Xellos's blog

IOI 2018 mirror
By Xellos, history, 6 years ago, In English

A while ago, a link for the mirror contest of IOI day 1 has appeared!

Day 1

If you want to discuss anything about the problems in comments, use spoilers.

UPD: Day 2

  • Vote: I like it
  • +46
  • Vote: I do not like it

»
6 years ago, # |
  Vote: I like it +1 Vote: I do not like it

Is there something wrong with scoring? I get this report on Werewolf: Verdict: OK Status: Full solution But my score is only 49 and I have implemented the full solution.

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

    Scoring in Werewolf looks OK for me, it seems that currently only scoring in combo is broken (95 instead of 100).

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

      Looks like they are regrading the solutions for combo now.

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

      I think that test cases are incorrectly grouped, I tried adding a 2s timeout on each test case in which the given graph is a tree, and this happens in testcases 18-35 while it does not happen in testcases 36-54, which should be the testcases for the last subtask. Also, my status is ok on every test case.

      Have you actually scored 100 points on this problem?

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

        not yet, but working on it! Feel free to message me through contest interface to check a specific submission

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

        I have the same problem, I get 0 points on group 5 but the verdict is OK.

        → Reply
  • »
    »
    6 years ago, # ^ |
      Vote: I like it +5 Vote: I do not like it

    Does anybody know if the scoring is ok now?

    → Reply
»
6 years ago, # |
  Vote: I like it -53 Vote: I do not like it

is it normal to put ad-hoc problem on ioi such as problem A?

→ Reply
  • »
    »
    6 years ago, # ^ |
      Vote: I like it +3 Vote: I do not like it

    Yes, ad-hoc problems are quite common on IOI. Only as far as recent problems go, Toy Train and Scales were pretty ad-hoc.

    → Reply
»
6 years ago, # |
  Vote: I like it -7 Vote: I do not like it

Will the second day mirror be held during the real contest or it will be the same as with the first day's one?

→ Reply
  • »
    »
    6 years ago, # ^ |
      Vote: I like it +16 Vote: I do not like it

    Probably not, considering Japan's timezone. Anyway, it's a virtual contest, so it doesn't matter when you compete.

    → Reply
»
6 years ago, # |
  Vote: I like it +10 Vote: I do not like it

Can anyone share their approaches for werewolf and seats ? (можно по русски)

→ Reply
»
6 years ago, # |
← Rev. 2   Vote: I like it 0 Vote: I do not like it

I'm doing the day 2 mirror and I have the "waiting" status on a problem for more than 15 minutes.Has this occurred to anybody else?

LE:My submissions were all tested,but I waited something like 30-45 minutes on average.Also,the scoring for highway might be broken,beacuse I got "ok" on the first 50% of the tests,and still I don't get any subtask.

→ Reply
  • »
    »
    6 years ago, # ^ |
      Vote: I like it +8 Vote: I do not like it

    Yeah, I'm getting same results, but after a while it gets tested.

    → Reply
»
6 years ago, # |
  Vote: I like it +47 Vote: I do not like it

You can solve problems from Day 2 here: https://oj.uz/problems/source/365

We believe our judging speed could be faster than Yandex's :p

Since we didn't have a model solution, we just set the time limit and memory limit as same as the limits of the contest. If you think they are too strict, please tell us.

→ Reply
  • »
    »
    6 years ago, # ^ |
      Vote: I like it +18 Vote: I do not like it

    Tasks from Day 1 are also available: https://oj.uz/problems/source/364

    → Reply
  • »
    »
    6 years ago, # ^ |
    ← Rev. 2   Vote: I like it +15 Vote: I do not like it

    Damn right it's faster, this year's Yandex mirror has Slowpoke as its spirit animal... one extra feature it has is showing exact grader outputs though.

    Highway would benefit from like a thousand extra tests, I got from 90 to 100 points with a few shitty optimisations without any ideas like in the official solution. (Only 1 test with 50 queries, the rest was actually <= 49.) This is why I don't like constant-optimisation interactive problems, betting on coincidence is the best way to get rid of 1-2 queries.

    → Reply
  • »
    »
    5 years ago, # ^ |
    ← Rev. 2   Vote: I like it 0 Vote: I do not like it

    Are you sure everything is fine with subtasks #1, #2 and #3 in doll? In all of them I get a result that it was tested on 2 tests and on first one output is not correct and on second one that it was halted even though I pass all tests from subtasks 4, 5 and 6 (which are supposedly much more general) (not with full score though, but with a decent partial one).

    → Reply
    • »
      »
      »
      5 years ago, # ^ |
      ← Rev. 2   Vote: I like it 0 Vote: I do not like it

      In test #1 of subtask #1, your code gives S = 13 when N = 9. Subtask #4 is not a super-subtask of subtask #1, #2 and #3.

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

        Ahh, I didn't note that this rule about partial scoring applies to subtasks #5 and #6 only and moreover on Yandex I got OK's everywhere, but its scoring is kinda broken, so it doesn't mean my output is correct and these two things led me to confusion. Thanks for clearing this out.

        → Reply
»
6 years ago, # |
  Vote: I like it +8 Vote: I do not like it

Lol, one of my submission to werewolf got 66 pts. So I passed tests where graph could be anything, but failed tests where graph needs to be a line (TLE on ~5 testcases). Is anybody else able to pull this off :D? Then I wrote simple solution to that subtask with line and got 100 pts xD.

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

    I've seen somebody with this on Yandex earlier too.

    → Reply
»
5 years ago, # |
  Vote: I like it +8 Vote: I do not like it

Why is scoring on Yandex in dolls and highway so broken? Everybody gets 0 pts for both of these tasks. Moreover it shows me "Ok Full solution" in both of these tasks even though I surely exceed query limit in last subtask of highway (I mean, I get up to 52 whereas for full score it is required to have <=50). oj.uz shows me that for dolls I exceed limit for S sometimes as well (limit required for full score), but only thing I see on Yandex is "ok" and "checker message: 0.00000"

→ Reply
  • »
    »
    5 years ago, # ^ |
      Vote: I like it +8 Vote: I do not like it

    The Yandex judge isn't very friendly to interactive/unusual problems, probably because it's all a volunteer effort and nobody took the time to make it so this year. If you look at older problems, there's something unsolvable too, several problems that are borderline-unsolvable because you have no idea how to communicate with the judge and one problem from 2003 or so that nobody scored more than 88 points for (I challenge you to solve it for 100). Try finding out who can fix this and cyberstalk (or just plain stalk) them with requests to fix.

    → Reply