LeetCode: Number Complement

Problem Statement

Given a positive integer, output its complement number. The complement strategy is to flip the bits of its binary representation.

Example

Example 1:

Input: 5
Output: 2
Explanation: The binary representation of 5 is 101 (no leading zero bits), and its complement is 010. So you need to output 2.

Example 2:

Input: 1
Output: 0
Explanation: The binary representation of 1 is 1 (no leading zero bits), and its complement is 0. So you need to output 0.

Note:

1. The given integer is guaranteed to fit within the range of a 32-bit signed integer.
2. You could assume no leading zero bit in the integer’s binary representation.
3. This question is the same as 1009: https://leetcode.com/problems/complement-of-base-10-integer/

Solution Thought Process

Bitwise problems are better understood if they are demonstrated via number.

First, take a short course on bitwise operations.

• When you XOR a bit with a 1, the value is toggled. This is one important property of XOR.

• When you have a binary number which has 1 in front of it and the later bits are 0, and you subtract 1 from the number, the 1 will become 0, and the 0's will become one.

For example, let's take 16 in decimal which is 10000 in binary. Let's subtract 1 from it. It becomes 1111 in binary.

1 0 0 0 0         1 6 

     -  1         - 1  

----------       ----
0 1 1 1 1         1 5

So for this problem, we need to XOR every bit of the given number with 1.

So if the number is 10100, then we have to XOR 1 with 0, then 0, then 1, then 0, and then finally 1 from right to left.

So, in a nutshell, we have to do XOR like this

    1 0 1 0 0 
^   1 1 1 1 1
------------------
    0 1 0 1 1

How can we do this?

First, we have to find out what is the length of this binary number. We can do it by using this logarithmic expression:

floor(log2(num)) +1;

How does it find it? Let's take the intuition from 10 based logarithms.

Suppose we have a number 1000, how many digits are in here?

floor(log10(1000)) + 1
= floor(log10(10^3))+1
= floor(3)+1
= 4

So we have the number of digits here, which means we need to have these five 1's. How do we find five 1's as a number?

Using the second bitwise lesson we can see that if we have 1 followed by five 0's, which is 100000 in binary, and then subtract 1 from the number, we will get 11111 in binary.

But how do we find 1 followed by five 0's?

We take the 1 and left shift this bit by 5, the number of binary digits in the given number. There's a shorthand way of doing so, that is

1 << 5
= 100000 (In binary)
= 32 (In decimal)

Voila, we now know where to subtract the 1 from and we will get the number by which we have to perform the XOR.

1 0 0 0 0 0
-         1
-----------
0 1 1 1 1 1

Let's perform the XOR. It will give us the number with bits flipped, 0 becomes 1, 1 becomes 0.

    1 0 1 0 0 
^   1 1 1 1 1
------------------
    0 1 0 1 1

Yan become Ying, Ying becomes Yan.

Gotchas

• Be careful when creating the mask. We may create overflow when we are creating the mask. So we perform a cast operation to long long int and then subtract 1 in the code, which will take care of the overflow problem. Why? Just give a little thought and you will figure it out.

Solution

class Solution {
public:
    int findComplement(int num) {
        int binaryLength = floor(log2(num)) +1;
        int mask = (long long int) (1 << binaryLength) - 1 ;
        return num ^ mask;
    }
};

Complexity:

Time Complexity: O(n), where n is the number of digits in binary. We will have to find out the length of binary digits which we can do by dividing the number by 2 repetitively and we have to do the masking operation. Both of them will take the max of O(n) time where n is the number of binary digits. In the case where we know that the integer is 32 bit, the time is constant. But when the length is unknown, it will loop over all the binary digits, giving us a time complexity of O(n).

Space Complexity: O(n) where n is the number of digits in binary. We have to have that much binary digits for the computation.