Doubt in Time Complexity of Atcoder League Problem

Правка en1, от code_warrior, 2020-09-15 19:00:05

Hello, everybody! Recently i was solving a problem League on atcoder.

I submitted a solution which according to me is O(n*n*n) and so under the given constraints (N<=1000) should not pass, but since it got accepted ,it suggests the time complexity to be O(n*n). Can someone help me finding its right time complexity.

My Code

My Idea:- We initially maintain an array of pointers ptr[] of n teams each pointing to the very first index in its row i.e setting ptr[i]=1 (in 1 based indexing) for all i from 1 to n. Then untill and unless we could not find any valid match we continue our search (searching two positions i and j such that a[i][ptr[i]]==j and a[j][ptr[j]]==i and mark all such two positions and increment ptr[i] and ptr[j] i.e ptr[i]++ and ptr[j]++ for every such pair of indices) and so in each iteration of the main while loop we increase day by one (initially day=0.)

At last we check if any of the pointer ptr[i]<n which indicates we are short of some matches then our ans is 0 else ans=day.

Why it's time complexity should be O(n*n*n)-: The total number of the outermost iteration can be n*n in the worst and in such iteration we are doing linear search through the array of pointers , so it total time complexity should be O(n*n*n). But surprisingly, it seems to be O(n*n). Can someone explain me why is it so?

История

 
 
 
 
Правки
 
 
  Rev. Язык Кто Когда Δ Комментарий
en3 Английский code_warrior 2020-09-16 11:21:40 4 Tiny change: 'y to be O(n*n). Can som' -> 'y to be O(N*N). Can som'
en2 Английский code_warrior 2020-09-16 08:14:03 6 Tiny change: 'o me is O(n*n*n) and so u' -> 'o me is O(N*N*N) and so u'
en1 Английский code_warrior 2020-09-15 19:00:05 1517 Initial revision (published)