Привет, Codeforces!
В Nov/28/2018 17:35 (Moscow time) состоится Educational Codeforces Round 55 (Rated for Div. 2).
Продолжается серия образовательных раундов в рамках инициативы Harbour.Space University! Подробности о сотрудничестве Harbour.Space University и Codeforces можно прочитать в посте.
Этот раунд будет рейтинговым для участников с рейтингом менее 2100. Соревнование будет проводиться по немного расширенным правилам ACM ICPC. Штраф за каждую неверную посылку до посылки, являющейся полным решением, равен 10 минутам. После окончания раунда будет период времени длительностью в 12 часов, в течение которого вы можете попробовать взломать абсолютно любое решение (в том числе свое). Причем исходный код будет предоставлен не только для чтения, но и для копирования.
Вам будет предложено 7 задач на 2 часа. Мы надеемся, что вам они покажутся интересными.
Задачи вместе со мной придумывали и готовили Михаил MikeMirzayanov Мирзаянов, Роман Roms Глазов, Адилбек adedalic Далабаев, Владимир vovuh Петров и Иван BledDest Андросов.
Удачи в раунде! Успешных решений!
А вот сообщение от наших друзей из Harbour.Space:
Hello Codeforces!
We are excited to announce that the Hello Muscat Programming Bootcamp registration is open! The camp will take place from March 9th to March 15th, 2019, and our early bird discount of 15% is going until December 15th!
This next edition in our Hello Programming Bootcamp will run in parallel with the traditional Moscow ICPC Workshop — both Bootcamps’ contests will be identical, and contestants will be able to see their position in the General Leaderboard. Every day, both camps will be competing simultaneously, 4,000 kilometers from each other!
We would also like to remind you that the deadline for applying to the Master’s in Robotics programme scholarship will close November 30th, so we encourage you to check out the website to see the all the requirements and apply!
UPD: Поздравляем победителей:
| Место | Участник | Задач решено | Штраф |
|---|---|---|---|
| 1 | I_love_Tanya_Romanova | 7 | 157 |
| 2 | theodor.moroianu | 7 | 330 |
| 3 | halyavin | 7 | 361 |
| 4 | Radewoosh | 6 | 158 |
| 5 | palayutm2001 | 6 | 190 |
Поздравляем лучших взломщиков:
| Место | Участник | Число взломов |
|---|---|---|
| 1 | halyavin | 131:-8 |
| 2 | zdw1999 | 55:-4 |
| 3 | MarcosK | 42:-1 |
| 4 | ismagilov.code | 64:-48 |
| 5 | garipov.roma | 48:-24 |
Было сделано 1354 успешных и 1065 неудачных взломов.
И, наконец, поздравляем людей, отправивших первое полное решение по задаче:
| Задача | Участник | Штраф |
|---|---|---|
| A | I_love_Tanya_Romanova | 0:02 |
| B | Dalgerok | 0:04 |
| C | I_love_Tanya_Romanova | 0:08 |
| D | hitman623 | 0:15 |
| E | lqs2015 | 0:11 |
| F | ko_osaga | 0:47 |
| G | RomaWhite | 0:08 |
UPD2: Разбор опубликован
Yoo!! Another comment by harrypotter0
It's been so long since MikeMirzayanov appeared as a problem setter(except div. 3) :)
except for div.3, yeah! :)
Mike was problemsetter at Mail.Ru round 3, which was on Sunday.
Idk if 3 days are so long
Oh, I missed that. Then it's a long time before Mail.Ru round 3.
I apologise, I missed that too. Good luck for the contest!
Can something be done please to avoid clashing with the SRM?
Good
(int) Round 55 Div. 2 = 28 November :D
ceil?
Судя по организаторам, раунд будет очень даже хорош
а, нет(((
We haven't seen div3 for long time.
Everytime I said "Hope we all have a good rank and I want to become expert"then got a bad rank,so I say hope everyone enjoy this contest this time :D
Yoo!! Another contest by Pik Mike How To lomk Him ? we Didint have Div3 For Long Time And Is It Rated ? that isn’t something to be happy about
"Is it rated?" Life Hack: Read the title!
Clashes with Carlsen vs Caruana tie-break
Auto comment: topic has been updated by awoo (previous revision, new revision, compare).
challenging problems are today.
How to solve G??
Make a flow network and represent each edge as a vertex. Connect each "edge" to the source with capcacity the weight of the edge. Connect every vertex V of the graph to the sink with capacity a[V]. Finally connect each "edge" with the 2 vertices it connects with capacity Infinity. The answer is sum of weights of all edges — min cut in this network.
Can someone elaborate why this works and what's the intuition behind it?
Think of it like this. Initially we have taken all "edges" to form our subgraph. If we cut an edge connecting source to "edge" it means we don't take the "edge" in our subgraph, similarly if we cut an edge from vertex to sink we don't take the vertex in our subgraph. If an edge is present in our subgraph then the 2 vertices it connects must be in our subgraph too. If an edge is not present in our subgraph then atleast 1 of the 2 vertices it connects must be out of our subgraph. In this network any cut will fullfill these rules(try to see why this is true) and to maximize the answer we just need to find the min-cut.
I can help you with that :P. In fact, the book "Algorithm Design" by Jon Kleinberg, Eva Tardos can be truly revealing about that.
Go to page 396(Chapter 7 Network Flow). This problem is known in the literature as "Project Selection". Enjoy.
PD: The book i pointed to you is amazing, there are other non straight forward applications. My suggestion: Read other chapters as well.
Is Dinic's algorithm enough for problem G? Seems the complexity should be something like O(n3).
No, in this case graph has special structure. All path from s to t has length 3, so there won't be n phases of using BFS, so it has the other complexity. As I understand this is something like O(nm).
Bonus: i am not sure, but as i understand if you use scaling complexity is better
halyavin is here.
that's not fair I actually wasted 15 min on A because i cant understand the second test case until the Global announcement.... maybe I would have solved C I'm just putting final touches on my solution
Really good contest, tricky A though :D
Can anyone share their approach for G? I saw G started getting accepted solutions much before E and F started getting accepted.
Make a flow network and represent each edge as a vertex. Connect each "edge" to the source with capcacity the weight of the edge. Connect every vertex V of the graph to the sink with capacity a[V]. Finally connect each "edge" with the 2 vertices it connects with capacity Infinity. The answer is sum of weights of all edges — min cut in this network.
yeah tricky with false test explaining
Is Dinic's algorithm enough for Problem G? It seems that the complexity is something like O(n3).
By the way, I got accepted with Dinic's after I got accepted with Ford-Fulkerson. Would someone tell me which submission the system will use to judge?
The last submission will be judged
If the first passes it will be considered. Else the second
No
The earliest submitted solution to get AC on systests will be considered.
That's sounds reasonable. Moreover, if the first one failed the systests while the second one passed, will I get penalty for the first one?
Yes, you will.
Ok. Understand it. Thank you!
That's odd though. On "normal" codeforces rounds, only the LAST submission to pass the pretests in judged on systests and that one submission will matter towards your score.
How come it's different for "educational" rounds?
Educational system is a modification to icpc system, not the cf one. Over there all the submissions after AC don't count, the same is here.
How to solve D?
Fix the two vertices with lowest a as the ends of the diameter. Put all vertices with a >= 2 between them. Then check if you can put the remaining vertices (they have degree 1) on them.
First, if sum(a)<2*(n-1) then you can't build a connected graph at all. Otherwise you need to build a tree that is maximally close to just a path. My approach was like this: first, build a path from vertices with degree 2+, then add a vertex with degree 1 (if there are any) to the first and last vertices of that path. Add other leaves anywhere respecting degree constraints.
Will I get a penalty if I hack unsuccessfully?
No.
And what if I hack successfully? Will I get a extra bonus or only the one who is hacked will lose his "accepted"?
You can indirectly get an extra bonus: if you hack someone that is higher than you in standings and he falls below you, you get a better position in final standings and a bit more rank.
Many thanks~ However, I think I should go to bed now because it is already 1: 30 in China now...
No one cares.
Would that really matter though? There will be system test to eliminate those wrong answers if nobody hacked them anyway right?
In Educational rounds the systest consists of the hack inputs used during the open hacking phase. If the input needed to hack the code is very specific, it's possible that it passes the other hacking inputs and thus the systests.
I'm still sort of confused about the Educational rounds. Will there be a system test phase after the open hacking phase? Or the contest just finishes once the open hacking phase finishes?
I see, that makes a lot of sense!
thanks~
could anyone please tell me what is this hell -> Diagnostics detected issues [cpp.clang++-diagnose]? only due to this i could not submit my B problem. although my code is working fine on my system.
Maybe this will help https://mirror.codeforces.com/blog/entry/55902
It says that "Diagnostics~" issue might be due to undefined behavior.
Make sure that you never try to access index -1 of an array or something like that.
how to solve C??
May God save me.
HELP ME PLS!!!
PROBLEM E
This submission 46340090 Wrong answer on test 11,
but on any other compiler 11 test — ok (and on my local machine).
I do not understand why and I think that the problem is in the cf compiler,
help pls.
//11 test//
input:
18 1
3 3 1 3 1 1 3 4 2 2 2 1 2 1 4 1 4 3
ans:
9
r = g[a[i]][r]; //UB
and...??? why is it wrong???
r always must be < g[a[i]].size(),
because g[a[i]].push_back(i) and r = lower_bound
=>
r always < g[a[i]].size()
why UB???
dp[x][j — 1] = ind; --> dp[x][j — 1] = lower_bound(g[x].begin(), g[x].end(), ind) — g[x].begin(); FeelsBadMan
Ohh, sorry.
Thank :3
Can anyone tell me why this code has been blowing up with undefined behaviour on Codefroces servers ?. It works fine on my machine and ideone.com . My Submission
Just check your IF statement carefully. you put a ';' just after the if (i < matrix[j].size()) statement which caused the wrong answer.
what can be the hacking input in problem B
100 GGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG
169 GSGSGSGGSGSSSGSSGSGGGSGGGSSSGGSGSSSSSGGGGSSSSGGGSSGSGGSGGSGGSSGGGGSSGSSGSSSGSGGSSGGSSGGSSGSGSSGSSSSSSGSGSSGSSSGGSGSGGSSSSGSGGSGSSSSGSGGSSGGGSGGSGGSSSSGSSGSSSSSGGGGGGGSGS
123 GSSSSGGGSSSGSGGSGGSGGGGGGSGSGGSGSGGGGGGGSSGGSGGGGSGGSGSSSSSSGGGSGGGGGGGSGGGSSGSSSGGGGSGGGSSGSSGSSGSSGGSGGSGSSSSGSSGGGGGGSSS
183 GSSSSGGSSGSGSSGGGGGSGSSGGGSSSSGGGSSSGSGSSSSGSGGSGSGSGGSGGGSSSGSGSGSSSGSGSGSGGSGSGGGGGSSGSGGGGSGGGGSSGGGSSSGSGGGSGGSSSGSGSSSSSSSSSSGSSGSGSSGGSGSSSGGGSGSGSGSGSSSSGGGSGSGGGGGSGSSSSSGGSSG
1000 SSGGSGSGSSSGSGSGSGSSGGSGSGGSSGSGGGSSSSGGSGSSSGSSSSSGGGSGSGSSSSGSGGSGGGSSSGSSSSGSSGSSSSGSSGGSGSSSSSSGGSGGGSSGSGSGSSGGSSGSSSGGGGSGSSSGSSGSSGGSSGGSGSGSGGSGGGSSGSGGSSSGSGGSSSSSSSGGSSGGGSSGGGGSGGGSGSGGGSGSSGSSGGSSSGGGGSSGGSGSGGSSGGGSGSSSGGSGSSGSSSSGSSSGSSSGGSGSSSSGSGGSSSGSSSGSSGSGGSSSSSGGGGGGGSSSGSSGSGGSGSGGGGSGGGGSGSGSGGGGGSGSSGGSSSSGGGSSSGGSSSSGGGSSSGGGGSGGGSGGGSSSGGSSSSSSSSGGGGSSGGGSSGSGGGGSGGSGSGGSGGSGGSGSGGSSSSGSGGSSSGGSSGGGSSGGGSGSSSSGGGSGGGSSSSSSGGSGGGSGGSGSSSSSSGGGSGSGSSGSSGGGGGGSGGGSGSGGSGGSSGGSSSGGSSGGSSSGSGGGGSGSGSSGSGSSSGGGGSSSGGGGSGSSGGGSGSSSGSSGSSGSGSGGGGGSGSGGGGGSSSSGSGSGGSSGSSSGSGSGGGGSGSGGGSGGSSGSGGSSGSSGSSSGGGSSSSSSSGGSGSGGSSSGGSSSSSGGSSSGGSSGSSGGSSGGGSGGGGGSGGGGGSSGSSSGGSGGGSSGGGSSSGGGSSGGGSGSGGGGSSGGSSSGSGSGSGGSSGGGGGSGGGGSGGSGGGSGGSGGGGGSSSSGGSSGGSSSSGGGSGSGSSGGSSSSGGSSSSGGSGSGGSSGSGSSSSSSSGGGGGGSGGSGGGSSSGGSGGGGSSSSSSSGGGSSGGSSSGGGSGGGSGGSSSGGSGGGGSSSGSSSGGSGSGSSGSGGGSGGGGSGSSSSSSSGSGSGSSSSSGSGGSSSSSGGGSGGSGGSSSGSGGSGSSGSGGGSGGSSGSGSSSSSSGSSGGSSGSGGGSGSSSGGGGGGGGGGGGGG
Another HackerForces contest!!!
Hello! idk if this is intended but I found out I could submit for practice before contest ended, this seemed really cool however when I get wrong answer I can see the test data!!! which means I can hack any code I want that gives Wrong Answer, is this the intention for the educational CF?
46340916
Why does Binarysearch WA for C?
I think the function is not strictly increasing
oh ;-; thanks!
For problem C I wrote code in c++ in ideone it ran for sample test cases well.But on submitting it is giving runtime error with "Exit Code is -1073741819" for the very first test case I don't know what the exit code means? Can anyone explain why and when that exit code occurs?
finally I still can't hack this 1965ms O(n^2) 46324307
50000*50000 just cost 2s.. amazing...
but I hacked another solution with java : 46336111
but in fact this solution's time cost is about 2400ms
so this tells us drop java and pick c++ nowjust joking XD
Intended solution for F? I can only bound my solution with
, where S is the sum of lengths of all strings (that also may be slightly wrong, to the better or worse), but the constant is incredibly low and I can't come up with a worstcase. It passed with 233ms or so.
(I will use N to denote size of a trie)
Don't know about intended; I did following DP: [subtree that we have to solve (N)][Depth of the first taken vertex on path from subtree root to the overall root(N)][how many vertices do we want to take in this subtree(K)]. So I have N * N * K states, and in order to solve the state I'm running knapsack over all sons of the vertex. Adding up those knapsacks for all vertices, we'll get N * K * K over whole tree (each vertex is a son of exactly one other vertex, and it will be considered in K * K possible transitions in knapsack), plus we have to run it N * K times for all possible combinations on two other dimensions of DP, which leads to the bound of N2 * K3 with a really low constant.
hi there i have a problem. recently, i got a message that said your round is going to be skipped. bu at the end, it said that if you think there is a problem, you can send a message here. i was getting +98 from this contest, so i think there was no reason to skip this round by my self. thanks.
When will tutorial be available?
Auto comment: topic has been updated by awoo (previous revision, new revision, compare).
python 6-lines AC solution to problem B
https://mirror.codeforces.com/blog/entry/60059 for more
Где топ участников и топ взломщиков?
Ждем...
Завтра ближе к обеду будет, из поезда что-то не получается сгенерировать
I tried to solve problem C with Binary Search but failed.. whats wrong with my algorithm? thx :)
here is my submission 46606310
I think there's some bug in the checker of problem D (Test case number 18). Diameter of the following code matches with the answer diameter, but it is still giving wrong answer with the comment " Given diameter is incorrect 388 389"
https://mirror.codeforces.com/contest/1082/submission/46993586
Please look into it. Thanks in advance.