Question

There is no array type in python, but to emulate it we can use lists. I want to have 2d array-like structure filled in with zeros. My question is: what is the difference, if any, in this two expressions:

zeros = [[0 for i in xrange(M)] for j in xrange(M)]

and

zeros = [[0]*M]*N

Will zeros be same? which one is better to use by means of speed and readability?

Answer 1

You should use numpy.zeros. If that isn’t an option, you want the first version. In the second version, if you change one value, it will be changed elsewhere in the list — e.g.:

>>> a = [[0]*3]*3
>>> a
[[0, 0, 0], [0, 0, 0], [0, 0, 0]]
>>> a[0][0] = 1
>>> a
[[1, 0, 0], [1, 0, 0], [1, 0, 0]]

This is because (as you read the expression from the inside out), you create a list of 10 zeros. You then create a list of 10 references to that initial list of 10 zeros.

---

Note that this will also work and it avoids the nested list comprehension:

zeros = [ [0]*M for _ in range(N) ]

If numpy isn’t on the table, this is the form I would use.

(Use xrange if you’re still stuck in the python2.x dark ages :).)