What a nasty comment...

haha, same, man

If they solve problems fairly then yes, they're will be shown in the leaderboard. This rule is there just to exclude the people who want to be top-5 by unfair hacks.

i just set https://clist.by/ as my browser home page

oh！ really ? 乛◡乛）

Good luck， I hope u can achieve it.

I will become green B) screen it

lol, guys, 2 rating units. I got 113 out of 115 needed))))

1598 so close XD

Don't post these kinds of shitz next time >(

No, but D is

it is.

I solve E with lower_bound in c++.I think it's the same as binary search.

I solve D and E with binary search，but I am not sure it will not be hack

Start from left of the array and if there is a zero in between two ones, remove the right one. Let this be ansA, do the same for the reverse array and let the answer of that be ansB. Final answer is min(ansA, ansB).

Sorry this is for B

Observation: We are checking if removing an element from the array can make the max element equal to the sum of other remaining elements. And in case we are removing the max element, we are just checking if we can make the second max element equal to the sum of other remaining elements. (Since there is no element with -ve value) So, we just have to find out the max and semi-max element and then iterate the array removing each element at a time and checking the above criteria.

for example, a= [5,2,1,3] Max= 5, Semi Max= 3 Removing 5(Max), a=[_,2,1,3] makes Semi Max, 3= 2+1 Removing 1, a=[5,2,_,3] makes Max, 5= 2+3 These two are only bad indices in this example.

dp[number added][last added]

Naive transitions work in O(n) for a total of O(n^3). If we use sliding window for each layer, we can transition each layer in O(n), for a total of O(n^2).

For E: Store the frequency of each topic in array and sort that array. Then let say we fix the no of topics in first contest be i, then binary search on this array to find if there exists a topic with frequency greater than i, if yes then take the leftmost topic and set it as new index (initially index is 0) and then check for 2 * i and repeat the process. If not found, just break at that index. Repeat the process for new i. Code: http://codeforces.com/contest/1077/submission/45834476

I think recursive find() function that continuously divides possible region for k in half might solve the problem in O(nlgn), but I ran out of the time...

It's easy to find an array that can by cut out x times. So just do a binary search for the biggest valid x.

No.
AC with segment tree (1731ms).

I think O(n^2) solution with a queue which supports amortized O(1) maximum extraction was intended.

condition is (x*k + k -1 < n) print(-1)

maybe because you need to write bool exist=false;

if you don't it will take arbitary value

Your cmp function does not define a "strict weak ordering", which std::sort requires. The short version of the explanation is that your predicate should behave like < and not like <=.

Failing to meet this requirement causes undefined behavior, which basically means that the verdict may vary arbitrarily.

P.S: there might be other problems with the source, this is just one I have noticed.

4
2 7 7 5

Answer:

2
2 3

The reasons for WA are various as for all of the following outputs I found a solution that produces that output:

0

2
2 2

4
2 3 2 3

2
2 3 2 3

me too bro :(

The question asks for the minimum lights you have to turn off for all peoples "sleep well".

same, the pairwise distinct stuff confusing me, i think that for sample test 1 answer should be 1 because it only need 1 pair to make not disturbed array,

wasted so much time understanding this problem >:(

No, it doesn't help to increase the rating.

Some people enjoy this :) And for cf: you don't need to test some solution if it already failed

WoW ! Looks like CF is real fassst.

Can you give the link of solution?

Then you have a different version of the compiler. On my machine with GCC 7.2.0 it runs almost instantly.

EDIT: Actually, you need to have -O2 as a tag for the compiler to make the solution fast, compiler version doesn't matter.

Compiler optimizations. P.S. This is what happens when a person jumps right into the programming, bypassing the CS.

You're using unordered_map; if you replace it with map, you should get AC. unordered_map can sometimes be slow, you can read more about it here: http://codeforces.com/blog/entry/62393

Дизлайк за ДО, когда есть очередь с амортизированным извлечением максимума за О(1).

После некоторых обсуждений мы пришли к выводу, что мы хотим, чтобы люди решали эту задачу при помощи очереди с поддержкой максимума на ней, но при этом если кому-нибудь (как Вам) захочется запихать ДО или Sparse Table — то пусть они пытаются пихать такие решения. Мы не являемся ответственными за ваши мысли. Мы и сами прекрасно понимаем, что отсечь при разумных ограничениях O(n²) от практически нереально при такой постановке задачи. Выбор, как решать, всегда остается за участником.

Между делом, в этом мы хотели дать еще и некоторую учебную составляющую. Такую, что кроме ДО и других стандартных структур данных существуют другие специфичные СД, позволяющие выполнять какие-то определенные вещи быстрее, чем вышеописанные, и что стоит уметь иногда реализовывать и такие структуры данных. Многие участники без проблем написали очередь с максимумами или ее аналоги.

please tell me that why this 45850644 passed and this 45850660 failed ..they are almost please help!!

MLE code : 45823888 AC code : 45852635 Here is my code.

Maybe list need some extra space to save pointer？

in div1/2/1+2

this is education mode, you can't get any beneficiary from successful hack

in few hour i guess

it seems it is already release

Even though a[i] <= 10^6, sum can be greater than that, so your line 42 where you tried to access pre[(sum-a[i])/2] would make you RTE

Magic has been hided after the dots, last number of the input is 532704.

if you "hack" someone by sending inputs that are meant to fail the code you get 100 points if that attempt was successful, likewise the person that was "hacked" will loose their points for the question, as their attempt will be considered a failure. I got hacked and dropped a rank because of it i feel.

I think it's because the predictor is considering non-trusted participants on the calculus

Same for mine too...it was showing +61 but actual increase is +39

same with my solution 45813690 why this fail, and when change map to ll get AC?

http://codeforces.com/contest/1077/submission/45868135 Accepted used multiset.lower_bound instead of lower_bound can anyone help why lower_bound is slower in the documentation both function complexity is O(log).

as for now not released yet

I explained it some time ago on stackoverflow: https://stackoverflow.com/questions/12054415/get-min-max-in-o1-time-from-a-queue

Hmm, it is really weird. I will add the test on which your solution will got TLE. It is very strange that nobody do it during the hacks phase. (And even more strange that I don't do it before the round start. Sorry about that.)

UPD: Woops, this solution got TLE now.

You have to use integer division (//) instead of float division(/)