Hello everyone,
I want to invite all of you the next edition of one of the monthly contests on HackerEarth- this particular challenge for the month of February is called the February Clash. The duration for the same being 12 hours, consisting of 5 problems, of various difficulty levels. 2 Easy-Medium, 2 Hard, and 1 Challenge problem to keep you occupied for 12 hours. You can register for the contest at: February Clash. Let's see all of you here to make the contest bigger and solve some really mind boggling problems as one of the perks.
Contributors : Bidhan, kunal93, anta, darkshadows, akashdeep and Xsquare
Date : 21 February
Time : Check time in your timezone
January Clash HOF:
1.) Xellos
2.) I_love_Tanya_Romanova
3.) Kmcode
Thanks for a contest :)
It was a nice one; however, I have few suggestions about improving it in future.
Talking about binary problems — what is the reason behind giving same problems for few contests in a row? Problem count some numbers using matrix multiplication was used for January contest, and today I saw it again. Just few values in matrix changed; is it really a good way of creating new problems — by using problems from previous contest? I hope there will be no problem count n-digit numbers with x%3==a digits 4 and y%3==b digits 7 in decimal notation next month:)
And few words about challenge problem.
First of all, when somebody beats top score for this task — leaderboard becomes incorrect for a long time; some scores are updated, some other aren't... Taking into account that binary problems are mandatory for good rank, and challenge problem determines your final place — I prefer to know my actual score for it and my actual rank:)
What is the reason behind making basically 1 testcase for challenge problem? Ok, there were 6 of them, but final score on 99.97% depends on last one. For a problem with full feedback and stable enough solutions — one can implement naive hill climbing for simplest "sort them" solution and make it fit good enough for given test, by simply picking few constants and making a lot of submissions.
What are reasons behind picking such problems and giving such constants at all? By giving so high constants for so strict TL authors leave almost no place for any solution which makes sense; I don't think it is possible to make something better than well-known greedy solution to work good enough to pass TL. And messy implementation of naive greedy solution will also give you a TL :) Mine works in 0.3s on random testcase at CF, and it turns out to be 0.96s at Hackerearth.
And in January ACM ICPC binary problem were given instead of challenge problem :) BTW, it played a bad joke with me that time, because before contest I thought that time will not matter (because of challenge problem) — and it turned out that instead of challenge problem I have to solve pretty standard ACM task, which can be seen often during ACM team contests :)
Thanks for your suggestion we will keep that in mind and in next contest will come up with more interesting questions.
In January, I also thought time wouldn't matter. And I thought there's plenty of time, that I can solve COCI in the meanwhile :D
... and I won in the end :^)
I dont think so at all. Well, I love these kind of matrix-exponentiation problems. The problem from previous contest was different. These are also meant for the beginners so they could learn from their previous experiences and come with the solution at the time of the contest from next time.
Hope you get the point here.
Thanks Hackerearth for providing such contests.
It's a "hard" problem. A "hard" problem isn't supposed to cater to beginners.