SecondThread's blog

By SecondThread, history, 11 months ago, In English

Meta Hacker Cup Finals 2023

The 2023 Meta Hacker Cup Finals will be happening this Saturday, starting at 6:00 AM Pacific.

We'll reveal the scoreboard and winner on a facebook livestream at 10:30 at 11:00. We'll include the stream on Codeforces if Codeforces decides to support Facebook livestreams, but otherwise, you'll want to visit facebook.com/hackercup to see the stream!

Good luck to the 25 finalists, we hope you enjoy the problems, and we hope everyone enjoys the livestream. You can also check out our teaser video before the contest!

  • Vote: I like it
  • +111
  • Vote: I do not like it

»
11 months ago, # |
  Vote: I like it +16 Vote: I do not like it

btw, does anyone know the CF handle of Keita Murase?

»
11 months ago, # |
  Vote: I like it +8 Vote: I do not like it

Waiting and excited to watch the final :)

»
11 months ago, # |
  Vote: I like it 0 Vote: I do not like it
»
11 months ago, # |
  Vote: I like it +10 Vote: I do not like it
»
11 months ago, # |
  Vote: I like it +20 Vote: I do not like it

Is the live stream happening ?

»
11 months ago, # |
  Vote: I like it +10 Vote: I do not like it

tourist orz

»
11 months ago, # |
  Vote: I like it +28 Vote: I do not like it

tourist clutch o7

»
11 months ago, # |
  Vote: I like it +13 Vote: I do not like it

tourist comeback god

»
11 months ago, # |
  Vote: I like it +76 Vote: I do not like it

I tried to solve F using an ILP solver (JuMP/SCIP). However, it did not solve the pretest in a reasonable time, which made me give up cause the ILP solver seemed to not work in this case. It turns out the ILP part doesn't even need a second, but it takes more than a minute to execute Floyd-Warshall in Julia for $$$n = 500$$$ which is kinda sus. I changed it to $$$n$$$ BFS, plus fixed a few bugs and upsolved it. My experience with Julia is really limited, so I don't think I was really close to solving it. upsolve code upsolve code with FW

I also learned how to solve F reasonably simply without these ILP shenanigans — you don't need to think about placing stuffs carefully to not blow up the distance. Just think of it as assigning integers to vertex: Yon need to find $$$d: V \rightarrow [0, K]$$$ such that $$$|d(u) - d(v)| \le 1$$$ and for vertices with $$$d(v) > 0$$$ there should be some vertex $$$u \in adj(v)$$$ such that $$$d(u) = d(v) - 1$$$. Now it is a simple $$$O(nk^2)$$$ DP on cactus, and for me this makes F the favorite problem of the contest.

  • »
    »
    11 months ago, # ^ |
      Vote: I like it +73 Vote: I do not like it

    I was happy to discover the same integer assignment approach for F during the contest -- it indeed made the problem more enjoyable for me.

  • »
    »
    11 months ago, # ^ |
      Vote: I like it +16 Vote: I do not like it

    Thank you! I hadn't even expected anyone to like the cactus problem to be honest :) Just wanted to give some problem with a short statement, a small size input and a implementation-heavy DP as solution. It's great that this observation literally made the implementation less annoying for you.

»
11 months ago, # |
  Vote: I like it +14 Vote: I do not like it

Will Facebook hacker cup be held next year 2024?

  • »
    »
    11 months ago, # ^ |
      Vote: I like it 0 Vote: I do not like it

    Should we be positive guys?

    @SecondThread