How To Swap Values Without Temporary Variable

My dear readers, this article will be invested in explaining about swapping two integers without the use of temporary variable. The reason I am writing this article is because it has been asked several times in an interview for a CS internship/job. In fact, a friend of mine who applied for an internship at a local company was asked the same question. The problem tends to test your analytical skills as well as understanding of programming fundamentals. However, in practice one should never use the following approaches which I will explain why towards the end of the article. Starting with no further delay, there are two approaches to the problem. 1. Using XOR operator

2. Using addition and subtraction operators

Throughout the article we use ‘^’ symbol for representing XOR.

Below is a table showing the results of XOR operation

0 0 0
0 1 1
1 0 1
1 1 0

This concludes that for different value of A and B, the XOR operation gives 1 while it gives 0 for the same value of A and B.

XOR also termed as exclusive-or has following properties

1. Self – inversed

A ^ A = 0

Any value XOR’d with itself gives 0.

2. Identity

A ^ 0 = A

Any value XOR’d with 0 remains unchanged.

3. Associative

A ^ (B ^ C) = (A ^ B) ^ C

The order does not matter.

4. Commutative

A ^ B = B ^ A

How to swap two integers using XOR

It is simple and underlies on the above mentioned properties/principles.

Say for an instance a = 2 and b = 3

The XOR operation is performed between two corresponding bits of the numbers.

In binary 2 = 0010 and 3 = 0011

For instance, while performing a ^ b, the most significant bit of a is XOR’d with the most significant bit of b and so on.

The following three statements swaps values of a and b.

a = a ^ b

b = a ^ b

a = a ^ b

Congratulations, you’ve swapped values in a and b without the use of a temporary variable. How?

What going on in the first statement, lets see it in a table.

a b Final value of a = a ^ b
0 0 0 (from self-inversed)
0 0 0 (from self-inversed)
1 1 0 (from self-inversed)
0 1 1 (from Identity)

At the end of the first statement, the value of a = 0001 = 1

Now the value of a = a ^ b,

What’s going on in the second statement, lets see it in a table.

a = a ^ b b Final value of b = (a ^ b) ^ b
0 0 0 (from self-inversed)
0 0 0 (from self-inversed)
0 1 1 (from Identity)
1 1 0 (from self-inversed)

Voila, we’ve got the initial value of a I.e 2 to be held by b now. It worked but how? It’s nothing but the combination of self-inversed property followed by identity over the integers. For the second statement

b = a ^ b // After the first statement the value of a is equal to a ^ b, so

b = (a ^ b) ^ b

This can be rearranged by associative property and written as

b = a ^ (b ^ b)

We know from self-inversed rule, anything XOR’d with itself gives 0 . So we may now write

b = a ^ (0)

From Identity rule, we know something XOR’d to 0 will leave the number unchanged. Therefore we get

b = a

That’s how we got the initial value of a to be stored to b at the end of second statement.

Moving on to the third statement which is

a = a ^ b // By now the value of a is a ^ b and value of b is the initial value of a

so we may also write

a = a ^ b ^ a

Now this can be written as

a = a ^ a ^ b

Using self-inversed rule

a = 0 ^ b

Using identity rule

a = b

Through Table

A = a ^ b B = a A = a ^ b ^ a
0 0 0
0 0 0
0 1 1
1 0 1

At the end of the computation, we have b = 0010 = 2 and

a = 0011 = 3

Why you shouldn’t use XOR for swapping values?

For an instance, say

int a = 2

int *aptr = &a

int *bptr = &a

In the above set of statements, aptr and bptr are pointers to a. Using what we just learned

*aptr = *aptr ^ *bptr

*bptr = *aptr ^ *bptr

*aptr = *aptr ^ *bptr

Now since both the pointers are pointing to the number in the same location so all the above statements are making change at the same location whose result is data loss.

Swapping using addition and subtraction operators

a = a + b

b = a – b

a = a – b

This is pretty straight forward. Say for an instance

a = 5 and b = 10

Then the result of the first statement will be

a = a + b yields

a = 5 + 10

a = 15 and b is unchanged I.e b = 10

Now when the second statement is run, we have

b = a -b yields

b = 15 – 10

b = 5 (which is the initial value of a. Note that the value of a has already been changed to 15)

Running the third statement

a = a – b

a = 15 – 5

a = 10 (which is the initial value of b. Note that the value of b has already changed to 5)

So what do we have at the end

a = 10 and b = 5 (values swapped)

Why addition and subtraction shouldn’t be used to swap values

Say for an instance the values of a and b are large integers. We are performing addition on the first statement which can lead to overflow since we have large integers.

Should you have any questions, do comment below.

Bhishan Bhandari [22] Brewing contents directly from the Himalayas of Nepal. I am a hobbyist programmer and enjoy writing scripts for automation. If you'd like a process to be automated through programming, I also sell my services at Fiverr . Lately, I like to refresh my Quora feeds. Shoot me messages at  

Leave a Reply

Your email address will not be published. Required fields are marked *