Hello, Codeforces!
We are glad to invite you to participate in Codeforces Round #449 (both Div. 1 and Div. 2) which will be held on December 2, 17:05 MSK.
There are seven problems, created by webmaster, ODT, dogther.
This round is about the happiest girl in the world — Chtholly Nota Seniorious~!
You will help Chtholly, Nephren, Ithea and Willem (characters in the great novel and anime "What Do You Do at the End of the World? Are You Busy? Will You Save Us?") to solve some problems.
(Image by gin_sei56(・.8・) on pixiv)
This round is our first round on Codeforces.
Thanks to zcyskyaa for helping us, Arpa, cyand1317 and Tommyr7 for testing the round, vintage_Vlad_Makeev and KAN for round coordination and MikeMirzayanov for Codeforces and Polygon platforms.
This round has 5 problems in each division and you have 2 hours to solve them.
The scoring will be announced shortly before the start of the contest.
The contest is rated for both Div. 1 and Div. 2 contestants.
It's recommended for both divisions to read through the Interactive Problems Guide before the round.
Wish everyone high rating and accepted submissions!
Upd: Scoring is 500-1000-1500-2000-2500
Upd2: Congratulations to the winners:
Div 1:
Div 2:
The editorial will be posted soon.
Upd3: Editorial
Auto comment: topic has been updated by ODT (previous revision, new revision, compare).
I would solve 2E if there was a little more time
The experience is so bad :(
Problems are really good
Hope there is no long queue next time
Looks like an interesting round!
I agree with you.
Is lxl rated?
I feel that the round will be rated
Too lazy coders use bits!
Ohhh bro, First of all, check it for "Is it DATED" !
However,the round is semi-rated.lol
Sounds like a poisonous round.
I think Round#447 is more poisonous! QAQ
Really? Round #447 is interesting and fun,isn't it?
被删除的Tommyr7 20:45:30
没有啊。。。div1E真的是字符串分治然后母函数然后FFT一下就做完了
被删除的Tommyr7 20:45:39
只是lxl表示你们知道也不会做，所以提前透露了
Translate:
Tommyr7
Not at all...Solution for Div1E is just string divide and conquer then generating function and FFT
Tommyr7
lxl leaked it before the contest and said you can't solve it even if you know the problem
UPD:This poisonous round must be unrated!
Maybe only students from NS can solve this problem!
As a tester, just joking.
In fact, I think the problems are okay as a codeforces round.
But in China, lxl (ODT 's Chinese name) is considered as a problem setter of the most difficult and troublesome problems.
And many contestants are afraid of lxl, and always think that his problems are bad. I just want to say that, the contest is quite good, and lxl and the others authors work very hard to make a good contest!
Wish you have fun in the contest!
See you all!
Chinese CodeForces div1 rounds' difficulty is above average anyways
Nope Nope Nope Nobody's gonna AK this Chinese round.jpg
Chinese CodeForces div1 rounds usually have less hacks， so I dislike them.
Why it isn't in the main page? UPD: It's now :D
Chtholly is good!
The happiest girl in the world!
Wish everyone enjoys the round and few bugs! And hope all the programs will be fast in practice :D
Chtholly is invading Codeforces... And, I really don't want a math contest! PLS!
It surely won't be a math contest. It might be a Data Structure contest, I suppose.
No tricks please
Where can I find these pictures?They are very interesting!
Auto comment: topic has been updated by ODT (previous revision, new revision, compare).
Auto comment: topic has been updated by ODT (previous revision, new revision, compare).
ke jiao xing guo(
The girl on the picture is not Chtholly :(
Interactive Guide — Div 2 I am comming back :D
Hope not to meet five data structure problems...
Hope not to meet five data structure problems, too.
Hope not to meet five problems with block dividing algorithm, too.
Data Structures Round? Very interesting! But I hope there will be less data structures problem. :)
I know you are good at solving data structure problems! Aren't you?
I agree with you.
YYQ AK!
YYQ AK!
YYQ AK!
I can't read the annoucement again...
Check your internet connection! XD
I mean,I have no bravery to read it again:(
Фалунь Дафа хорош!
Is Chtholly rated?
What a Taiwan Round
Dear ODT, why not to darw chtholly with me? It will be fun, my dear friend!
Some interesting IDs belonging to ODT upvote this annoucement...
But I prefer to practice FalunDafa rather than take part in this round!
В конце концов, Чжэнь Шань Жэнь это единственный критерий для оценки хорошего или плохого человека.
Chtholly is good! hope the contest will be the same as usual. also hope everyone high rating!
Could you resize the image to be a bit smaller? It looks really weird on the front page.
DULIU mcfx HENTAI lxl
I will never participate in a DS round! (emmm... and I advise ODT to prepare for National Higher Education Entrance Examination in a hurry (wulxl :I can pass the exam easily and taking lessons does nothing to me!So I will make more DULIU problems to make money!(:
Chtholly Nota Seniorious!
orz krydom ,upstairs ↑.
Chtholly Nota Seniorious
I think this contest will be very interesting!
But "Is Chtholly rated?"
another weeaboo garbage jesus fuckign christ what happened to this website
No, we practice Falun Dafa, we are real and kind and forbearing.
The police is coming for you.
Let them come. A real Falun Dafa studier is not afraid of commies.
I guess you'd better shut your f**k**g Falun mouth before you attract more police to our computer room.
You d***less person get away
Don't be so rude, my friend :)
Yes, at least we are all anti-commies.
zcy!!! is he the man who has hung and beaten the trained team ? %%%%
I like this round of contest
I think too many big images and words in animes appearing in the contest will make me feel #%#$@^(&#! when I solve problems.
I hate people who put big image of Chtholly everywhere!
死宅真恶心
Not so big. KAN will sure that the image in a small size :(
ODT YOU NOT YELLOW MATE
Does that mean he can't be a good problem setter?
By the way, ODT is very excellent!
Love this Anime and LN so much :> Hope everyone high rating and have fun coding :>
Looking forward to such a 'colourful' contest!
Well, except troublesome Data Structures.
Uh, I'm wondering if there will be big pictures in desciptions of problems.
This picture is really nice. I wonder where do you find it.
this picture is famous, dont't you know? Use baidu before asking! ( www.chtholly.ac.cn is also a nice choice!) :)
well, I didn't find the picture above on baidu indeed.
cyand1317 confirmed there won't. Thanks cyand.
So after consideration and discussion, we've reached this decision.
There won't be large, distracting images. Some problems contain smaller images that may act as a delimiter, though — start reading the statements after the image if you'd like to.
I agree it's the best decision about images in problems. I'd like some lovely images and decriptions or scripts about animes, though. Therefore, I can decide if I'm interested in that anime and whether to see it later. However, many feel sick about images of cartoons. I suppose this is the best balance, making ODT able to share and express, while not disgusting others. Anyway, lxl just can't help putting his beloved into problems.
Wanna know if you'd like that anime? Go to MAL, not cf.
I like some anime, but I don't want it mixed into problems.
It's like saying, "wow I like soda, gonna make some soda-flavored pasta :D"
Again these shitty anime pictures covering the whole screen...
Yet another anime round,but with DULIU problem setters...
But cheerful with Chtholly :)
I think its not good to use Interactive Problem because many Div. 2 contestants. don`t know it for me i didn't see it before
So why don't you go through the interactive problems guide as suggested in the blog?
I did but i suggest it will be good to use it in educational round or unrated one
This might be harsh — but you are not going to solve anything and improve with this attitude.
So what is your suggestion ?
Solve an old interactive problem if you don't feel confident.
e.g. http://codeforces.me/contest/680/problem/C
I love these anime rounds :D
I also like animes
Wow anime round ! what could be better!
Wish round will be without system errors! And of course, wish problems will interesting and enjoyable ...)
lxl(aka. nzhtl1477, ODT) is really a toxic tumor!
Oh look, another anime that every weeb have watched it except me. Somehow people really like to recommend this anime but I am so triggered by it's title so I skipped it because of it's title is literally longer than my cumbersome comment.
PS: Weebs are taking over ICPC in EC in case if you don't know
(full login password: I love Megumi, Google Megumi Kato)
Pffft, Megumi from konosuba is wayyyy cuter.
EDIT: Oops, this is Megumin not Megumi. Sorry.
Except this is Megumin.
'n' is often added to Japanese names to make them cuter.
Which means megumin is cuter than megumi >:D
Hope everyone will have fun with Chtholly in this round~ P.S. I love Nephren forever!
When there will be a "Nephren Round"?
I like this contest! Maybe I'll participate in it.
Maybe Chtholly will get high rank in Chtholly round
Not maybe, It should be: "must be" :P
You did it and done great ... congrats !!
The amount of discussion on this thread is growing faster than exponential running time! :O
Yeee, anime round, we all were missing it.
An anime round?
Or a lxl data structure round?
This won't be a lxl data structure round.
glhf
On one hand, data structure problems are nice, but on the other hand, we have interactive...
Hope the network is good tomorrow. The server was down for several times this week.
How to solve D? I used binary search + FFT but it was TLE.
You need to use online dynamic persistent red-black tree with square root decomposition + some dp optimizaton + FFT
Thanks a lot. I finally figured out the TLE test case of my solution:
But seriously, why most of light novel need to have a ridiculously long title? This, Moshidora and Oreimo each seems like they have about 50++ characters in their title >_<
So 51 characters?
nah it returns 50.
you were right
V I R A J
M A D D U R
WeEEeeeEeeEeEEEbs
It seems to be an unsolvable round?
Five difficult data structure problems!
这这这……what is this? 好像混进了什么东西。
How to solve Div1 A? I code for 10KB but got WA.
You only code for 10KB? How could you do that?
What? How can you solve it within a 10Kb-code? I have code 30Kb but I haven't finished it yet! :(
What makes you sure that you can solve it within a 10kb code,bro?
Beacuse I think it's Div1 A, so I think the code won't be too long.
Have you forgotten who are the editors of this round? XD
Wake up mate, it's ODT's problem, the code can't be short.
Oh.. I forgot the editors of this round! thx bro, I will recode soon! :)
But I wrote a 5KB code and got accepted :)
You are so strong :) I have just complete it in 35Kb. (QAQ)
In fact, I missed a 0. So it is 50KB :)
But you got AC.
I got WA199 at last :(
I lose hope on recoding :( It's too long..
Why not try to get the answers with a Brute force and just print the answers for each tasks? Certainly you need to beg others in your room not to hack you XD
An FST will be ready for you :) You know, those problem setters are hackers :)
Oh no I hope that setters are lazy and just made weak data,though I'm sure that is impossible. :(
You can submit it at the last second so that no one can hack you.
Oops, Problem setters can :)
I got accepted and the time complexity is only O(logn) !
How to solve Div.1 F? It's a challenging problem with a 30s time limit. :(
Wake up bro, the article has said : This round has 5 problems in each division and you have 2 hours to solve them. No problem F in Div1 :)
I know that :) So why is there an F? emmmmmmmmm maybe a DULIU problem setter?
Maybe an extra problem. So you can imagine how difficult it is...
What means 'DULIU'?
There are many kinds of DULIU problems,such as:
The author's solution runs 0.9s,and the time limit is 1s.
The author uses 5 hours to write the solution and the length of the contest is 3 hours.
The author uses some strange algorithms to solve the problem.
DULIU (毒瘤) means tumor in Chinese
It refers to those problem setters who set up exceedingly difficult and troublesome problems, thus making most problem solvers revolted so that they can't help complaining that the problems and their setter are really DULIU.
If you can solve all the problems and the total length of the codes is less than 50kb,you can see the Div.1 F.
How to sovle Div1C with block dividing algorithm? I got TLE over and over again.
use
std::bitsetand do some constant optimizations.Deleted
Let's wish Good Luck codeforces servers)
CONTEST: Omae wa mou shindeiru!
ME: NANI?
What's up with this weird discussion going on? Is ODT bad problem setter? Can you guys link me to some of his contests? This is really interesting.
http://www.lydsy.com/JudgeOnline/problemset.php?search=YNOI
if u r a true anime fan u should watch boku no pico great anime with intresting plot
yoooooooooooooooooo I'd NEVER watch that! It's revolting. I'd like lovely Chtholly better.
I can confirm this is true, I've watched that many time before. 10/10 would watch again though.
P.s: Pico best girl.
Every 3rd round on codeforces since I have joined is unrated due to some tech issues. I am scared today :p
The queue is insane.
Long queue for anyone else?
yes!
I just waited for an hour and got a WA.
welp, guess i'm doing something else of my afternoon than a codeforces round...
Lol this will probably be unrated
6-minute-queuing (until the moment I write this comment). Last time I started to complain about queuing issue, it subsided a few minutes later. Hope it will work this time as well :<
I've submitted a code with 22 minutes. No response yet. So, I it's a 16min+ queue time.
Well this is starting to get ugly...
30+ pages of queue? How is anyone supposed to time manage like this :/
My submissions can't even pass "pretest 1", because they TLE on "in queue"
37 pages queue...
No server problems so far but the queue is just as bad!
In queue = unrated pls
Long queue. Submitted B 20 minutes ago and still no verdict
Submitted B 20 minutes ago and only now I get a response that I missed it. This is pretty unfair and I request it becomes unrated.
time to wait another 20 minutes to see if I got it
The math says system testing will finish before pretest :-P
Seems fair that chtholly might win the contest
nearly 40 pages queue, should I continue the contest or not? :P
I'll go watch WorldEnd while waiting for the queue .__.
The nice problemset was simply ruined because of the queue :/
Unrated
Unrated again ?
Maybe unrated Sad :(
What are you doing at the beginning of codeforces round449？ Are the judger busy？Will it judge our submissions of this contest？
After 40 minutes I got my verdict Wrong Answer ! And after re-submitting when I will get my verdict ? After the contest ends?
I submitted 1st question in 9 min and got a score of 488 and after 2 min it was reduced to 428. Can anyone explain why? All the pretests passed and the total score was still 470 and yet they gave me 428?
Your solution got submitted twice. Even if pretests on the first submission pass, it is counted as 1 penalty.
I won't save a poisonous problem setter even if I'm not busy at all ：（
Dear codeforces, please hire some good engineers . those who can manage your servers
Did you know that 55 is not only the fifth palindrome number with 2 digits, but it is also THE NUMBER OF PAGES IN THE QUEUE?
I just waited 30 minutes to find wrong answer verdict
Oh this is just great. (35 minutes waiting for that verdict)
And also this one. "Working on"? — I guess it should be an unrated announcement instead... 
UPD: Seemed like many people's opinions are going to the void.
Waiting for round getting declared unrated and Mike's apology blog
Was testing system replaced by a group of Chinese? :)
Btw, problems are very good.
problems was good ? oh man are you kidding ? problem C / Div2 , A / Div 1 was good and beautiful ?
Not beautiful, but good. B and C are better than A(div1).
It's just my opinion.
I got a friendly advice to make a standalone blog instead, but couldn't delete the original message, because it was too old.
I think we will need BigInteger queue, because long queue will overflow :'D
It seems to be unrated :(...So sad to hear about it...
The announcement should be "This contest is unrated!"
I have lost hope to solve the problem. Because of [unsigned long long] queue and the HARD problem(Div.1) :(
But, it seems to be unrated?
And here's another unrated contest -_-
The girl in the post is shedding tears because of the loooooooong queue...
Steelix used unrated attack It was super effective! Togekiss fainted!
i think that the end of the queue is at the end of the world...
Unrated please; the server problems this contest are much, much worse than last contest, and the last contest was unrated, so this one should be too.
It's not the first time such long queues on Codeforces, it's time to buy normal servers. And yes, the round should have been unrated
Dear Mike Mirzayanov,
I think it's time to announce this round unrated because of failing to fix the issue with the judging machine. It's 11-12 pm in some time zone and it's time to sleep if the contest is declared unrated. Thank you for understanding us.
Best wishes.
Please unrate it.Getting W/A after waiting for 35 mins straight is just so frustrating.
What are you doing at the end of the queue? Are you busy? Will you test us?
This is my second codeforces round! And I will be unrated until 2018 =)
And you knew that the length of the last line in task C, is equal to the NUMBER OF ASSIGMENTS IN THE QUEUE!!!!!
That wonderful feeling when you submitted your solution about 50 minutes ago and now it gets RE 7. Love Codeforces servers!)
Round extended by 15 minutes? Do you think this will make the round any better?
Maybe he just want to get 15 more minutes to consider whether making this round unrated due to technical problems. :D
Increase the time by merely 15 min and decrease the score like a landslide? I don't think 15 min is enough though ! This has to be unrated !!!
And after all this they extend the contest by 15 minutes!!! Do they still expect to make it rated??? I don't know whether to laugh or cry!!! That's damn funny!
In Educational Codeforces Round 33 (Rated for Div. 2) , I waited for the long queue but that round was still rated.
The queue in that contest was at most 10 minutes (I think), which I feel is STILL fine, compared to today (It's not ideal, but it's manageable). Today, people had to wait for 30-40 minutes to receive a verdict.
Lagforces round #449
I will leave CF if this round is rated.
Still in long queue!! RIP CF
genuine question: What will you do now since the round is semi-rated ?
My rating has been changed, so I leave CF.
вы что CF ? ?? что вы делаете разве не анрейт? афигеть
What are we going to do in 15 minutes if verdict is given after 40 minutes? -_-
When you are Codeforces, and have a 40+ page queue
:')
It won't be a surprise if the comments on this thread are also queued LOL!
Waited for 25 minutes & got WA
we could use this waiting time for Hacking. _/_ Codeforces.
too hard problems for me
What do you do at the end of the World? Are you busy? Will you save Codeforces?
damn that problem C, I know I did something wrong,but the problem feels really frustrating, like hardcoding,my eyes hurt after reading that statement "will you save us...." also was waiting long long time before verditc came "wrong answer on pretest 5" like , FUCK NO,not attempting that again after 30 minutes :(
Make it unrated...!!
The problems are very good . thanks to the writers . but long queue :(
Dear "whoever decided to extend contest by 15 minutes",
Do you really think that additional 15 minutes would make this competition fair? Or it is just some desperate attempt to make this contest rated?
If it's the first case: - no competition would be far from fair, I've been waiting about half an hour to see that my program does not work. Then I tried to fix something... guess what — I've been waiting another half an hour to see that a bug I found was not the only one. The point is — I should know what I know now max 10 minutes after my first submission (and it was 60, where are my 50 minutes?). My first concern is about amount of additional time. Why 15 minutes seems to be enough to make this contest fair? I don't feel like this is enough. The second thing is what ranking should be — is a person who get accept after X minutes at first try better or worse than person which would get accept after several trials in the first Y minutes (especially for X >> Y)? Because long testing queues affect the second one, but not the first one.
If it's the second case: please don't — just admit that it did not work out, that's what adult people do.
Sincerely, Kroma
I agree with you, but extending the round doesn't mean they are going to make it rated. I am saying this because, in a very recent contest, there was an extension of 30 minutes (!), but the contest was still declared unrated (for the same reason as today).
Yeah, and the server was faster that day than today ;)
If that's the case, then I'm sorry that I exposed you to my butthurt. To explain myself I can only say that I would bother with extending time iff there is a chance to make it rated. So I assumed that "whoever decided to extend contest by 15 minutes" may have similar approach.
The queue has disappeared... I don't know whether it's a good news or a bad one. Does it mean that the round will be rated? What a sad story... But I do think that a lot of contestants have go to bed after seeing the extremely long queue one or half an hour after the contest began. So do you think it is fair to make it rated?
Although the queues have been cleared but please declare the round as unrated, 1. Some of us waited for long just to get a wrong answer. 2. Many of us lost the hope to continue the contest and hence going by the past trends we already assumed that it's gonna get unrated.
After only 40 minutes waiting only 15 minutes ???
I submitted 2 solutions to problem 1 , in 9 min and in 11 min. Both got accepted and I got a penalty as well. Is it normal in codeforces or there is some malfunctioning going on? Initially I got 488 score and after 2 min it was reduced to 428. Sad :( MikeMirzayanov
there's a 50 point penalty for resubmission
It's normal. There's a 50 point penalty for resubmission regardless of the correctness of the previous submission.
div.2 a,b aren't good for hackers......
This has to be unrated... many people have lost 30+ minutes because they got WA after a super long queue.
What's worse is that resubmitting needs another 30 minutes to get another WA on the same test case. :(
Queue lag + Server lag = Unrated pls?
I even lagged out while trying to post this lol.
I really wonder why CF is not being open sourced ??(it's not a profit making site, right ?) .The website suffers every other round, I'm sure are many people willing to help CF.
Your contest is very stupid please don't repeat this awful... Your problems are unprofessional.
:| why unprofessional ?
if you cant solve them it's not their fault!
And what is a "professional" problem? I know these problems could be improved, but the big issue about this contest was the long queue
MikeMirzayanov Make it rated please ^^
I hope that this round won't be rated. I waited for the result for half an hour,then the Web told me that I was wrong, then I used 1 minute to fixed the bug .But it didn't work again,and I waited for another half an hour. So I used 3600 seconds to fix a bug .Time is money.We shouldn't waste our time. I'm Chinese student ,I have so much homework to do. I tried my best to save time to do the round .If the round will be rated ,not only me ,but also many other people will get very angry. (My English is not good ,I just tried my best to explain my opinion)
Agree with you
When almost everyone asking for an unrated round has performed poorly
I don't know what's worse: The long queues in every other round of CodeForces or the way CodeForces handles these rounds (Extending the round and keeping the participants in the dark about the contest being rated/unrated, and finally making it unrated after the contest ends — A desperate attempt to ensure that the participants take part in the contest and do not leave it midway).
You nailed it! Instead of stopping confusion as soon as possible they prolong it even further.
What are you doing at the end of the queue? Are you busy? Will you test us?
Good problems but awful contest.
what is test 5 for C ?!
D is just https://en.wikipedia.org/wiki/Catalan%27s_triangle + compute binomial for arbitrary modulo.
"How do we make this question harder?"
"Make it modulo anything instead of 1e9+7"
Always work with a charm. (**Smacks my own face for not keeping another combinational template**)
But how do you handle sum from l to r(and C-type customers)?
Fix the number of paying. Sum from l to r is reduced into two binomial terms.
Div 1. A. I wrote a program that kept on failing pretest 1 even though it worked on my computer. What ended up fixing it was doing my preprocessing before reading the input, which should not have changed anything. This issue took me one hour to catch since it took a good 30 minutes for the test results to get back and another 30 minutes for me to figure it out.
How to solve C and E?
On E I had a
solution which didn't pass. Did someone pass it with such complexity? Also is there a way to solve it without SQRT decomposition?
PS: I think the round should definitely be unrated, because with such a queue it's not worth looking at the easy problems (Div. 1 A and B) as you may do a stupid bug and then be punished really hard. And so a lot of people start solving the harder problems and the chance of them failing the competition increases (for example I did so — I didn't even read B). Also a lot of people probably gave up after seeing such a queue.
PS2: The people which passed E all have the
solution. The constant of my solution should really be large xD
On C operations are random and the queries don't depend on the order of numbers so expect a lot of adjacent numbers to be equal...
msqrt(n)logn + nlog^2n algorithm using binary search tree?
yep (treap)
The constant of this algorithm is too big... You can optimize it into O( msqrt( nlogn ) + nlog^2n ), but I think it still wont pass.
for n=1; this is the string "What are you doing while sending What are you doing at the end of the world? Are you busy? Will you save us?? Are you busy? Will you send What are you doing at the end of the world? Are you busy? Will you save us??" and in this string the 194th char is 'y' but in the given test case it is 'b'! please help me if i have build the string in the wrong way please guide me!
You are missing the quotation marks when you reffer to "f(i-1)"
You missed including the quote " right before and after the nested string.
The 194th char is 'b'. There are 218 characters in string f1, so if you start counting backwards from the last character you'll end up at 'b' in 'busy' on char 194. Make sure to count spaces and punctuation as well.
Edit: I think putting in quotations will fix the problem.
Is D all about Catalan number + DP? I regret ignoring an algorithm which computes (nCi mod non-prime for all i) in O(n) time. :/
There is no argument defending that the round should be rated. Please make it unrated as it has affected many of the participants badly.
Unrated please dude
how to solve div2 B
i th number is of format ireverse(i) for example 10th number is 1001
.................... I'm so stupid.......
ME to realized tooooooooooooooooo late
I tried to generate all this "zcy" numbers, but still got WA 3 )) http://codeforces.me/contest/897/submission/32875222
"zcy number" is actually just a positive number, followed by its mirror.
(Like "153351" is just "153" and its mirror — "351")
So you can iterate from 1 to k, generate the zcy number and add its value. Remember the modulo.
Wow, I think this is such a clever idea!! I did not notice this during the contest...
Instead, I used another method to generate the first 100000+10 zcy numbers, and calculate their sum.
when in queue ( but expecting correct answer ) : PLEASE MAKE IT RATED after getting WA : PLEASE MAKE IT UNRATED
after mains when everyone gets WA : PLEASE MAKE IT RATED
It would really be unfair to rate this round because many of us solved the problems faster than our submission times show; many of us submit, wait for a very long time, and find out our submission was incorrect, resulting in us losing more time than we should have. Today, the time lost was significant; some people lost up to 1 hour waiting for submissions. Personally, I lost a total of about 1 hour of time just waiting for the result of my submission (you don't know whether to continue working on a problem or not until you see the verdict), and I'm sure that many other people are in the same situation as me.
In addition, Codeforces Round 445, which happened a few weeks ago, featured servers that were awfully slow, yet still faster than today's. Given that it was still unrated, today's contest should also be unrated.
I got compilation error after waiting for 30 minutes lol
I learned that my solution for div1 A was wrong after waiting in queue for 45 minutes, which was in fact a minor error I made while copying and pasting a lame ass string from the problem set. It will be absolutely unfair if this round is not marked unrated. Adding 15 minutes doesn't reconcile the situation by any means.
how to solve div1 B?
How to handle N > 55 for div1 A/ div2 C? Won't this overflow 64 bit int?
it doesn't matter just take min(size, INF) where inf = 2e18
This is because k <= 1e18
This blog is starting to get downvotes.
I think the problems were good. The issue was in the long queue, so I don't think this blog should be downvoted
Ya except div 1 A/div 2 C, making sure you copied the right string (especially since there are quotation marks) is a nightmare.
Hello, community!
Recent weeks I did a lot of work to improve performance and stability of the website. It seems my efforts were not in vain. Today and on the previous two rounds the website worked good.
But today we’ve encountered with long queue because of Codeforces default strategy to rerun each solution if verdict is TL in exclusive mode. On each judging machine there are ~3 invoker services which judge your solutions. Exclusive mode guarantees that at most one solution is currently running. Because of many TLs many judging machine worked de facto in single invoker mode (instead of 3 of them). Sorry for long queue. I effectively diagnosed the issue and temporarily disabled this feature (it will be “on” on the system testing).
Sorry for huge queue. During the round I did a lot of work to understand the reasons. It seems I succeeded in it. I personally promise that after website optimizations (and it seems now it is OK), my next step is to improve judging performance to eradicate such situations.
We decided to make this round rated with weight 0.5 (semi-rated). It means that all rating changes will be divided by two.
Thank you for taking part in the round. And many thanks to the writers and the coordinators!
Mike.
First time I seen the semi-rated round...:D
Thanks for your great efforts keep up
But there are people who thought according to tradition the round will be unrated so that we can concentrate on other important works(I have Exam ) and left the contest in middle.Please once think about this
It's so called rotten compromise. Nobody is satisfied ;)
If somebody was affected, he definitely wants the round to be unrated and if not, he would like it fully rated.
Mike, this is just not fair. I took off the contest after waiting half hour for a Wrong Answer. Is just ridiculous to think about doing this round rated, please, think again about your decision.
do you know what 1488 is used for? gtfo
Yes! It means 1st April 1988, the first day when black humor got widely used among many people all over the world!
Umm no, it's a nazi code, idiot. I wonder why people upvoted your retarded picture.
Because they don't see anything nazi in 7 / 44.
you are hilarious
Looking forward to 0.25-rated, 0.625-rated, 0.265625-rated contest for near future!
Why not -0.5 rated contest to spice it up?
Is it quaterated?
Well, that's rational
a π rated or an e rated round sounds preety good. maybe even an i rated round
While the round wasn't perfect, I agree this is probably the best compromise between appreciating our efforts and alleviating the negative effects of early judgement delays on some competitors.
What is the point of making it semi-rated? This is a terrible solution, because:
People who didn't suffer from the long queue and didn't drop writing this contest are going to be sad.
People who suffered and/or decided to drop the competition (because 2 contests back there was a similar situation, but then it was made unrated) are going to be sad.
I think it's better unrated... If any one feels bad in this contest please downvote me...
You don't deserve downvotes. Long queues and system failure aren't the problemsetters fault.
I thought the problems in today's round were excellent. System issues aren't something under your control. If you ask me, you deserve extra upvotes for having your efforts eclipsed by queue problems.
I would say it's better rated... The impact was not that huge, especially in Div2 and the round was extended for 15 mins.
Maybe 3/4-rated would be ideal solution? :)
Semi-rated seems to be a bad idea. If you do well in the contest,you do not like semi-rated. If you do badly in the contest,you do not like Semi-rated,too. So,two choices:rated or unrated. I think it shouldn't be rated because we spend lots of time in waiting the result.
My own opinion: you can emulate the situation of a rated contest by mass-submitting in the problemset sections during idle times (of course, the tempo should be based on the current submitting rate), so you can test and debug the system without interfering the nature of any contests henceforth.
Dear MikeMirzayanov, it's great that your efforts were not in vain. On the other hand I would like to get an answer to the questions I posted in another comment. It might seem that I only wanted express my frustration, but I genuinely would like to know answers to those questions:
Sincerely,
Kroma
PS
also, I've noticed that @moejy0viiiiiv seems to be not affected by his performance in this contest. So now we have additional parameter Z which says that if your rating is >Z then you can be treated in a special way?
Please be consistent. If you make this thing 'semi rated', then at least all people should be affected by this semi change of ranking. Otherwise it is no different that just giving points to people you like a bit more, which makes whole ranking system garbage.
I submitted a wrong solution to Div1.A at first. Half an hour later, I got an Wrong Answer. The same thing happened to me on Div1.B. This led to my very low score. Actually, the total time I spent fixing the bugs was only a little. But I will surely lose lots of ratings because of the system trouble. It's quite unfair. Mike, please think twice.
I can't wait to see non-integer rating.
It makes me lose rank1 TAT
Looks like someone listened to my heavily downvoted comment
I naively thought I will be "expert" after getting rank 170 in Div.2,but when I viewed my rating,it is: Contest rating: 1599 (max. specialist, 1599) 1599 1599 Yes,guys,I'm fine.
I hope that your efforts during this period will make codeforces become 100 percent ready for December interesting Contests ....
Pretests to div1 B were very weak. I passed them using solution, which could have added maximum element in the middle and then add something smaller in the end. I hope that real systests cover such bug :/
Which category is DIV 2C and D?
Div 1. C solution anyone?
Bruteforce (keeping track of constant intervals in a map) works fine, because of random input and queries of type 2. After ~30 queries, we expect that all values between n/4 and 3n/4 have been set to the same value at least once.
Actually, does anyone have a proof that it works? That is, can there be a malicious (N, seed) input for which this solution is really slow? Of course I also believe it should be fast, but have no strict proof. Anyone?
There is a proof.
Here are some quick Mathematica calculations for the limit case, after infinitely many operations (there might be a couple of off-by-ones, but you get the idea).
We first calculate the probability that i and i + 1 are in the same interval. If we only consider the first two operations (where we actually modify things), we split i and i + 1 with probability 2 / n (given that we do operation 1 or 2, we split 2 of the n positions).
The probability that we merge i and i + 1 is
. The factor of is 1 / 2 because this only happens in the second case, 2i(n - i) is the number of pairs (l, r) where we merge i and i + 1, and n2 is the total number of pairs.
The probability psi that i and i + 1 are not in the same interval is now given by the probability that the last event was a split.
The expected number of intervals is just the sum of the probabilities that i and i + 1 are not in the same interval. This returns 43.
The expected number of intervals in a given query is different, because elements close to n/2 are more likely to be in the interval. Taking this into account, the estimated number of intervals covered by a single query is only 5.
EDIT:
I ran some stats on my code, and it seems that the average number of intervals in a query is 8. This is because the first couple of queries will be much larger. After a 100 queries, the number of intervals is roughly 4000, and it first drops below 100 after 6000 steps.
EDIT 2:
If we approximate ps by
and next do the summation 
, this is just a constant! In particular, for any n, the average number of intervals in a query is just 5!
Actually, I meant something different. This
rnd()function is not really random, it's only pseudorandom. (And problemsetters/hackers can control the seed!) What if there is seed for which we get no type-2 operation for very long time? Even if these occur, how do we (surely) know that the input hasn't been crafted so that some early type-2 operations are applied to very short segments?E.g. for N=100000, seed=546202549, we have no assignment operation until 65th query. My submission processes this test in 592 ms, compared to 327 ms on systests. What if there are even worse cases?
I think we can estimate that as well. The total size of the input space is 1028, so roughly 230. If we want to delay the first query of type 2 (which happens with probability 1/4), I think we should expect the worst case to be somewhere around the 60th query. That would imply your example is about as bad as it can get. At least we can be quite sure there are no cases where the first 100 queries are not of type 2.
I mean, this assumes that the generator is random (e.g. outputs random integers or generates a random permutation of all remainders mod 10^9+7), not only "looks random". (For instance, 7 is not a generator mod 10+9 + 7, which makes it a bit worse linear PRNG. Moreover, the generator has a fixed point x fulfilling
. Fortunately, this x produces only type-2 operations. If it produced other types, we'd be screeeewed. Edit: actually not, I forgot this would generate a constant array.)
I'm also "quite sure" there are no bad cases; I can't prove it, though.
Half the people want rated, half the people want unrated.
Makes round semi-rated
Wow, two of my solutions are judged after 30mins and more than 1 hours.
Rated contest. Good game.
m3l4
How To Solve Div 2 Problem C?
It's easy to work with n=0. With n>0, we will take this into consideration:
The string f[n] will be divided into 5 segments:
What are you doing while sending "f[n-1]"? Are you busy? Will you send "f[n-1]"?You can point out which segment the k-th character will lie on in the string f[n].
If it is in segment 1, 3 or 5, the problem will be similar to the case of n = 0.
Otherwise, turn the problem into finding the (k-seg1.size())-th character (if segment 2) or (k-seg1.size()-seg2.size()-seg3.size())-th character (if segment 4) of string f[n-1] — it's a recursive process, which stops when the character you need to find is on segment 1, 3 or 5; or the string you're working on is f[0].
Since maximum k is 10^18, any string length higher than this value will be assigned with a self-defined "positive infinity" value. (Personally I used 4e18).
Complexity O(n) for each query, O(q*n) for the entire problem.
Can you Explain Second Part in More Details With Example
Seemed like I'm getting too complex again :)) Alright.
At first, I'll give some strings out:
f[0] = "What are you doing at the end of the world? Are you busy? Will you save us?"— 75 characters.f[1] = "What are you doing while sending "What are you doing at the end of the world? Are you busy? Will you save us?"? Are you busy? Will you send "What are you doing at the end of the world? Are you busy? Will you save us?"?"— 218 characters.f[2] = "What are you doing while sending "What are you doing while sending "What are you doing at the end of the world? Are you busy? Will you save us?"? Are you busy? Will you send "What are you doing at the end of the world? Are you busy? Will you save us?"?"? Are you busy? Will you send "What are you doing while sending "What are you doing at the end of the world? Are you busy? Will you save us?"? Are you busy? Will you send "What are you doing at the end of the world? Are you busy? Will you save us?"?"?"— 504 characters.Well I did forget something. It'd be wise to save the starting index of each segment in each string f[n] (with n > 0).
For n = 1: 5 segments' starting points are, respectively, 0, 35, 110 (or 35+75), 142 (or 110+32), 217 (or 142+75).
For n = 2: those starting points are, respectively, 0, 35, 253 (or 35+218), 285 (or 253+32), 503 (or 285+218).
In general, they are: 0, 35, 35+f[n-1].size(), 67+f[n-1].size(), 67+2*f[n-1].size()
Let's work with some cases featuring n = 1.
If k = 13, it will lie on the 1st segment. Therefore, the character we need will be the (13-0)-th character of that segment — or
' '(whitespace in this case).If k = 42, it will lie on the 2nd segment. To be precise, it is the (42-35)-th, or 7th, character of that segment (which is f[0]). The character we need will be
'e'.If k = 151, it will lie on the 4th segment — which is, again, f[0]. To be precise, this one is its (151-142)-th or 9th character. The answer is
'y'.More examples with n = 2:
If k = 61, it will lie on the 2nd segment. To be precise, it is the (61-35)-th, or 26th, character of that segment (which is f[1]).
In the string f[1], the 26th character belongs to its 1st segment, therefore we need no further recursions. The answer in this case is
'e'.If k = 81, it will, again, lie on the 2nd segment. To be precise, it is the 46th character of that segment, or f[1].
The 46th character of f[1] belongs to its 2nd segment (which is f[0]).
In other word, the character we seek is the (46-35)-th or 11th character of f[0], which is a whitespace (
' ').Hope all those examples help! ;)
For your convenience, you can re-check the answer by using this site: Identify Character or Line Position — Online
Is “ a single character ???
Yes. Quotation mark is counted here (with n>0, since f[0] has no such characters).
In the strings given above, the first and last quotation mark are for clarifications. In other words, this is what f[1] looks like when we print it out:
What are you doing while sending "What are you doing at the end of the world? Are you busy? Will you save us?"? Are you busy? Will you send "What are you doing at the end of the world? Are you busy? Will you save us?"?Thanks I was utterly confused whole round with ‘ and “
Since maximum k is 10^18, any string length higher than this value will be assigned with a self-defined "positive infinity" value. (Personally I used 4e18).
How does it work? Can you explain some details?
The concept behind that "positive infinity" value is: using a value higher than 1e18 (maximum k) to mark any string length longer than 1e18 characters or any segment start index higher then 1e18.
Why marking? To avoid any undefined behaviour caused by data type.
Like, you add 1 to a "long long" variable with value 2^63-1 (maximum signed long long value), and it immediately become -2^63
(Well, Python users will have no worry about this part, but I still advice using it, as the number will become really huge, and there's no telling if analyzing them could even surpass the time limit of the problem.)
To the strings with +inf value, we know that we will only care about the first 1e18 characters of it, though its true size can be a whole lot longer.
To the segment with starting index of +inf, it means that segment is ignored, as no query would have ever reached it (its starting point is higher than any k value inputted).
My submission for example: 32874794
It was quite buggy when handling +inf with the "thiStart" vector (4e18 * 2 will overflow in signed int64 data type), but it had little impact, since the "secStart" vector was handled correctly.
Actually, it's not O(n)
as the size of the string gets larger so fast, it can reach 1e18, i.e. max K, within ~53 iteration, any farther computations are useless!
so I think that each query complexity is just O(min(log(k), n))
Ah yes, I forgot that. Thanks for re-clarification ;)
'Is it rated?' is outdated. Now it is: Is it semi-rated?
How to solve Div1-B? Is it related to Pigeonhole theorem?
Btw, when you thought that the round will be unrated and then heard that it will be semi-rated
It never matters whether a contest is rated or unrated wht matters is how to solve problem C div 2.My approach was to dp the size of f-ith string.dp[i]=c+2*dp[i-1].Thats how you calculate dp[i] string size. Now you take k keep subtracting to find which segment of the string the character belongs and print it.This maybe wrong as size of dp[i]~2dp[i-1] .So size rises exponentially.So cant store sizes maybe thats where i went wrong.Help pls Problem C
if i>55, dp[i]>=1e18(>=k) ,so you can assume that dp[i]=1e18(i>55). you can seek the answer using dp[] and recursive program.
thanks,got AC by applying your suggestion .I used 55 and got ac .Was there some reason you start initialzing after i = 55 . Thank, Archit
I simply seek i such that 1e18<=dp[i]<=long long max.i=53,54,55 are met.
Algorithm is useless. Let's study computer architecture!
http://codeforces.me/contest/896/submission/32868418
http://codeforces.me/contest/896/submission/32871177
and they passed system tests......
how did they ? i dont understand
My idea is :
i see well isnt it an idea to solve many data structure problems like this one ? even the ones that can be solved like normal data structure problems ?
I'm not sure I understanded you correctly, but to cite several reasons for this incident :
I've read the assembly of Shik and MrDindows and I've noticed both of them abuse SSE SIMD instructions, but Shik's solution doesn't work if you comment out any of his GCC pragmas, same with Dindows' solution. If you comment out
#pragma GCC optimize("Ofast")in his solution, it will not pass in time limit. Though as I believe, SSE instructions are in assembly regardless of these things (-O2 should be enough to include them), so I'm slightly confused why does it improve the perfomance of their code.
The compiler really does not hurry to replace the usual code with SSE without enabled Ofast. https://godbolt.org/g/YVDP8f
Study, at least in order to avoid the errors with constraints, which task author did.
Btw, as for me, it's so obvious to solve it in O(N2), so I expected a lot of such solutions to the end of the contest, and I absolutely do not understand how the author and testers allowed this.
Can you Explains the tricks you did to Pass Time Limit and your complexity is O(N^2) ?
i'm sorry i don't get it .. thanks in advance
Yes, I can.
Tricks from my solution:
1) Fast IO.
2) No branching.
3) Loop unrolling.
4) SIMD.
Explanation:
1) No need to spend valuable time on slow IO.
2) Avoid branching. May be it's hard to believe, but
if (a[i] > x) a[i] -= x;may be slower thana[i] -= a[i] > x ? x : 0;3) Make less loop operations, i.e. less checks whether the loop ended. Also it's better to write loops like
for (i = 0; i < n; ++i)orwhile(n--), it's like you say: "Hey, compiler, I have N loop iterations here, you may optimize them as you want."4) SIMD is may be used, when you can perform several loop iterations independently. I didn't used SIMD operations by myself, but with loop unrolling I showed to compiler, that it can use SIMD. Also pragma`s with -Ofast and SSE are important here.
https://godbolt.org/g/YVDP8f — Here are some comparing of different implementations.
It's also worth pointing out that the compiler itself can use optimizations 2-4 even without your help, Also sometimes it's like a lottery, to guess which optimizations compiler will use, and which will give more acceleration.
in point 2
can you tell me why a[i]-=a[i]>x?x:0 is not branching ?
as far as i know it's similar to if condition
We can see at godbolt that in second case compiler does the following:
That's not exactly what I wrote, but in ternary operator instead of
ifI say to compiler, that it should perform subtraction in any case, but I expect that the value to subtract will be calculated from simple condition, and it can be performed with some CMOV instruction without branching. Compiler optimized my code a bit more to not use zero value, so it's likea[i] = a[i] > x ? a[i] - x : a[i];.But consider, such behavior is nowhere defined, so we can only rely on our experience or expectations. For example, clang compiler even simple loop with
if (a[i] > x) a[i] -= x;optimize with SSE instructions, and it unrolls unrolled loop from my solution even more to process 16 elements of array at each loop iteration.Summarizing, it is necessary to understand that there are some conditions, that can be performed without branching. For example:
a = cond ? b : c;orif (cond) ++val;I'm terrible sorry for this...
I know little about computer architecture, and I haven't even imagined that a O( nm ) algorithm may pass it...
My code uses about 800ms, in order to let some algorithms with larger constant or larger complexity pass, I set time limit to 3000ms.
My fastest brute force algorithm uses about 5600ms, so I thought that no brute force algorithm may pass...
I'm too naive.
Well, if you didn't use any special tricks in this fastest brute-force solution they I'd say that it is quite expected that some participants managed to double the speed and pass. The gap between 5600ms and 3000ms is way too low.
I used some tricks.
The most naive brute force takes about 9000ms in the slowest test case.
I optimized it into 5600ms, and that's the fastest I can do.
However...I didn't know anything about pragma GCC ( In most of the contests they are banned )
It seems that by using this can improve the speed of brute force algorithm from 5600ms to about 2500ms...
how does pragma GCC commands speed up execution time ?
sse, sse2, sse3, ssse3 and others are CPU extensions which allow to compute several values at once, e.g. there are commands to compute sum of 4 pairs of numbers. If you have a simple loop
compiler can sometimes unroll it to process 4 values at once
Which can sometimes be performed simultaneously using special sse instructions. By default it seems that compiler doesn't perform those optimizations, but with pragma it does.
I read this the last time it happened but still couldn't think of it. RIP
I'd still like to say a thank you to the writers of today's problemset; although I was disappointed by how the contest ended up playing out, I would like to say that the problem authors still did a very good job with the problemset and wrote some very interesting problems. The issues with the servers are entirely not their fault, and I don't want them to feel that their efforts are not appreciated. Thank you, problemsetters, and I hope to see more of your problems on Codeforces in the future.
What is semi rated ? And where is that announcement ?
Hey there! I prepared a problem and I would like to share it with the community. Here goes the statement:
Uvuvwevwevwe Onyetenyevwe Ugwemuhwem Osas
Iuisdgsjhsdq is playing a game with little eastern dwarf mongooses.
She gives them an infinite array of strings, f0...∞.
f0 is "Uvuvwevwevwe Onyetenyevwe Ugwemuhwem Osas".
She wants to let more people know about it, so she defines fi = "Uvuvwevwevwe "fi - 1" Onyetenyevwe "fi - 1" Ugwemuhwem "fi - 1" Osas" for all i ≥ 1.
It can be seen that the characters in fi are letters, question marks, (possibly) quotation marks and spaces.
Iuisdgsjhsdq will ask the little eastern dwarf mongooses q times. Each time she will let them find the k-th character of fn. The characters are indexed starting from 1. If fn consists of less than k characters, output '.' (without quotes).
Can you answer her queries?
Do you think this problem is very funny, entertaining and interesting? Share your thoughts so we can make this an even better community by expanding this innovative form of humor. After all, it makes a problem more interesting, right?
This is the 2nd funniest thing I've seen today. Funniest thing is the announcement of this round's being semi-rated!
Where is the announcement ?
See this comment
You should increase image size to 1920x1080, in order to make better contest
That was just a teaser. Gotta make them ask for it before I give it, you know ;))
While I may understand your butthurt because of the use of anime pictures in the statements, at least this time it had readable english, compared to some other rounds, which is a lot more important. It seems anime pictures get a lot more backslash than the bad english rounds.
In the end, it's the author's round and he is allowed to put anything he wants in the statements, as long as it follows the guidelines, and as far as I know they don't restrict the use of "anime pictures". Personally, I think that the statements should be as lightweight as possible, and should probably follow the csacademy style, removing useless stories. As for the announcement, I think the author can put anything he wants as long as it's safe for work.
These are just my opinions, though they could be biased as I also have an anime picture in my profile...
I don't think he mainly criticized anime pictures. They forced contestants to read some long phrase they like. It feels like a headmaster talking about their favorite poems for hours
Yeah, that's also true. But that's what happens almost every second round — unnecessarily long statements with strange stories. That's a general problem, but people don't usually complain so much.
Being forced to read
"What are you doing at the end of the world? Are you busy? Will you save us?"
happens in every second round?
This phrase appeared only in the first statement. Yes, the statements had unnecessary details in them which are irrelevant for the problem. This round it's even more noticeable because of 2 reasons: it's anime, so people will obviously complain, and the stories involve characters from an already established story, so the author tried to reference the story in the statements.
For me, this is no worse/better than the usual "Mike has an array", "Vova is trying to solve an interesting problem", etc.
This is a general problem, not only this round, so I don't think the author should get so much hate.
If I am forced to count number of words in
"Mike has an array, he has queries too, and queries have to be answered"
"I’m like TT. Just like TT. Ireon nae mam moreugo neomuhae neomuhae"
I would be triggered in the same way too. I don't know why you are trying to address it as "anime problem"
Well, what's wrong with this phrase so that it cannot be in the problem statement?
I said it's not a problem with phrase itself
this anime girl in your picture is sooooooo kawaiiiii
best contest 10/10
-every weeb on codeforces
Well done, bro.
I wonder why a character in a Japan Anime has an alien name.
Another example: The shelter in Fate/Grand Order Mashu Kyrielight, neither Japanese nor English.
It's now 3:00 AM in my country, my family all fell asleep except me, and I accidentally laughed out loud because of this; OMG! Hope my parents don't wake up...
fi has no question marks. :(
Чё пацаны, аниме?
I think at least you should let the competitor choose to remove his own participation if he is affected seriously like CF round 415.
+1, some people are affected more greatly by the bad server than others, especially today, where the queue took more than an hour to get resolved by some.
Yeah, that'll be fair.
What you have done to being not affected ?
Why Semi-rated?
This semi-rated thing is ***********...
Who got high score: "Why semi rated!"
Who got lower score : "Why not UnRated!"
CF Predictor DIV 1 DIV 2 and yeah don't forget to divide the change in ratings by 2
This round should be unrated, this semi-rated thing is *******...
I guess from next round people will start asking is it semi-rated instead of it is rated..
Realizations.
Semi-rated? What kind of joke is that? This round must be unrated. 40 min queue for a submission and then get WA! will you continue the contest? :D Btw, with such low quality contests, cf rankings are not important at all.
Why getting run time error on test 7 A — Scarborough Fair code: http://codeforces.me/contest/897/submission/32864871
I think you are reading the spaces in the input by accident
when i submitted your solution. status is Accepted: http://codeforces.me/contest/897/submission/32880683
what happend hare MikeMirzayanov Run_time_error:http://codeforces.me/contest/897/submission/32864871 Ac: http://codeforces.me/contest/897/submission/32880683
You're right I see nothing wrong with your solution
В Жопу CF
What is semi-rated actually..??
I have finally had a semi-decent performance in a round only to find the round to be semi-rated
Ok so after waiting for about 40 minutes in queue, I decided to just go and sleep instead. Then I come back and see they actually made this contest "semi-rated". Are they accounting for the fact that people left instead of waiting for submission to stay through queue?
While everybody complains semi-rated contest, I found a tiny leak from 1B=2D TCs.
My code fails when m>n and c=1, but fortunately every c=1 TCs were given with m=n so that I got AC. Yikes.
To get rank 1 in Div.1,you just need to solve A and B efficiently,and then turn to E and quickly figure out that using computer architecture can get the problem solved.And you come first in Div.1.
(Just joking)
The contest is rated,no? If it's, how much time we have to wait until the rating change? P.S(I am Romanian, don't judge me).
ow men forever newbie is very sad , and forever specialist too!!
What's wrong with the judge now ? Getting "Judgment Failed" when submitting at problems from this contest now...
MikeMirzayanov архив сломался, у половины посылок Judgement failed
Please stop use semi-rated. If use it once, we have to scared to p-rated(0 < p <= 1) even though contest server is completely broken. We have to fight against it and keep pushing F5 until to recover it. It's not programming contest.
BTW, I like Div1C very much, it is interesting problem with PRG. Thanks to writers!
What on earth! What kind of verdict is this?
Top 4 submissions are exactly identical except for some whitespaces.
And all will get accepted. Just wait.
Anyway, the bug seems to be fixed now. They now all changed to AC.
please someone reply me is rated or not??
"We decided to make this round rated with weight 0.5 (semi-rated). It means that all rating changes will be divided by two"
I wonder what happens if someone were to have an odd value of rating change. Will they get rating changes with 0.5 now??? :)))
How did @moejy0viiiiiv manage to have his contest unrated?
I guess because of this.
Oh, but I guess they should have made it unrated for all. Considering the eliteness of moejy0viiiiiv it is clear he would have performed much better had the contest environment been better with the judge not failing midway but it seems unfair to the not so good rated people who possibly have had lost their rating in this contest because of it.
The point is, why he is not affected and everyone else is.
This will create a discussion about the advantages that you get based on CF color
Newbies' lives matter
I think that he wasn't in the contest.
ref: http://cf-predictor-frontend.herokuapp.com/roundResults.jsp?contestId=896
I have him added as my friend so I followed him through the contest. You can even check out his submissions, you can see that he had A/B/C submitted during the contest. Unless CF decides to remove them as well.
ref: http://codeforces.me/submissions/moejy0viiiiiv
I don't see any subimissions from todayish contest. So the only trace is a comment
@Kroma I suggest you give a closer look
Last submission I see on the list is 32405811 (for problem 891D). So either I don't understand what is on this list or there are no submissions from today (yet).
Also, I gave a closer look, but in a different place — I've found submissions in a contest-submissions list, but if you check one link it says that author is "Out of competition:". So maybe that's why I don't see those submissions on the list you've linked?
He participated in this contest.See this - His submission entries
My bad, by default I had checkbox "show unofficial" empty. So those submissions are there, but are unofficial.
That's unfair ... if it is unrated to moejy0viiiiiv it should be unrated for all of us. Just be logical and make it unrated for all.
It seems my solutions are judged much more later than others. The judger skipped my solution and tested all the solution before and after mine. And this solution is judged after the queue is epty(I'm not sure because I don't know the exact time it got judged). In the system test, my solution is also judged after finishing judgment of all other solutions. I described this situation to the admin. Although I haven't received the answer yet.
https://imgur.com/a/rR4vK That was the answer that I received
It's kind of unfair to make the contest unrated for some people and semi-rated for others. If I knew the contest was rated I would have solved B (and not had gone for E), so I also feel affected by the long queue (also I had to wait some time for my A to give WA). If you want to stick with your "SEMI-rated" decision, at least apply it to everyone.
rating doesnt matter much :)
Sure, but it's still unfair to make the round unrated to someone, just because he is a legendary grandmaster :D
Seriously codeforces a queue of an hour and you are making it rated my B was in queue for ~40 minues thats why i left the contest and you made it rated.
What the hell is semi rated if you are not giving us even a chance to show that we are affected. This CF community has now become biased had never thought that things like this will happen but problem setters and cf coordinator have stopped caring about community.
Seriously, wtf is this!!! First educational round and now this round. I abandoned both these rounds in the middle because of some fucking server problem, which I'm not responsible for. Still I get my rating decreased.
One time is OK, but this has become a regular thing at Codeforces, really disappointing :(
Codeforces has new way of avoiding these server problems. Do things like making this contest rated and decrease server load for next contest by making people loose interest on cf and leave cf rounds .
It took 20-30 minutes for each submission to be judged so you could submit at most 4-5 codes during the contest but It's rated cause they added 15 minutes and they think they somewhat fixed the issue during the contest (they didn't or if they did it was too late) which is great banter..
When we can change Nickname?
Christmas.
Can someone tell me how to solve div2 E ??
Please, write here!
Well, it seems like you can just support some structure to track the sequence of segments of equal numbers and their values (for example, using a segment tree) and calculate responses directly. Due to the test random generation algorithm soon you'll have a lot of long segments with equal values so it won't take too long. (Like with probability of order 1/64 on each request all the numbers on [n/4, 3n/4] become the same.) I cannot prove it strictly, but I don't think there is a more efficient way to answer type 4 requests.
so the main secret is in the input generation algorithm
0.5 ranking contest and 1.5 level problems!
Seems funny! Looking forward again fraction ranking contest!
Since the editorial is not yet out. Can anyone help me why this submission fails.
Submission
NVM, found the mistake. Thanks
div 1 E test cases so weak, naive solutions passed...
Tests are good enough.
Are you seriusly?! Why this contests is rated?? Long queue, many participant leave the contest before it end, dude
How is 15 minutes extra justified? If I would've gotten 5 more minutes, I could've solved C and improved my rank by 600 at least. Just 15 minutes extra for 40 minutes queue is totally senseless. I want to know who decided this. Please let me know how you arrived at that number. Thanks.
Could you please post the editorial? It would be much appreciated. Thanks.
In particular I want to read the logic behind B in detail.
UPD: solved this one, but still looking forward for the others
LOL @ the fake segment tree problem...
Also, did anyone solve E the legit way? I looked through all the submissions but it's all GCC pragmas everywhere. I thought of doing segtree with lazy BBST in each node for n log^2 n but I could not code lazy BBST at all. I would be grateful if anyone could share their implementation with me.
I think E is a block dividing problem rather than a data structure problem.
I think block problem is a part of data structure problem.
More than half of the comments are memes. I guess Codeforces became 9gag already.
What are you doing when the contest running? Are you busy? Will you fix the Evaluation machine?
Auto comment: topic has been updated by ODT (previous revision, new revision, compare).
How to solve Div2 D? (How to prove that solution is right?)
How to solve Div2 D? (How to prove that solution is right?)
@ODT
The brief idea is to maintain a non-decreasing list during the whole game, and insert the given number at the correct place. You can notice that after writing down a number, the sum of all written numbers will decrease so that the number of rounds won't be much.
I'll explain the method as well as the proof in details later in the editorial.
"the sum of all written numbers will decrease"
why? if we fill in the numbers from behind and replace one, then the sum may be fewer?
For each round, you either replace an existing number with a strictly smaller one, or fill in an empty sheet. Thus each sheet can be modified at most c times, giving n * c rounds in total. To pass all the tests you only need to maintain 2 such lists from the beginning and the end.
How to prove we are always possible find an answer by this way?
Not a strict proof, but notice that N*ceil(C/2)<=M.
Using this algorithm any cell can be overwriten at most C/2 times, so M moves should be sufficient to fill the whole array.
Was there problem in judgement because one of my solution for div2 C gets AC after the contest though it gave WA as verdict.
I dont know why i got WA in div.2 B on test 3.Your text to link here... i knew that it needs to get mirror number,but failed?
you just do
sum+=data[a]%p, it means you add some value that smaller than p, however it still have possibility to make your some overflow, so you have to dosum=(sum+data[p])%pWhere is the editorial? :)
How to solve Div2 D and Div2 E?
What was the idea of problem Div 2. Problem D "Ithea Plays With Chtholly" ?
What are you doing at the end of the contest? Are you busy? Will you post the editorial?
Auto comment: topic has been updated by ODT (previous revision, new revision, compare).
I'm really confused ('_') ODT .... five minutes ago my rating was restored but now its again updated :??
Someone please explain the Div 2E in detail. I do not understand it from the editorial
I got how the equal intervals are being created. They happen when we perform the operation 2.
zcysky beat trained team！！！