How to initialize a two-dimensional array in Python?

Each Answer to this Q is separated by one/two green lines.

I’m beginning python and I’m trying to use a two-dimensional list, that I initially fill up with the same variable in every place. I came up with this:

def initialize_twodlist(foo):
    twod_list = []
    new = []
    for i in range (0, 10):
        for j in range (0, 10):
            new.append(foo)
        twod_list.append(new)
        new = []

It gives the desired result, but feels like a workaround. Is there an easier/shorter/more elegant way to do this?

A pattern that often came up in Python was

bar = []
for item in some_iterable:
    bar.append(SOME EXPRESSION)

which helped motivate the introduction of list comprehensions, which convert that snippet to

bar = [SOME_EXPRESSION for item in some_iterable]

which is shorter and sometimes clearer. Usually, you get in the habit of recognizing these and often replacing loops with comprehensions.

Your code follows this pattern twice

twod_list = []                                       \                      
for i in range (0, 10):                               \
    new = []                  \ can be replaced        } this too
    for j in range (0, 10):    } with a list          /
        new.append(foo)       / comprehension        /
    twod_list.append(new)                           /

Don’t use [[v]*n]*n, it is a trap!

>>> 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]]

but

    t = [ [0]*3 for i in range(3)]

works great.

You can use a list comprehension:

x = [[foo for i in range(10)] for j in range(10)]
# x is now a 10x10 array of 'foo' (which can depend on i and j if you want)

This way is faster than the nested list comprehensions

[x[:] for x in [[foo] * 10] * 10]    # for immutable foo!

Here are some python3 timings, for small and large lists

$python3 -m timeit '[x[:] for x in [[1] * 10] * 10]'
1000000 loops, best of 3: 1.55 usec per loop

$ python3 -m timeit '[[1 for i in range(10)] for j in range(10)]'
100000 loops, best of 3: 6.44 usec per loop

$ python3 -m timeit '[x[:] for x in [[1] * 1000] * 1000]'
100 loops, best of 3: 5.5 msec per loop

$ python3 -m timeit '[[1 for i in range(1000)] for j in range(1000)]'
10 loops, best of 3: 27 msec per loop

Explanation:

[[foo]*10]*10 creates a list of the same object repeated 10 times. You can’t just use this, because modifying one element will modify that same element in each row!

x[:] is equivalent to list(X) but is a bit more efficient since it avoids the name lookup. Either way, it creates a shallow copy of each row, so now all the elements are independent.

All the elements are the same foo object though, so if foo is mutable, you can’t use this scheme., you’d have to use

import copy
[[copy.deepcopy(foo) for x in range(10)] for y in range(10)]

or assuming a class (or function) Foo that returns foos

[[Foo() for x in range(10)] for y in range(10)]

To initialize a two-dimensional array in Python:

a = [[0 for x in range(columns)] for y in range(rows)]

[[foo for x in xrange(10)] for y in xrange(10)]

Usually when you want multidimensional arrays you don’t want a list of lists, but rather a numpy array or possibly a dict.

For example, with numpy you would do something like

import numpy
a = numpy.empty((10, 10))
a.fill(foo)

You can do just this:

[[element] * numcols] * numrows

For example:

>>> [['a'] *3] * 2
[['a', 'a', 'a'], ['a', 'a', 'a']]

But this has a undesired side effect:

>>> b = [['a']*3]*3
>>> b
[['a', 'a', 'a'], ['a', 'a', 'a'], ['a', 'a', 'a']]
>>> b[1][1]
'a'
>>> b[1][1] = 'b'
>>> b
[['a', 'b', 'a'], ['a', 'b', 'a'], ['a', 'b', 'a']]

twod_list = [[foo for _ in range(m)] for _ in range(n)]

for n is number of rows, and m is the number of column, and foo is the value.

If it’s a sparsely-populated array, you might be better off using a dictionary keyed with a tuple:

dict = {}
key = (a,b)
dict[key] = value
...

Code:

num_rows, num_cols = 4, 2
initial_val = 0
matrix = [[initial_val] * num_cols for _ in range(num_rows)]
print(matrix) 
# [[0, 0], [0, 0], [0, 0], [0, 0]]

initial_val must be immutable.

For those who are confused why [['']*m]*n is not good to use.
Reason:- Python uses calls by reference, so changing one value in above case cause changing of other index values also.

Best way is [['' for i in range(m)] for j in range(n)]
This will solve all the problems.

For more Clarification
Example:

>>> x = [['']*3]*3
[['', '', ''], ['', '', ''], ['', '', '']]
>>> x[0][0] = 1
>>> print(x)
[[1, '', ''], [1, '', ''], [1, '', '']]
>>> y = [['' for i in range(3)] for j in range(3)]
[['', '', ''], ['', '', ''], ['', '', '']]
>>> y[0][0]=1
>>> print(y)
[[1, '', ''], ['', '', ''], ['', '', '']]

t = [ [0]*10 for i in [0]*10]

for each element a new [0]*10 will be created ..

Incorrect Approach: [[None*m]*n]

>>> m, n = map(int, raw_input().split())
5 5
>>> x[0][0] = 34
>>> x
[[34, None, None, None, None], [34, None, None, None, None], [34, None, None, None, None], [34, None, None, None, None], [34, None, None, None, None]]
>>> id(x[0][0])
140416461589776
>>> id(x[3][0])
140416461589776

With this approach, python does not allow creating different address space for the outer columns and will lead to various misbehaviour than your expectation.

Correct Approach but with exception:

y = [[0 for i in range(m)] for j in range(n)]
>>> id(y[0][0]) == id(y[1][0])
False

It is good approach but there is exception if you set default value to None

>>> r = [[None for i in range(5)] for j in range(5)]
>>> r
[[None, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None], [None, None, None, None, None]]
>>> id(r[0][0]) == id(r[2][0])
True

So set your default value properly using this approach.

Absolute correct:

Follow the mike’s reply of double loop.

To initialize a 2-dimensional array use:
arr = [[]*m for i in range(n)]

actually,
arr = [[]*m]*n will create a 2D array in which all n arrays will point to same array, so any change in value in any element will be reflected in all n lists

for more further explanation visit : https://www.geeksforgeeks.org/python-using-2d-arrays-lists-the-right-way/

use the simplest think to create this.

wtod_list = []

and add the size:

wtod_list = [[0 for x in xrange(10)] for x in xrange(10)]

or if we want to declare the size firstly. we only use:

   wtod_list = [[0 for x in xrange(10)] for x in xrange(10)]

Initializing a 2D matrix of size m X n with 0

m,n = map(int,input().split())
l = [[0 for i in range(m)] for j in range(n)]
print(l)

Matrix={}
for i in range(0,3):
  for j in range(0,3):
    Matrix[i,j] = raw_input("Enter the matrix:")

If you use numpy, you can easily create 2d arrays:

import numpy as np

row = 3
col = 5
num = 10
x = np.full((row, col), num)

x

array([[10, 10, 10, 10, 10],
       [10, 10, 10, 10, 10],
       [10, 10, 10, 10, 10]])

row=5
col=5
[[x]*col for x in [b for b in range(row)]]

The above will give you a 5×5 2D array

[[0, 0, 0, 0, 0],
 [1, 1, 1, 1, 1],
 [2, 2, 2, 2, 2],
 [3, 3, 3, 3, 3],
 [4, 4, 4, 4, 4]]

It is using nested list comprehension.
Breakdown as below:

[[x]*col for x in [b for b in range(row)]]

[x]*col –> final expression that is evaluated
for x in –> x will be the value provided by the iterator
[b for b in range(row)]] –> Iterator.

[b for b in range(row)]] this will evaluate to [0,1,2,3,4] since row=5
so now it simplifies to

[[x]*col for x in [0,1,2,3,4]]

This will evaluate to
[[0]*5 for x in [0,1,2,3,4]] –> with x=0 1st iteration
[[1]*5 for x in [0,1,2,3,4]] –> with x=1 2nd iteration
[[2]*5 for x in [0,1,2,3,4]] –> with x=2 3rd iteration
[[3]*5 for x in [0,1,2,3,4]] –> with x=3 4th iteration
[[4]*5 for x in [0,1,2,3,4]] –> with x=4 5th iteration

As @Arnab and @Mike pointed out, an array is not a list. Few differences are 1) arrays are fixed size during initialization 2) arrays normally support lesser operations than a list.

Maybe an overkill in most cases, but here is a basic 2d array implementation that leverages hardware array implementation using python ctypes(c libraries)

import ctypes
class Array:
    def __init__(self,size,foo): #foo is the initial value
        self._size = size
        ArrayType = ctypes.py_object * size
        self._array = ArrayType()
        for i in range(size):
            self._array[i] = foo
    def __getitem__(self,index):
        return self._array[index]
    def __setitem__(self,index,value):
        self._array[index] = value
    def __len__(self):
        return self._size

class TwoDArray:
    def __init__(self,columns,rows,foo):
        self._2dArray = Array(rows,foo)
        for i in range(rows):
            self._2dArray[i] = Array(columns,foo)

    def numRows(self):
        return len(self._2dArray)
    def numCols(self):
        return len((self._2dArray)[0])
    def __getitem__(self,indexTuple):
        row = indexTuple[0]
        col = indexTuple[1]
        assert row >= 0 and row < self.numRows() \
               and col >=0 and col < self.numCols(),\
               "Array script out of range"
        return ((self._2dArray)[row])[col]

if(__name__ == "__main__"):
    twodArray = TwoDArray(4,5,5)#sample input
    print(twodArray[2,3])

This is the best I’ve found for teaching new programmers, and without using additional libraries. I’d like something better though.

def initialize_twodlist(value):
    list=[]
    for row in range(10):
        list.append([value]*10)
    return list

Here is an easier way :

import numpy as np
twoD = np.array([[]*m]*n)

For initializing all cells with any ‘x’ value use :

twoD = np.array([[x]*m]*n

Often I use this approach for initializing a 2-dimensional array

n=[[int(x) for x in input().split()] for i in range(int(input())]

The general pattern to add dimensions could be drawn from this series:

x = 0
mat1 = []
for i in range(3):
    mat1.append(x)
    x+=1
print(mat1)


x=0
mat2 = []
for i in range(3):
    tmp = []
    for j in range(4):
        tmp.append(x)
        x+=1
    mat2.append(tmp)

print(mat2)


x=0
mat3 = []
for i in range(3):
    tmp = []
    for j in range(4):
        tmp2 = []
        for k in range(5):
            tmp2.append(x)
            x+=1
        tmp.append(tmp2)
    mat3.append(tmp)

print(mat3)

The important thing I understood is: While initializing an array(in any dimension) We should give a default value to all the positions of array. Then only initialization completes. After that, we can change or receive new values to any position of the array. The below code worked for me perfectly

N=7
F=2

#INITIALIZATION of 7 x 2 array with deafult value as 0
ar=[[0]*F for x in range(N)]

#RECEIVING NEW VALUES TO THE INITIALIZED ARRAY
for i in range(N):
    for j in range(F):
        ar[i][j]=int(input())
print(ar)

Another way is to use a dictionary to hold a two-dimensional array.

twoD = {}
twoD[0,0] = 0
print(twoD[0,0]) # ===> prints 0

This just can hold any 1D, 2D values and to initialize this to 0 or any other int value, use collections.

import collections
twoD = collections.defaultdict(int)
print(twoD[0,0]) # ==> prints 0
twoD[1,1] = 1
print(twoD[1,1]) # ==> prints 1

lst=[[0]*n]*m
np.array(lst)

initialize all matrix m=rows and n=columns

I use it this way to create mxn matrix where m = no(rows) and n = no(columns).

arr = [[None]*(n) for _ in range(m)]

from random import randint
l = []

for i in range(10):
    k=[]
    for j in range(10):
        a= randint(1,100)
        k.append(a)

    l.append(k)




print(l)
print(max(l[2]))

b = []
for i in range(10):
    a = l[i][5]
    b.append(a)

print(min(b))


The answers/resolutions are collected from stackoverflow, are licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0 .

Leave a Reply

Your email address will not be published.