Блог пользователя zxqfl

Автор zxqfl, история, 9 лет назад, По-английски

SRM 663 is tomorrow at 11 AM EDT (local start time). I am the author, dreamoon_love_AA and rng_58 are testers, and misof is the language reviewer. Thanks to them for their invaluable help.

Good luck and have fun!

  • Проголосовать: нравится
  • +108
  • Проголосовать: не нравится

»
9 лет назад, # |
Rev. 2   Проголосовать: нравится +44 Проголосовать: не нравится

Just curious — did they change rules? Some time ago such announcements were not allowed.

Anyway, thanks for reminding! Looking forward for an interesting problemset :)

  • »
    »
    9 лет назад, # ^ |
      Проголосовать: нравится +23 Проголосовать: не нравится

    Oh, I didn't know about that rule. Now I know for next time. Hope you enjoy the problems!

»
9 лет назад, # |
  Проголосовать: нравится +17 Проголосовать: не нравится

I have never solved hard problems(srm 659 and TCO 2B) written by you. Hope I can solve it this time 0w0... And I really enjoy your problems. ^_^

»
9 лет назад, # |
  Проголосовать: нравится +9 Проголосовать: не нравится

Does anyone know if the Topcoder web arena is working properly now? Last time I tried, it is very laggy (cannot load scoreboard most of the time, cannot challenge..)

  • »
    »
    9 лет назад, # ^ |
      Проголосовать: нравится +9 Проголосовать: не нравится

    I used web arena to write two of TCO 2nd rounds this year (including last week) and the experience is a bit painful. First and worst of all it's often lagging on typing in the code, sometimes I type in a full line of code to see it appearing after several seconds. Also in both rounds chrome tab with web arena crashed one or two times per contest. Fortunately the code was safe (at least if you compile it). Also I think there are some bugs there, I've seen once that when contest started it was showing me "XXX opened 500-pt problem, YYY opened 500-pt problem" while people were opening 250-pt problem.

»
9 лет назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Thanks. I hope unlike most TC authors recently, you take publishing the editorial seriously.

  • »
    »
    9 лет назад, # ^ |
      Проголосовать: нравится +9 Проголосовать: не нравится

    TC has independent editorialists, editorials are usually not published by authors.

    • »
      »
      »
      9 лет назад, # ^ |
        Проголосовать: нравится 0 Проголосовать: не нравится

      Yes but it's not like they force the authors to sit back and watch as the "community" (of which TC is so fond) tries to find the solutions, is it ?

      • »
        »
        »
        »
        9 лет назад, # ^ |
          Проголосовать: нравится +1 Проголосовать: не нравится

        The point is that authors are not obliged to write the editorials. Plus, last SRM had an editorial posted here, occasionally other SRMs have them, and there's always discussion of the problems, which is as good if not better IMO (just not in one place and harder to extract info from).

»
9 лет назад, # |
  Проголосовать: нравится +47 Проголосовать: не нравится

  • »
    »
    9 лет назад, # ^ |
    Rev. 2   Проголосовать: нравится +8 Проголосовать: не нравится

    I figured out a lot of re-submissions on 300 from high rated coders
    I'm worry :( Passed :D no worries :D

»
9 лет назад, # |
  Проголосовать: нравится +16 Проголосовать: не нравится

Fail. I was absolutely sure that my straightforward solution for 550 is O(n2) (and was really surprise to be the first one who submitted it). It is, but for a single test case. Sigh.

I determined it's incorrect, like, 5-10 minutes before the end.

»
9 лет назад, # |
Rev. 2   Проголосовать: нравится +1 Проголосовать: не нравится

Lol, looks like "systest at 99.94%" means "systest started"... it's been stuck at 99.94% for a while.

UPD: Done. Meh, better than nothing.

»
9 лет назад, # |
Rev. 3   Проголосовать: нравится 0 Проголосовать: не нравится

Who else receive strange test result in practice room for D1 550? I receive failed test #3, which is example 3 :| (when I run batch test it passed)

Edit: Nevermind, I got WA3 when I run test individually. I'll never use Batch test from now on..

  • »
    »
    9 лет назад, # ^ |
      Проголосовать: нравится 0 Проголосовать: не нравится

    batch test uses two fields, success and correct example. Your solution passes the case if correct example is true. I hope this was the problem.

    • »
      »
      »
      9 лет назад, # ^ |
        Проголосовать: нравится 0 Проголосовать: не нравится

      This surely looks like correct to me

      [TEST CASE #3]:
      Success: true
      Status: 0
      Return value: [I@e20834c
      Correct example: true
      Execution time (ms): 14
      Peak memory used (KB): 106704
      
      • »
        »
        »
        »
        9 лет назад, # ^ |
          Проголосовать: нравится 0 Проголосовать: не нравится

        What was the value returned when you tested individually?

        • »
          »
          »
          »
          »
          9 лет назад, # ^ |
            Проголосовать: нравится 0 Проголосовать: не нравится

          It returns no.

          Thank you for your concern, though I'm quite fed up with this & I'm going to sleep now until maybe when my friends wake me up to see the final judge result.

»
9 лет назад, # |
  Проголосовать: нравится +74 Проголосовать: не нравится

The system test (and practice) results for 550-point problem look wrong.

For the test {1, 1, ..., 1}, {1, 1, ..., 1}, {999999, 999999, ..., 999999} (length 2000 each), the report goes like:

Expected:
...397979577, 397979577, 397979577}

Received:
...397979577, 397979577, 39797}

That surely looks like string truncation at some point.

Moreover, the only passed solutions are in Java, so the problem is probably language-specific.

The admins said they are already looking into it.

  • »
    »
    9 лет назад, # ^ |
    Rev. 2   Проголосовать: нравится -16 Проголосовать: не нравится
    Looks like there may be a test case or two that causes an issue with the testing system. We will likely remove those cases, and re-test. Will have results completed as soon as possible, but it may take a little time. Thanks for your patience.
    
    • »
      »
      »
      9 лет назад, # ^ |
        Проголосовать: нравится +40 Проголосовать: не нравится

      Removing might be a bad idea, the test case with lots of 1s is an important maxtest.

    • »
      »
      »
      9 лет назад, # ^ |
        Проголосовать: нравится 0 Проголосовать: не нравится

      So, you actually had to change the test cases: e.g., test case 40 which was [1] * 2000, [1] * 2000, [999999] * 2000 became [1] * 2000, [1], [999999]. Did you have to remove all cases where the output length exceeds a certain limit?

      In my opinion, it is a shame that the system could not have been fixed instead to handle such cases correctly.

      By the way, does MPSQAS still support only Java solutions? Might be no proper way to test for such bugs in advance then, since Java solutions did not seem to be affected.

      • »
        »
        »
        »
        9 лет назад, # ^ |
          Проголосовать: нравится 0 Проголосовать: не нравится

        The setter's solution has to be in Java, the rest can be in other languages AFAIK.

      • »
        »
        »
        »
        9 лет назад, # ^ |
          Проголосовать: нравится +16 Проголосовать: не нравится

        I could have caught the bug because the testers' solutions were in C++, but I neglected to run a final test (I had tested the first time I added data, but when I added those new cases, I didn't). It's unfortunate because it seems like it was just a few characters off, so if I had put Q ≤ 1999, I think it would have been fine.

        I think the overall results were unaffected, since maxtests still worked. The problem only manifested itself when every query was identical, since in that case every answer was the full 9 digits.

  • »
    »
    9 лет назад, # ^ |
      Проголосовать: нравится +16 Проголосовать: не нравится

    Funny, this is actually my challenge case for O(n^3) solution.

    It was most likely added to the system tests in the end as a successful challenge, that's why the truncation bug on it was not caught before the round.

»
9 лет назад, # |
  Проголосовать: нравится -30 Проголосовать: не нравится

Thanks for your (easy)550, otherwise my rating would have taken a dive that would have made a peregrine falcon downright jealous :D

»
9 лет назад, # |
  Проголосовать: нравится 0 Проголосовать: не нравится

Idea behind Div21000? Looks like a dp. How to do that?