affix's blog

By affix, 8 years ago, In English

We've got n nonnegative numbers (a[i]) . We want to find the pair with maximum gcd. For example if we have:

2 4 5 15

gcd(2,4)=2

gcd(2,5)=1

gcd(2,15)=1

gcd(4,5)=1

gcd(4,15)=1

gcd(5,15)=5

The answer is 5.

n<100,000 and a[i]<100,000

I have an O(n*sqrt(n)) algorithm is there more efficient algorithm like O(n*logn) or O(n)?

 
 
 
 
  • Vote: I like it
  • +7
  • Vote: I do not like it

»
8 years ago, # |
  Vote: I like it -7 Vote: I do not like it

What's the limitation for the numbers itself?

Or we can suppose that a[i] = O(n)?

Otherwise, you can factor all numbers and do something with it. In that case assympotics will include sqrt(max(a[i])) and probably it will be better.

  • »
    »
    8 years ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    a[i] < 100,000

  • »
    »
    8 years ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    I can optimize my algorithm to O(max_a[i] * sqrt(max_a[i]) ) but it is not fast enough. I heard there is O (max_a[i] * log(max_a[i])) algorithm! do you have any idea?

  • »
    »
    4 years ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    i did factor and get the gcd by this Mark all divisors of each element, in O(sqrt(n)). Then find maximum element of mark array that is marked more than twice, in O(n). but i really also need to know for which such pair is it the max wrt this solution how do i calculate which pair it is ?? Please help and reply asap

    • »
      »
      »
      4 years ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      You can maintain an auxiliary "vector of array" for that

»
8 years ago, # |
Rev. 2   Vote: I like it +15 Vote: I do not like it
for d = 1..10^5
  for i = 1..10^5/d
    sumCnt += cnt[i * d]
  if (sumCnt >= 2) ans = max(ans, d)
»
8 years ago, # |
Rev. 2   Vote: I like it 0 Vote: I do not like it

I am thinking of a way a little bit strange to solve this problem. You just randomly pick up two numbers and calculate the gcd. Run it 100000 times. Your possibility to get the right answer is very close to 1. Try it out.

  • »
    »
    8 years ago, # ^ |
    Rev. 3   Vote: I like it 0 Vote: I do not like it

    It is not true. Test case 2 4 1 1 1 .. 1 1 1 (99998 times one). Probatility to choose first and second numbers is 2 / (n * (n — 1)).

    • »
      »
      »
      8 years ago, # ^ |
      Rev. 2   Vote: I like it 0 Vote: I do not like it

      Every time the possibility I didn't choose 2 4 is p, then p = (1 — 2/(n * (n — 1))). So in the end, the possibility that I didn't come up with the right answer is p ^ 100000. I think is close to 0. Is that it?

    • »
      »
      »
      8 years ago, # ^ |
        Vote: I like it 0 Vote: I do not like it

      OK, sorry for my idea without test. In fact, p^100000 is about 0.9998....So in this problem, n' scale is 100000, this method didn't work out. However, when n is about 400 or smaller, 100000 times can almost gives a right answer.

      • »
        »
        »
        »
        8 years ago, # ^ |
          Vote: I like it 0 Vote: I do not like it

        When N is so small we could check all possible pairs. I think your idea doesn't work for these kinds of problems.

»
4 years ago, # |
  Vote: I like it +17 Vote: I do not like it

This question is close with HackerRank WoC 34 #2 (https://www.hackerrank.com/contests/w34/challenges/maximum-gcd-and-sum), please don't answer comments like written today and from "fake" accounts till end of competition. However, there is answer in comments...

Good luck