Блог пользователя csacademy

Автор csacademy, 7 лет назад, По-английски

Hello, Codeforces!

We're going to host a new contest at csacademy.com. Round #54 will take place on Wednesday, 25/October/2017 15:05 (UTC). This contest will be a Div1 + Div2, with 7 tasks of varying difficulty that need to be solved in 2 hours.

We are glad to have khadaev as a problem author.

Contest format:

  • You will have to solve 7 tasks in 2 hours.
  • There will be full feedback throughout the entire contest.
  • Tasks will not have partial scoring, so you need to pass all test cases for a solution to count (ACM-ICPC-style).
  • Tasks will have dynamic scores. According to the number of users that solve a problem the score will vary between 100 and 1000.
  • Besides the score, each user will also get a penalty that is going to be used as a tie breaker.

Prizes

We're going to award the following prizes:

  1. First place: 100$
  2. Second place: 50$

About the penalty system:

  • Computed using the following formula: the minute of the last accepted solution + the penalty for each solved task. The penalty for a solved task is equal to log2 (no_of_submissions) * 5.
  • Solutions that don't compile or don't pass the example test cases are ignored.
  • Once you solve a task you can still resubmit. All the following solutions will be ignored for both the score and the penalty.

If you find any bugs please email us at contact@csacademy.com

Don't forget to like us on Facebook, VK and follow us on Twitter.

  • Проголосовать: нравится
  • +79
  • Проголосовать: не нравится

»
7 лет назад, # |
  Проголосовать: нравится +6 Проголосовать: не нравится

Just a reminder, the round starts in 7 hours.

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

Why contests are always on Wednesdays?

  • »
    »
    7 лет назад, # ^ |
      Проголосовать: нравится +5 Проголосовать: не нравится

    Empirically, we noticed this is when we have most users. Are there more people who would like another day/hour?

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

      Well, it would be better if contests are on Saturday/Sunday in my opinion.

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

      I think that it would be better if contests are on different days.

      For each day and time there are some(or a lot of?) users that can't participate at this day and time. For example, i have classes on Wednesday evenings, so it is hard for me to participate.

    • »
      »
      »
      7 лет назад, # ^ |
      Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

      I wonder not only about Wednesday, but also contest time. Recently, it seems CS Academy contest starts at 00:05 UTC+9 / 15:05 UTC. I think changing starting time is better.

»
7 лет назад, # |
  Проголосовать: нравится +4 Проголосовать: не нравится

can someone post last problem's statement if they have it open already. (The site is down)

  • »
    »
    7 лет назад, # ^ |
      Проголосовать: нравится +13 Проголосовать: не нравится

    In CF and AtCoder first thing I do is to open all problems in different tabs to prevent exactly such situation as yours. But on csacademy doing that is a real pain :(. I can't click "open in new tab" on a name of task :(

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

      If you click on the "Tasks" button in the navbar (yes, it's also a button, even though on hover it acts as a dropdown), you can then open each task in a different tab by ctrl+clicking.

      P.S.: As I was trying to check if this is working I discovered a bug on Windows, it will be fixed later today.

»
7 лет назад, # |
  Проголосовать: нравится +13 Проголосовать: не нравится

Is it only me or is it showing 'Bad Gateway' for everyone?

»
7 лет назад, # |
  Проголосовать: нравится +39 Проголосовать: не нравится

Hmm, someone blamed codeforces for being slow and now I cannot access problems in csacademy even though there are only 900 people in the contest...

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

Probably will be unrated... Finally I could get to yellow (I did well) and then it's unrated...

Edit: Seems like it isn't.

»
7 лет назад, # |
  Проголосовать: нравится +3 Проголосовать: не нравится

Will it be rated or unrated? (If you've decided it, we are happier if you announce it sooner.)

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

I have waited 30 minutes to submit D (it get AC now). It cost me a lot of penalty :(

»
7 лет назад, # |
  Проголосовать: нравится +3 Проголосовать: не нравится

"Closest Numbers" is online version of this.

  • »
    »
    7 лет назад, # ^ |
      Проголосовать: нравится +10 Проголосовать: не нравится

    oh, I was an author of our icpc quarterfinal in 2015, and I prepared very similar problem (online too, but less constraints).

    endagorion was a tester for this problem. times later he prepared task that you shared.

    now there a CSA Round #54, I have registered... but in my house turned off the electricity!!!

    now its on, and I see what I missed...

    anyway, as I understand, solution in editorial on csa better, bcs my used nsqrt memory.

    • »
      »
      »
      7 лет назад, # ^ |
      Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

      Did I miss anything? What does logic to write the contest where you already know some problems?

      • »
        »
        »
        »
        7 лет назад, # ^ |
          Проголосовать: нравится +8 Проголосовать: не нравится

        There is no logic. I just noticed funny situation.

        Anyway, you dont know problem until read it. But in div1 round I never read last problem first. So if I participated in this round, most probably I would see this problem after the end.

        And tell me, where is the edge of "knowing" problem? What if I see problem where all needed is a classic segment tree (and maybe some additional thinking)? I need to leave contest? I need to blame myself for cheating? No, bcs this kind of logic does not work here.

        And where is edge of "knowledge" of the problem? When I see this 645F - Cowslip Collections problem on contest, I say "wow, I have prepared similar, but easer, problem for camp some months ago, and I know half of the solution already, the only needed is to bit thinking to completely solve this". And I solve it. And win T-shirt. It is cheating? It is unfair? No. This logic does not work here.

        And last example. Past summer I decided to participate in Snarknews Summer Series. And during one round I... saw MY problem! Just title and some terms are renamed. But this is problem that I prepared for our local competition some months ago. What was I should do? I already solved some problems on contest, and I came to compete in this contest. And I just write my solution and submit them... And I dont feel anything bad. Does I guilty? Does snark, creator of contest, is guilty? NO. This logic does not work here.

        After writing this big answer, I noticed, that your question probably meaning something else. Maybe I word "logic" need to replace for "pleasure"? If yes, then ok. I dont feel competition pleasure in those situations. Only good feeling I is like — Oh, I created this/similar problem before, so its not shitty problem...

        • »
          »
          »
          »
          »
          7 лет назад, # ^ |
          Rev. 11   Проголосовать: нравится 0 Проголосовать: не нравится

          O, I wasn't right if you understood me not in full. Of course, I meant "What is the logic to write the contest [which you write only for pleasure] where you already know some problems, [and why you said "I see what I missed"]". And also I just was interested in this. I didnt mean anything bad. Good round :)

»
7 лет назад, # |
  Проголосовать: нравится +75 Проголосовать: не нравится

Is it CSAcademy or Codechef? Because taking an old problem and requiring online approach by using persistent trees is not the most glorious way of problemsetting. http://mirror.codeforces.com/contest/765/problem/F

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

    By the way my solution with persistent segment tree gets OK on csacademy, but receives MLE on CF, although the constraints on csacademy are higher, and MLs are equal.

    • »
      »
      »
      7 лет назад, # ^ |
        Проголосовать: нравится +58 Проголосовать: не нравится

      It seems that limits on csacademy are actually 5 * 104 and not 5 * 105.

      • »
        »
        »
        »
        7 лет назад, # ^ |
          Проголосовать: нравится +28 Проголосовать: не нравится

        I checked and this is true.

        ¯\_(ツ)_/¯

        khadaev wtf?

        • »
          »
          »
          »
          »
          7 лет назад, # ^ |
            Проголосовать: нравится +8 Проголосовать: не нравится

          How would you expect to pass otherwise? :)

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

            For N 500.000 if you have a constant-friendly pure sqrt, it shouldn't be that hard to fit in 4 seconds (nsqrt ~ 350 millions). If i knew N was that small, I would've used persistent segment trees...

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

              I tried mo on codeforces problem and it got TL.

            • »
              »
              »
              »
              »
              »
              »
              7 лет назад, # ^ |
                Проголосовать: нравится +5 Проголосовать: не нравится

              "If i knew N was that small, I would've used persistent segment trees..." — but persistent segment trees are , aren't they? If so then there's no reason to be afraid of them, in partciular it should be significantly faster than which you convince us may pass.

              • »
                »
                »
                »
                »
                »
                »
                »
                7 лет назад, # ^ |
                Rev. 2   Проголосовать: нравится 0 Проголосовать: не нравится

                Yes, but the other part of the algorithm was NlogNlogVmax and my implementation used more segment trees than needed, so I was afraid about the first part of the algorithm, not about the persistent segment trees themselves. logN*logVmax is about the same as sqrt for the given constraints, so everything depended on the constant, which was huge in my code.

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

                  Ahh, yes, I forgot that we perform n log V queries which gives additional log to complexity (even to memory). Now it makes sense, sorry for that

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

                Persistent tree creates too many nodes — my solution doesn't fit into ML in CF problem for example. With 5·105 queries it would not work too.

          • »
            »
            »
            »
            »
            »
            7 лет назад, # ^ |
              Проголосовать: нравится +20 Проголосовать: не нравится

            I am sure MrDindows will be able to squeeze O(nq) in TL :D

        • »
          »
          »
          »
          »
          7 лет назад, # ^ |
            Проголосовать: нравится +10 Проголосовать: не нравится

          Oh, know I understand why I got AC.

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

        Yeah, it was a terrible idea to prepare this contest in just two days :(

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

        Wut xd? I thought for a moment about exploring "n^3/2 ideas", but disregarded them as unlikely to fit into TL :(