MarioYC's blog

By MarioYC, 12 years ago, In English

Hi everyone, so I found this problem today, and it seems like a really obvious bipartite matching, but I've been getting TLE in it. Is there some faster way to solve it?

  • Vote: I like it
  • 0
  • Vote: I do not like it

»
12 years ago, # |
  Vote: I like it 0 Vote: I do not like it

O(VE) bipartite matching algorithm obviously gets TLE on dense graph with such limits. I think solution which runs in O(sqrt(v) * E) will pass. For example you can use "Hopcroft–Karp algorithm".

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

    Sorry, forgot to say I used Hopcroft-Karp.

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

      Strange... Show your code, maybe i can help. Solution with O(N ^ 2.376) exists, but i think Hopcroft-Karp must pass.

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

          Your algorithm works in O(VE) time and it isn't Hopcroft-Karp. Just try this testcase: 500 500

          2 2 2 2 2 ... (500 times)

          2 2 2 2 2 ... (500 times)

          Copy this a few times and then add "0 0". Your algorithm will work more then 5 seconds (and will get TLE).

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

            Thanks, I found the mistake, I was using the || operator instead of |= when updating the variable change. With this change it still isn't Hopcroft karp because i'm not taking shortest paths, but that only makes it a bit slower.

»
12 years ago, # |
Rev. 2   Vote: I like it +5 Vote: I do not like it

Dinnic works here, my solution accepted in 4.86 seconds.