Hello Codeforces Community, I am glad to share HackerRank World Codesprint 9 starting on 27th January 2017. The contest duration is 24 * 2 =48 hours.
The winners of the contest will win up to 2000USD cash prizes. Also, there are opportunities to win T-shirts and Amazon Gift cards.
The contest will be rated. If two person gets the same score, the person who reached the score first will be ranked higher.
There will be 7 normal algorithmic challenges and 1 approximate challenge in this contest. I highly recommend you to read all the problems. Most of the problems contain cool figures and subtasks!
The problems are prepared by zemen, Wild_Hamster, kevinsogo, allllekssssa, forthright48, nabila_ahmed and bishop15. Thanks to danilka.pro, niyaznigmatul and wanbo for helping to test the problems. And again special thanks to Wild_Hamster for helping to finalize everything.
(By the way, danilka.pro is now known as danilka.pro!)
Update: Contest is starting in less than 5 minutes.
Update: The contest ended. We will soon announce the winners. Meanwhile, let us know if you liked or didn't like the problem-set.
Happy Coding!
Why can't CodeSprint's approximation problems always be a tie-breaker? Why do the problems have very coarse rounding? (This is a real question.)
Below is just my opinion.
Approximation problems that supposed to be a tie-breaker have sometimes poor quality. This is because the setter and tester don't test enough. Also, the most part of setters/testers barely has experience with approximation problems compared to non-approximation problems. Another reason is that approximation problem can be made "easily" because one can choice a random NP-Hard problem for example and sometimes one don't even think about the problem at all.
You all should be careful when set an approximation problem. Careless creation of a "tie-breaker" problem should be avoided. If you have to set an approximation problem, do test enough. You probably have to get counsel from experienced marathon match players.
Once I heard that second place is not bad too :P
I am not experienced with approximate problems at all, so I can not tell whether it is well prepared or not. But from achieved scores it looks the problem has enough strong testcases to make difference even between top users (maybe not between top 5) ..,
Also I was talking with Shafaet about scoring for approximate problems, now HR doesn't have option to compare solutions betweeen users ( something like 100 % for the maximum total answer, 99,99% for the second and so on — I know it is needed better formula ). For my side improvement with having separated subtasks is more important now than approximate problem. At the end I think way of rounding score is changed, so now difference between real score and achieved score is less than before :)
We had 80+% solution for approximate problem. Anyway, imagine if we made approximate problem even harder, what will be success rate in that way? Currently only 45 out of 7000 participants got 50+% score on this problem. If we increased n or (xi, yi), decreased number of moves or changed formula for the answer, what would be success rate in that way? Will it be right, if only 5-10 participants will get normal score for this problem? I don't think so. You can think differently.
I agree, that problems without ceil score 100%(codechef type approximate problems) looks better, but currently hackerrank doesn't support that type of problems.
I get "Incorrect output" when printing 9 digits after the decimal point. Please check!
UPD: missing read statement maximum 8 digits. But still "Incorrect output" when printing just 6 digits.
UPD 2: It seems display that message when RE.
In this particular case, I think, to just change some parameters of test generation doesn't solve the issue at all. I'm not experienced at approximation problem either so I'm not sure but the problem itself is not very interesting I think.
Perhaps we can learn something from the past problems of TopCode Marathon Matches?
Though, I'm not saying that the problem on the contest is particularly bad quality. I'm (in my original comment) saying in general. To be honest, in my random opinion, quality of the problem is above average level.
Also, can you please answer the question: why do the problems have roundings at all? HackerRank platform requires to doing so?
In my opinion without rounding results will be more random. Imagine 2 participants, that got the scores x and x + 0.00001 with same solution and because of some random parameters they will be in different conditions. That's just my opinion, I can be wrong.
Thank you for your answer.
Your answer is the same as what I thought. But I think in that way rounding is unnecessary.
We can submit multiple times. In that way, we can always decrease variance (kind of).
I can be wrong either but look at Marathon Matches. Virtually no rounding are there as far as I know. Thus I think it should be in that way.
An only issue I can think is when everyone gets the almost same score (when the difference is, for example, less than 0.1%). In that case, a kind of randomness arise. But rounding doesn't solve the issue in that case either (at least the problem won't be a tie-breaker). The problem is very bad in that case anyway.
I see that the top 3 contestants at the moment have the maximum score at the approximate problem. My understanding would be that they all found the optimal solution on each test case (given how the score is computed). Do you mean that there's a tie at the top mostly because of the rounding? (i.e. the scores were not necessarily identical, but they were sufficiently close to the optimal score, that they were rounded to 100%). If yes, then this seems like a big problem and defeats the purpose of the approximate problem of breaking ties (instead the opposite is happening — solutions of not quite equal quality are forced into a tie).
The logic behind rounding is to ignore the small difference in scores due to random parameters or biased test cases (it's hard or almost impossible to create test cases which are not biased to some heuristics more than others). wild_hamster told it earlier in this comment.
Let us know what is the best and most fair way to break ties? Your opinion will help us to improve.
Did anyone recieved prizes of womens codesprint? https://www.hackerrank.com/contests/womens-codesprint-2/challenges Its already been more than 10 weeks, also I didn't get prizes of 1st womens codesprint. Do hackerRank really give prizes, PS: I got tshirt but no prize, I was in top 100 both times
this is really disappointing, do anyone know whom to contact for prizes? Shafaet , allllekssssa
Sorry to hear that, I have sent you a message in inbox. Generally, if prizes are delayed, the email id to contact is [email protected].
Will there be a plagarism check or are the standings final?
There will be plagiarism check. Consider it final only after the rating is updated.
I got the money (bitcoin) but I think I didn't receive the t-shirt (I believe the one I got was from 1st women codesprint, but I'm not sure)
First of all, nice problemset!!! But tests are very weak. For example, my slow solution O(2^n) for Kingdom Division passed for 50 points.
But I liked problems.
Change the tshirt design pls. Your current design is bullshit. I mean the one with "world champion" written in front of it. Even something like writing just "hackerrank" instead of "world champion" sounds more cool.
Shafaet
Nice problems. I think there is some bug with the editor. I made some changes to my code in the code editor and pressed "Run Code" , and suddenly code in the editor changed and reverted all the changes i made in last few minutes.
I can't see other people's solution. When will they become public ?
They are public now.
What is the medal criteria? And when will the ratings be updated?
See this for the medals.
Can see other's solutions for just first 2 problems. Not for rest. Can you confirm that because my browser has been behaving weird lately and this issue might be just with me ?
Try again now.
Works now. Thanks !
How long will it take for ratings update?
It's being updated now, will take around an hour or two.
any hint on how to solve the Toll Cost Digits Problem?
Editorial
thanks but didn't want to see the whole editorial.. got hint from a friend//
Can you share the hint please? :D