Hi! would you provide solution here?

Learned a lot of DS tricks on F2.

Maybe it's same for almost everyone. For me, if I cannot find some useful observation within 30 minutes, I would have no chance to solve it independently.

Regardless of the ridiculously tough leaderboard – way tougher than my Div.1 performance – it's still a good contest. The idea of problem DIFST is great.

upd I read the editorial for APDIS2 and found that I've got all the key observations for this problem during the contest. This problem is wonderful!

emmm.. 'cause almost all the top rankers on the board are Indians?

Or maybe people from other countries are not likely to participate in CC nowadays

I found I'm competing with all silicon brains. When I solved the first two questions and look at the rank board, I found there are lots of participants have AKed the contest.

Indians, you win. I recommend you to participate in one contest called Codeforces Blitz Cup, and in my view yall have the chance to get the champion!

Am I doing ARC or a stupid guessing game?

I have no idea why 1D could pass ~150 people. I think about this problem for 2 hours when testing but can't figure it out!

As a tester, I lost the right to gain positive or negative rating delta in this contest.

queueforces.

I think I'm doing div1 instead of 1+2

Wow that's amazing! Got it.

The percentage of GM or over is far more than I imagined.

btw how do you calculate the "probability to reach next rank"?

Could you explain how it works? I can't understand that.

You say it shows the solved/total problems, however it shows solved/attempted problems. Bug or feature?

First step of D1D is the same as the previous ICPC East Asia Final task.

in 1e1, I enumerating at each transition, which position to modify next and what character to change it to, and then determining how far the resulting string can be matched.

To compute this matching, I implemented an algorithm with a complexity of ( log^3 ): I binary search on the answer, then perform a binary search on positions in the suffix array, and when comparing with a certain suffix in the suffix array, I further binary search on the LCP length.

The overall time complexity is like O(26mlog^2mlogn).

However, it runs extremely slowly in practice.

Is there any possibility to optimize this approach?

In my mind, it didn't work for many years.

Got it. Thanks.

wtf

Thanks to div.1 C for making me IM again. It's too hard to me.

I think the most simple one called rand%2 can bring such probability correctly.

If you can't have fun from something, then it's time to give up and try another thing.

Try to firstly match the smallest numbers with it+1

As a Chinese CPer, I don't know how the guys in high school become LGM. Thus, I believe there would be a secret website they don't want to let us know.

Or their coaches all have superpower and they can invent super ultra awesome problems...

I'm not a high school student anymore so I can't get the secert key either

I've held the same view one year ago, and I used to think that it's completely impossible for me to become a Master or even Grandmaster, because I thought that I've 'trained very hard for two years' without any improvement.

Excuse me! Can you tell me how linyue's mind works?

This is a great advice. I will do that. Thank you!

Interesting E2.

it's actually right.

consider you have an sorted array a[] of length 2n. you want to pair them up so that sum(|a[p[i]]-a[q[i]]|) is maximum. this condition is equal to that there aren't two pairs (a[x1] a[y1]) (a[x2] a[y2) (x1<y1,x2<y2) that x1<y1<x2<y2.

if the p is a permutation of 1~n, q is a permution of n+1~2n, it's obviously that this condition is satisfied.

Same! in the contest I told me that I can randomly make a spanning tree then the answer is O(n), and firstly I dp for n times and got WA, then I changed it to 2n and got AC.

after the contest when I'm discussing the solution with my teammates, I found that my proof is totally wrong and feel very nervous until it passes system test.

bashkort will win IOI 2025!!!

seems to be interesting.

1. i=1 and j=0 means you didn't let any i in 2-n to be already paired so that all i in 1-n can be paired with token 1, it's of n-i-j+1 choices. n-i is i+1 to n, j is the places already being paired, +1 is i itself

2. Because no matter which number you write in ai it has this certain way to pair so it means there are i ways this certain way to contribute to the final answer That's like for each pairing ways I calculate the number of corresponding a[] and get the sum, so one a[] can be calculated many times

let f[i][j] be (i~n) choose j token to take

then in i-th place:

1. don't take that token. f[i][j]=f[i+1][j]

2. take it. there are (n-i-j+1) places to put that token. when you have the position i then the a[i] can be 1~i so f[i][j+1]=f[i+1][j]*(n-i-j+1)*i

then the final answer is f[0][0]+..+f[0][n]

D, very good problem, I love it

E, shit, if it's put on C then >1000 people can solve it. Ai is only used in a process of *800 difficulty, just brute force Onlog^2n dp without any thinking can pass it

you didn't write anything about the scoreboard. however i think the most important reason that the announcement being downvoted is the weird scoreboard. So now I still want to know what happened with this colorful Indian participants on the top of the board. I don't believe that these ones have evolved overnight.

another problem is that why you decide to cancel the div.1 part because I think there may be some other solutions just like play a 7 problem 1+2 or delay the round to make a new G.. cancelling a div.1 round is disappointing

third I want to ask my own question that WHY EVERYONE OTHER THAN ME(EVEN CYAN ONES) CAN MAKE LOTS OF PROBLEMS AND HOLD CONTEST BUT I CANNOT COME UP WITH ANY INTERESTING IDEAS???????

I WANT TO PLAY DIV.1 ROUND

I don't think D1 is a proper problem that can exist in the contest. First, the solution of D1 and D2 are totally different. Secondly, D1 worths 1250 point so that solve D1 first can gain more score than solving D1 and D2 at the same time (maybe D1-500p, D2-1750/1500p would be better).

Now you have the CF Round 1024 (1 << 10) as a Div1 + 2 Round on May 11.

As a International Grandmaster, I solve ABCD within 30 mins and then stuck in E for almost 2 hours.

PermutationForces.

Why I can't use it immediately? It says that I need to wait.

Could anyone explain why I'm downvoted?

Become IGM probably. I've come up with the solution of C2, but it's too hard for me to implement.

The best coders in the best university in China can NOT EVEN COUNT THE EXACT NUMBER of problems.

Sad.

Where is the score distribution?

As an administrator of Luogu, I think this is an interesting post.

orz

Ask 0b101010101 and 1010101010

Arcaea is a great game. However this round is not so great as Arcaea.

But I gain rating from this contest, that's good enough.

I agree with you! The blitz game is very interesting, I really enjoy it.

Golovanov won me but he didn't win tourist. Sad.

The type of game is very interesting. I like it very much.

I found that your code got a compile error. It's of a great fun.

I read the code and found my solutions are exactly the same with them. Maybe it's just I'm too naive to found an easy way to implement. Thank you.

Or maybe I'm just too naive to come up with an idea which would be easy to implement.

Task D2 and E are sh*t. I can come up with the idea within 2 seconds, and I need to code 1 hour to solve them. They shouldn't appear at a codeforces contest.

Stop solving too hard tasks. Solve the tasks that you almost don't need to read the editorial. Do VPs and try to solve *2200 tasks in real contests.

look at contests like CFR1001. diff of D is 2100 and the next task is *2900.

in contest 'Goodbye 2024', I only solve A~E(diff up to *1900), and I have a positive delta.

so you only need to solve easy tasks fast, you can win.

also you can do some vp to know some easy tricks, which would help you to solve template problems easier.

EXTREME huge gap between D/E1 and E2/F.

It seems the link is invalid now. Could you fix that?

Oh, I have already been GM now. That's great.

It says my expected rating will be only 2132. It's interesting that I've never reach ~2100 rating before.

1993E - Xor-Grid Problem Great task.

emmm, when I finished reading the statement of D, I come up with a similar task 1998E )

maybe it has become a very well-known idea and it has been used for too many times

I think task C and D are easier than task A and B. By the way, I think B is an amazing task.

It seems I get rank 5 and become Master.

I think maybe I can't go to sleep tonight.

I practice very hard for two months, and I finally make it.

when could us to discuss the tasks of the contest?

I want to know how to solve task F :(

May I ask how long the window of qualification round would take? It seems I cannot find some useful information related to this... I need to decide when to start once I could know the duration of the window :(

I practice a lot and it makes me to deal with the easy tasks very fast, so that I can have more time to think about the harder tasks and to solve them

Finally, I (would) reach yellow through get a 2400 perf in this contest.

I have struggled 4 years since I play my first contest in AtCoder.

It seems that you only fix it in the solution part, but in hint 2 there also exists one to be fixed

Asia east,china,Xi'an Jiaotong University,https://codeforc.es/team/53320

As a tester, I tested the round.

Using simulated annealing and passed task Ex, maybe the data is too weak?

Hi! may i give you a suggestion that it's not a good habit to have 'system("pause")' at the end of the code. it may cause some unknown problems in competitions.

wish you be better on your CP journey.

In the solution of D2, an 'equal' is wrong spelled.

and if it's equels 1).

But this problem is really good!

the gap between C and D is too large.

First,you need to express your request clearer so that we can know what you mean.

Second,your request is irrational for normal people.

Your English is too bad to read.

A Hieroglyph.

Because you used the Segment Tree, you need to make the array $$$4$$$ times bigger($$$200000 \to 800000$$$ or more).

915F is more difficult because we should turn the vertice's value into the edges in that case. Then they're completely the same.

1) Try to solve tasks in a specific category, do not solve tasks by rating. In the categories, I think it's important to solve some calssic tasks, then challenge some hard tasks. The gap in their ratings can be 300~500. 2) First, read some tutorials, then code by following the tutorials, the last step, try to solve the same tasks by yourself. I usually spend 2~3 hours per day. 3) Codeforces is a great website, but I use Atcoder more. 4) Very old codeforces contests can also provide you with good tasks. The tasks good or not DON'T depend on whether they are new or old.

And sorry for my poor English.

Why there is no this task?I think it's harder than them, and there were very few people solved the task correctly.

Could anyone tell me whether we can solve the task E with Trie?

Count me in