Efficiently initialize 2D array of size n*m in Python 3?

Question

I am using Python 3 and am trying to create 2D integer arrays of a known and fixed size. How can I do this?

The arrays I have in mind are the Python-provided array.array, although I am not averse to using a library like NumPy.

I know this can be done via lists, but I want to use an array since they are much more space efficient. Right now, I am initializing lists in this way, but I want to replace it with an array:

my_list = [[ 0 for _ in range(height) ] for _ in range(.width)]

For example, in C, I would write

int my_array[ width ][ height ];

How can I do this in Python

So, what exactly do you mean by array? numpy.ndarray? Or array.array? — juanpa.arrivillaga
– juanpa.arrivillaga, Commented Jul 30, 2017 at 22:33
Please post a minimal example. What is the actual data type? docs.python.org/3/library/array.html? — wwii
– wwii, Commented Jul 30, 2017 at 22:33
An array.array. I am not averse to using NumPy. I'll update my original question; thank you. — lynn
– lynn, Commented Jul 30, 2017 at 22:36

Jonas Adler · Accepted Answer · 2017-07-30 22:43:16Z

8

You can use numpy:

import numpy as np
my_array = np.zeros([height, width])

For example:

>>> np.zeros([3, 5])
array([[ 0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  0.],
       [ 0.,  0.,  0.,  0.,  0.]])

Note that if you specifically need integers, you should use the dtype argument

>>> np.zeros([3, 5], dtype=int)
array([[0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0],
       [0, 0, 0, 0, 0]])

answered Jul 30, 2017 at 22:43

Jonas Adler

10.9k5 gold badges50 silver badges76 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

juanpa.arrivillaga Over a year ago

@tpepin96 yes, they are. Although, given the dimensions you are talking about, even an list of lists wouldn't be an issue on modern hardware.

juanpa.arrivillaga Over a year ago

@tpepin96 They are primitive arrays, so each element is whatever size the dtype implies...

o11c Over a year ago

@tpepin96 scipy.sparse exists for cases that are actually large (1000x1000 is not)

jp48 · Accepted Answer · 2017-07-31 02:52:03Z

The numpy equivalent of the code you posted is:

import numpy as np
my_array = np.empty([height, width])

empty reserves uninitialized space, so it is slightly faster than zeros, because zeros has to make sure that everything is initially set to 0.

Note that, like in C/C++, you should write values before reading them, because they initially "contain garbage".

Docs:

https://docs.scipy.org/doc/numpy/reference/generated/numpy.empty.html

https://docs.scipy.org/doc/numpy/reference/generated/numpy.zeros.html

Collectives™ on Stack Overflow

Efficiently initialize 2D array of size n*m in Python 3?

2 Answers 2

3 Comments

Comments

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Related