expression

print(type(l))  # <type 'list'>
print(type(g))  # <type 'generator'>

print(sys.getsizeof(l))  # 9032
print(sys.getsizeof(g))  # 80

print(l[4])   # 8
#print(g[4])   # TypeError: 'generator' object has no attribute '__getitem__'

for v in l:
    pass
for v in g:
    pass