how to make a variable immutable in python

Question

My python script:

N = 2  # 2*2 matrix

a = N * [0]
b = a

print(b)  # prints [0, 0]

for i in range(N):
    a[i] = N * [0]

for i in range(N):
    for j in range(N):
        a[i][j] = 0

print(a)  # prints [[0, 0], [0, 0]]

print(b)  # prints [[0, 0], [0, 0]]

Why does my second print(b) change? How to make it immutable? I'd like my b to still contain [0, 0].

Answer 1

Your understanding of "objects" in Python and variable assignments is flawed.

In a language like C, when you define a variable (say int a), a tiny area of memory is allocated and reserved for this variable and a is now something that refers to this area of memory. You can poke into this area, change it and find that a "has a" different value now. A statement like a = 2 (or a = b where b is another variable) takes the value of the right hand side and writes it into the memory location reserved for a. Now you can modify b as you wish but a will still retain the original value. This is why you can do things like a++ in C which means, "get the value of what a refers to, add one to it and write it back to the same location".

In Python, when you say x = [], a new list object is created and x is made to "point" to that list. Now any change you make to x is affecting this object. Suppose you say y = x, you will get another reference to the same object. Changing y (or x for that matter) will change the object which is now pointed to by x and y. This is what you've done with the B = A assignment. All things done to this object via A will be visible when you access it via B since they both point to the same object. In this sense, all variables in Python are like pointers in C. You can also understand why we don't have a ++ operator in Python since it's meaningless to modify the contents of a memory location like in C.

The solution suggested by others so far is suggesting that you make a new object with the exact same contents as the list pointed to by A and make B point to this copy. This way, if you modify A (or what A points to), B (or what B points to) will remain unchanged.

This however doesn't make B "immutable" (i.e. an object which cannot be modified in place). However, I think you have simply used the word erroneously and meant that you didn't want the aliasing to take place.

Answer 2

When assigning objects in python, you assign references (something like pointers in C).

There are several ways to circumvent this but the IMHO most idiomatic is using copy:

import copy
B = copy.copy(A)

In some cases you even may want to used deepcopy(), have a look at the documentation for details.