PWC188 - Total Zero


On with TASK #2 from The Weekly Challenge #188. Enjoy!

The challenge

You are given two positive integers $x and $y.

Write a script to find out the number of operations needed to make both ZERO. Each operation is made up either of the followings:

$x = $x - $y if $x >= $y


$y = $y - $x if $y >= $x (using the original value of $x)

Example 1

Input: $x = 5, $y = 4
Output: 5

Example 2

Input: $x = 4, $y = 6
Output: 3

Example 3

Input: $x = 2, $y = 5
Output: 4

Example 4

Input: $x = 3, $y = 1
Output: 3

Example 5

Input: $x = 7, $y = 4
Output: 5

The questions

The wording is very precise but anyway clear. My initial WTF moment was about the $x == $y case, but the using the original value... part made it clear (for me at least).

As I said, the wording is precise, in that the either and the or are interpreted inclusively, i.e. there are cases where you can apply both operations.

The solution

This challenge started crying EUUUUUCLIIIIIID since the first mental attempt at calculating the result for the first example.

Why Euclid? Well, the operations that I thought I would apply were very close to the typical algorithm for calculating the Greatest Common Divisor between two positive integers. Which, by the way, is this:

sub gcd { my ($A, $B) = @_; ($A, $B) = ($B % $A, $A) while $A; return $B }

OK, back to the challenge. Let’s first put it recursively and very similar to how it’s described:

  • if $y == $x, return 1 (in one step they both go to 0)
  • otherwise, subtract the lower one from the higher one and recurse.

The repeated subtractions are clearly a distraction. If I subtract $x from $y, and I end up with a greater $y, I know that I have to continue subtracting $x. Hence, as long as $y keeps greater than $x, we will subtract $x. That is, we divide $y by $x and this is the number of operations that we do, counting them.

Well, well, not necessarily so fast. There are two cases:

  • $y is not a multiple of $x. We will end up with an integer number of subtractions (i.e. int($y / $x)) and we will be left with 0 < $y < $x. Rinse, repeat.

  • $y is a multiple of $x, i.e. $y = $n * $x. In this case, we have to carry $n - 1 operations with $y > $x, then a final operation because the remaining $y is equal to $x. This is $n operations in total.

Where does this bring us to? Here:

  • if $y == $n * $x, return $n (this includes the old case about $y == $x, right?)

  • otherwise, calculate $n = int($y / $x) and $r = $y % $x, then return $n plus the recursion over $x and $r.

This is where the Euclid algorithm starts to pop up, because it’s basically the same thing, except that the return value is different (i.e. it’s not the sum of the integer divisions, but the final non-zero rest).

So, here’s an iterative implementation:

sub total_zero ($A, $B, $n = 0) {
   ($A, $B, $n) = ($B % $A, $A, $n + int($B / $A)) while $A;
   return $n;

I’m cheating a bit by declaring variable $n directly in the signature, sparing a line. It’s a game, right?

Here’s the Raku version:

sub total-zero ($A is copy, $B is copy, $n is copy = 0) {
   ($A, $B, $n) = $B % $A, $A, $n + ($B / $A).Int while $A;
   return $n;

I hope I was clear enough… otherwise ring a bell!

Stay safe!

Comments? Octodon, , GitHub, Reddit, or drop me a line!