Feedback of August Long Challenge 2015



Hi All,

I hope you would have enjoyed participating in August Challenge 2015. Please feel free to share your feedback as the answers or comments of the question. We would be really thankful for your responses :slight_smile:


when will the final rankings of August Challenge 2015 be made?


**(Y) (Y) (Y) **



This time the August challenge was quite interesting.

The first 2 questions were adhoc.

The 3rd one ADMAG was a beautiful question. The proof was simply amazing. Had the time been set to 0.5 seconds, it would have made the question very tough to solve.

The 4th one WAYOUT was a regular BIT/segment tree problem. To make the users think about O(n) solution rather than O(nlogn), the time limit could have been made strict.

The 5th question was the most beautiful one. It was more based on removing the unnecessary calculations in ur solution. The time limit set was very good. It actually ensured many solutions got TLE for the last subtask. Even the test case data was very good. Many of my friends got 50 points just because they could not solve only 2 case in subtask 3. Kudos to the problem setter.

The 6th question was again a very good question. But the time limit I think was on the lenient side. 10 day long contest is meant to encourage users to get around the best algorithm and learn more. Had the time been set to 1 sec, it would have been much better. But the test cases were good, even containing the corner cases. I was happy as I got this one AC in 0.00 sec after struggling for about a week.

The 7th question to 8th question were based on graphs, which I currently didn’t knew. So, I can’t rate it.

The most funny was the 10th question, Simple queries. The subtask 6 was actually meant for us to check whether we can do updates in O(logn) or better complexity. Since, nothing was required to be printed, even TEXT submissions with nothing written on it got 5 points. (A link to one such solution is Actually, this meant that anyone who took part in the contest should have got atleast 5 points. :slight_smile:
Although, the question was mind boggling as the number of users who solved it completely suggests.

Overall, I feel the contest was the best one amongst the ones I had taken part in since March’15. Although the number of submissions to the first 4 question was huge, but 5th question onwards it really was an uphill task to get AC.

Really nice work by the problem setters. Also, the Editorials are also were good, explaining each and every minute detail with links to useful concepts.


The questions were really logical this time specially ADMAG.

But I would like to report a fault in the question Simple queries.

Even if you don’t calculate anything (simply input the queries), you’ll get 5 points. This is because there is a sub task which requires you to solve queries 2,3,4 which gives no output.
I know 5 points won’t make a huge difference but still such subtasks should not be included.

Overall problems were really challenging especially CHINSM and DCGAME. Others were beyond my limit :stuck_out_tongue:


Hi to all…

This time long challenge I have attempt very sincerely and learned lot.

I really enjoy this challenge

  • first two problem was adhoc nothing say about that…:stuck_out_tongue:

  • ADMAG was really beautiful problem …I just want to say for this problem…wow!! what a nice problem.

  • WOUT was based on BIT / segment tree

  • all others I haven’t able to complete because these are not my cup of tea…

thanks to problem setter…



although i got 100 points in DCGAME but i am not happy with the weak test cases
my O(n^2) solution get accepted because of no test file has already sorted element (inc or dec) .Initially i started thinking about some algo and DS which could solve it but finally i decided to get at least partial point but when i submit the code i shocked i didn’t expect it from you :frowning:

and also for 6 subtask of simple query when i looked the subtask then i thought for a while but you know now i am average submitter i didn’t thing my self to bother about that subtask i though it already include 1 and 5 type query but i know it was my fault that i didn’t bother. but now i will will will not believe any more on anyone :slight_smile: THANX for it.

and also i am missing (counting sub string) problems like STRQ (feb 15) ,STRSUB (march 15) and DEVSTR (april 15) :frowning:

well it was my first contest when i cross 600 i am happy :slight_smile:


I would say that the time limit for problem WOUT was bit too more. It should have been more strict. I solved the question by lazy propagation and my complexity was O(nlogn) while people were able to solve in O(n). Therefore the time limit should have been a little strict.

Overall the contest was an interesting one (y).


Nice contest! :slight_smile:


Hardest two problems CLOWAY and DISTNUM were not popping out of papers and googling like other times and I think it was the best part of this contest.


7 easy problems - of course, from a relative point of view of someone who tries to fully solve the nine binary ones…

I don’t have much to say about the challenge problem, since I’ve only tried for a short while to mine out a bit of points. I didn’t have much time for the contest, overall; most of my time working on problems was spent on thinking in parallel to doing unrealted stuff.

I really liked CLOWAY as a hard problem, it has a lot of parts that are pretty hard already, and they fit together somehow. A better modulo would’ve allowed for a smaller TL (with NTT), though, without giving out more than “you should use NTT in one of these many parts”, which I knew anyway because many hard problems want fast convolutions recently.

DISTNUM was yet another “let’s jam several quite standard problems into one and give it a crazy TL” type problem. Queries 3 and 4 can be handled by sqrt-decomposition in almost no running time, turning the array into a bigger one where we don’t add/remove, but “enable/disable” elements; query 2 is equal to both of them. Then, queries 1 and 5 are the same thing, we mainly want sums of distinct 0-th to 3-rd powers, so it’s a standard point-update/unique-range-sum problem with a constant of 4. And the hardest part of that is finding a really fast 2D segment tree. I don’t like problems about finding really fast data structures; the time limit is a complete overkill, but at least there are subtasks.

Related to that, the score now is maximum over all submits (sum of points over all subtasks); why not make it sum over all subtasks (maximum over all submits)? In other words, why not score each subtask independently? Not forcing people to code subtask identification wouldn’t take away from the contest and it’d make it more comfortable, so there’s one argument for it.


DISTNUM editorial is missing


The contest was very interesting (although I didn’t have time to try the hardests problems)


Interesting and learning one. Gravity guy and way out had two methods to solve them each. ADMAG was a good,sorry not good but a awesome problem (got 2 wrong logics at first, 3rd one got me AC :smiley: ), for DCGAME, kudos to the problem setter. Was able to get partial in 5th and 6th (although knew the logic of 5th :frowning: ). All in all a nice one :slight_smile:


@anupam_datta: We will release it soon after accounting up cheating related cases. Though we are going to change the rating system, which will be coming up in few weeks.


You don’t even need to input the queries. Just ensure your programs executes, terminates well and outputs nothing. :stuck_out_tongue:


Hello. I apologize for the delay; I’ll finish writing it soon.


@kevinsogo Editorials please… for DISTNUM


If you have missed it, the editorial for DISTNUM is out now :slight_smile: