Sums of Perfect Squares (4kyu)

Sums of Perfect Squares (4kyu)

Description

The task is simply stated. Given an integer n (3 < n < 109), find the length of the smallest list of perfect squares which add up to n. Come up with the best algorithm you can; you’ll need it!

Examples:

• sum_of_squares(17) = 2
  17 = 16 + 1 (16 and 1 are perfect squares).
• sum_of_squares(15) = 4
  15 = 9 + 4 + 1 + 1. There is no way to represent 15 as the sum of three perfect squares.
• sum_of_squares(16) = 1
  16 itself is a perfect square.

Time constraints:

5 easy (sample) test cases: n < 20

5 harder test cases: 1000 < n < 15000

5 maximally hard test cases: 5e8 < n < 1e9

1000 random maximally hard test cases: 1e8 < n < 1e9

Solution

The task is based on Lagrange’s four-square theorem, any positive integer can be written as the sum of four or fewer number’s square.

Specially, if n = 4^k * (8 * m + 7), n can not be the sum of 4 squares.

So, there are cases:

1. n equals to the square of its root as u64, return 1.
2. Use a loop to find out if n is sum of only 2 squares.
3. Check if n applys to the equation above, return 4 if yes or 3 else.

fn sum_of_squares(n: u64) -> u64 {
    //case 1
    let num = (n as f64).sqrt() as u64;
    if (num*num) as u64 == n{
        return 1;
    }
    //case 2
    let end = (n as f64).sqrt() as u64;
    for i in 1..=end{
        let left = ((n - i*i) as f64).sqrt() as u64;
        if left*left == (n - i * i) {
            return 2;
        }
    }
    //case 3
    let mut remaining = n;
    while remaining % 4 == 0{
        remaining /= 4;
    }
    if remaining % 8 == 7{
        return 4;
    }
    //case 4
    return 3;
}