GCJ will start in less than 24 hours.
Live Stream: https://www.youtube.com/watch?v=rh_EYIu7Ztc
Unfortunately I don't have time to edit this table, but rowdark will participate, and JoeyWheeler won't.
Participants:
| GCJ Handle | Country | CF Handle | CF Rating | TC Handle | TC Rating | R2 Rank | R3 Rank | Previous Finals |
|---|---|---|---|---|---|---|---|---|
| Gennady.Korotkevich | Belarus | tourist | 3503 | tourist | 3766 | 1 | 3 | 2014 (1st) |
| vepifanov | Russia | vepifanov | 2963 | Kankuro | 3232 | 33 | 10 | 2011 (8th), 2012 (4th), 2013 (17th), 2014 (8th) |
| qwerty787788 | Russia | qwerty787788 | 2947 | qwerty787788 | 2899 | 24 | 20 | |
| rng..58 | Japan | rng_58 | 2941 | rng_58 | 3468 | 3 | 1 | 2010 (7th), 2011 (1st), 2012 (19th) |
| yeputons | Russia | yeputons | 2783 | yeputons | 3092 | 103 | 19 | |
| Merkurev | Russia | Merkurev | 2758 | Merkurev | 2546 | 39 | 21 | |
| bmerry | South Africa | bmerry | 2722 | bmerry | 3295 | 60 | 11 | 2008 (3rd), 2009 (10th), 2010 (21st), 2012 (6th) |
| peter50216 | Taiwan | peter50216 | 2637 | peter50216 | 2974 | 2 | 24 | |
| TankEngineer | China | TankEngineer | 2634 | OierRobbin | 2636 | 23 | 26 | |
| AngryBacon | China | BaconLi | 2619 | AngryBacon | 62 | 16 | ||
| tkociumaka | Poland | tomasz.kociumaka | 2613 | tom612pl | 2717 | 8 | 2 | |
| dzhulgakov | Ukraine | dzhulgakov | 2607 | dzhulgakov | 3168 | 142 | 14 | 2008 (44th), 2009 (5th) 2012 (9th), 2013 (13th), 2014 (9th) |
| simonlindholm | Sweden | simonlindholm | 2570 | 20 | 8 | |||
| semiexp. | Japan | semiexp | 2507 | semiexp | 3005 | 6 | 27 | |
| pashka | Russia | pashka | 2488 | pashka | 2862 | 11 | 15 | 2008 (84th), 2009 (17th), 2010 (8th), 2011 (7th) |
| ishraq | Australia | JoeyWheeler | 2482 | izrak | 2086 | 85 | 18 | |
| betaveros | Taiwan | betaveros | 2463 | betaveros | 2181 | 384 | 17 | |
| Romka | Belarus | Romka | 2452 | _Romka_ | 2474 | 446 | 28 | 2014 (10th) |
| iwi | Japan | iwiwi | 2427 | iwiwi | 3100 | 7 | 6 | 2008 (59th), 2010 (9th), 2014 (15th) |
| tczajka | Poland | tomek | 2393 | tomek | 3204 | 18 | 7 | 2003 (4th), 2004 (4th), 2006 (6th) |
| JAPLJ | Japan | wrong | 2367 | wrong | 2465 | 67 | 22 | 2013 (23rd) |
| wuzhengkai | China | wuzhengkai | 2343 | wuzhengkai | 2575 | 135 | 23 | 2014 (23rd) |
| fagu | Germany | fagu | 2306 | fagu | 2259 | 126 | 12 | |
| kevinsogo | Philippines | kevinsogo | 2236 | kevinsogo | 136 | 9 | ||
| Xhark | South Korea | Xhark | 2132 | Xhark | 1677 | 35 | 4 | |
| linguo | UK | linguo | 36 | 5 | 2008 (74th), 2010 (22nd), 2011 (22nd) |
If you want you can add this into tomek previous finals- 2003(4th), 2004(4th), 2006(6th)
Thank you, it's difficult to find old results.
Your welcome. And best wishes for final.
The size of your graphic list is disturbing the page Please fix it.
Is it good now?
No yet! perhaps in monitor like mine (less than 10 inch)it doesn't appear well.
my ****** is longer than your screen width -_-
Just to clarify, screen length measured diagonally.. so I assumed height = width . Doing calculations, we get screenwidth < = 10 / √2 = 7inch
What less than 10 inch screen is square? Seems like a bad assumption.
why people assume ****** means that. I meant my "kitten" , is longer than your monitor, so it has 6 '*' , instead of 4 and 5
I don't think this is a place to talk about how big your "kitten" is ...
It is well-known that the people who brag about the size of their kittens have the smallest kittens ;)
unfortunately I must agree with you
We still have no idea why these competitions don't attract a whole lot of female participants.
Hope you guys the best of luck :D
Auto comment: topic has been updated by rng_58 (previous revision, new revision, compare).
Let us see how many different languages linguo uses today.
In Code Jam 2009, he used a total of 14 different languages, uptil Round 2.
I think iwiwi's TC handle should be [[iwi]].
In this context, the optimal strategy for tourist to make headlines could be to tie his right hand behind his back, and then still come 5th or thereabouts.
It's interesting to recall Dota 2 (computer game) International finals, which I observed in the same Seattle city just a week ago. The prize pool was $18 mln, and the the members of the winning team got about $1.2 mln each.
And now compare it to $15 thousands for the winner of Google Code Jam.
I make no judgements or comments here. Merely giving the info.
So where did the prize pool come from?
Mostly — players. They were buying a special virtual book, and Valve donated 25% of the proceeds to the pool.
The book is not required to play Dota, it gives some virtual things like hats or costumes, tickets to watch the tournament games online and a number of games in the game.
The audience of the game is tens of millions. There are ~600,000 playing at any single moment of time concurrently.
Good luck everybody :)
Who is apiad and why he's not mentioned in the list? He ranked 13th in round 3
Too young jqdai0815
I'm also too young to attend so my name really shouldn't be on this list...
I think tourist will win.
Why would you make such a hasty conclusion?
It is not easy to win finals, you need bit of luck. Last year as far as I know petr mentioned that eatmore was close for large data set in problem F and problem E large set has been expired for him. From last couple of years Petr failed in problem C in Yandex, Facebook hacker cup 2014 result was close too, as tomek is not involved in competitive programming regularly and he still do well that's quite awesome. In round 3, rng_58 won, he has won the competition previously too. Yes tourist has been in great level now but about finals you can't say. He is good but still with little bit of bad luck he can fail or somebody other can outperform him. If you see the list there are many onsite champions and ICPC champions are there. And some are new but has good progress in recent years. So you never know who will win. Its about the competition hours.
It would be really cool if linguo wins.
Why?
For the glory of Satan of course.
This
He has close to nil competition programming experience as compared to the rest.
Oh I thought it was because of this
Hahahaha, that's actually a better reason.
You may miss rowdark.
~
Is there a scoreboard link that shows current results other than youtube link ?
I suppose there will be one here, as for online rounds: https://code.google.com/codejam
Please everyone that will compete today, if you have the option for screencast, do this! It's a great orportunity for "us"(coders) to see how the bests programs.
All the best rng_58 :)
Anyone willing to bet on rng_58 to win the Finals ? :D
yep I think rng_58 will do :)
meeeeeeee
I would like to see yeputons win this year :p Good Luck to him
rowdark also made it to finals. And my TC handle is cvcvb_lyp. :)
Auto comment: topic has been updated by rng_58 (previous revision, new revision, compare).
Very red page :]
Started. Related links:
may I ask if the large constraint for B is N<=10^6 or N<=106? it just looks similiar
10^6 (one million).
Angry bacons keyboard is soo noisy
He use personal mechanical keyboard and type very fast.
He sounds like this.
Seems Something is going wrong with tourist!!!!
Not sure about that , because Penalty equal to the time of the last submission and not the sum of submissions like in ICPC so i think he is keeping large input to check later .
Just watched him leaving his chair and walking around. Besides noticed him reading the question for quite a while!!
Precisely. There is very little value in submitting your solutions to large subproblems early. In the finals a very valid strategy is to solve the small subproblem separately, then to solve the large one using an efficient algorithm, and then to run both implementations on many random inputs to make it more likely that your solution of the large subproblem is correct. You only download and submit the large subproblems after what you expect to be your last solved small subproblem. One extra benefit: your opponents receive less information.
And it's probably worth mentioning that this strategy is pretty much only viable for tourist.
Unless you're a big favorite, minimizing the variance is not the way to go. You're losing your time in the hope of increasing the accuracy. Also, there's a huge psychological benefit of being done with the problem. If you have three problems that you're unsure of, then it will be harder for you to focus on the new one.
Sure, the strategy does require some balls :) but many of the contestants there should be able to pull it off.
I would certainly go for a separate solution for the easy anyway, at least in problems where it's easy and fast to write it (which is the case for most problems this year). My experience with CodeJam finals is that consistency is what matters the most.
Once you implement a solution for the hard subproblem, running the random tests is often only a 1-minute overhead: 30 seconds to write a simple generator and 30 seconds to do some bash magic. IMHO it's a pretty small cost to pay.
(If I wasn't clear enough before, note that the overhead in writing the generator and running the tests is the only extra cost I would be paying. Instead of, for example, solving and submitting in the order A1 A2 B1 B2 C1, I would solve in the same order but submit in the order A1 B1 C1 A2 B2, or possibly A1 B1 A2 C1 B2, if everything goes well. If we ignore the overhead from running the tests, the time of the last submission is the same in both cases.)
I'm still saying that's a very bad strategy for almost everyone there. It may sound good on paper, but it's horrible in practice.
In most cases, the information on the leaderboard you provide is worthless unless you're in the lead, or you've solved a hard problem.
You have to write a separate solution to easy. It may take only 3-5 minutes sometimes, but it's still a significant cost. I agree that sometimes it may be a good idea even if you're behind. But it's definitely very far from being a golden rule.
30 seconds for generating test cases? Um, maybe in some trivial cases. The cost of switching your mindset, the potential cost of doing a bug in test generator. These things add up.
And again, most people have to click that "submit" button in order to get their mind fresh again. Sure, that's something that could be worked on.
I'll also add "If I wasn't clear enough before" clause: I'm not saying that doing that for any problem is a bad idea. I'm just against advocating it as an near-optimal strategy for every GCJ onsite participant.
Yeah, I agree with all you are saying. I also never claimed that this is a near-optimal strategy for everyone, I just said that it is a valid strategy worth considering, and it is certainly something I would strongly consider.
It is still worth to consider each problem separately. Sometimes, it is obvious that if a program solves the easy input, it's correct. In such cases, why bother testing :) but in many problems the hard can be tricky and failing it because of some dumb bug is very costly.
Yeah, I guess we both agree on everything.
It's the usual problem of trying to express in a short and precise way. Which makes everything sound much less subtle than it really is. Ahh, teh internetz :)
What means "Time expired"?
It means that the contestant downloaded the large input but failed to submit the correct output within the allowed 8 minutes. If you see "Time expired" in the scoreboard, it means that the contestant cannot solve that subproblem any more.
He downloaded a large input, but did not send a output in time.
Why they block chat on stream? =( On the ACM ICPC finals live stream chat was blocked too.
But why? I really miss it =)
What a technique of tourist ! First he is solving the small input problems and make him sure that if it is correct or not? I think he will submit all the large input problems after solving all the problems at a time.
Last year the stream was really better! This year the camera is always in the same place!
I think it was better showing each contestant and computers.
Is he or she? xD
OVER, bmerry First Place!! CRAZY CONTEST
That's not final results :) It will be announced later.
>implying solutions to large subproblems never fail
What happened to tourist's penalty?
He resubmitted D-large, apparently.
Resubmitted D-large.
You can resubmit as many times as you like during 8 minute window, but the time of the last submit will count.
no extra penalty 20 minutes?
What do you think,are there gonna be any upsets/surprises in the final results this year ?
tomek is the beast.
GG gennady why did I ever doubt you?
Yet again, tourist is the Google Code Jam Champion! :D
Not Again :D :D
Congratulations to tourist for back to back titles :)
yeah, He successfully defends his title, congratulations to tourist
That resubmission on D saved tourist. Well done bmerry and rng_58.
I think tomek also did really well. He does compete only in few major competition. And according to his submission time it looks like he took more time in B which may result in applying optimization algorithm in D and E. May be rustiness cost him :)
Yeah, he did really well, that one test case added last night failed his solution for D and similar situation with his E.
And I think I shouldn't say it luck but rank 1 always finds tourist.
im wondering how did he realize his submission was wrong so that he did a resubmit?
They have showed a part of the process on the stream. This may not be entirely accurate, but from what I have seen it looked as follows:
It did not seem to be the case that he ran his solution, then found and fixed a bug in his code, and then resubmitted. It seemed that he had two versions of his solution. As soon as the first one produced some output, he submitted it. The second one (presumably one that was slower but tried more possibilities? maybe one for which he wasn't sure whether it will be fast enough?) ran longer but also finished in time. When he compared the outputs he saw that some of the outputs the slower solution produced were slightly better. This was still within the 8 minutes, so he could resubmit the new output.
He coded two solutions!?!!?!?!!?
This guy is crazy
They may differ in some constants.
tourist's screencast: https://youtu.be/TO1Y_BrAJHc?t=3h48m16s
(Starting from his first submission of D Large)
Interesting, tomek submission of D large, Simulated Annealing, 99/100 test cases pass, last one added yesterday.Really!
He used Simulated Annealing for E. as well. It didn't pass unfortunately.
Awesome tourist Comeback !
Does anyone know the practice round problem from this years code jam? Apparently it was a finals problem they found out was similar to a recent problem and had to remove it from the set. (According to the broadcast)
I really like one thing from tourist and learned from it that sticking to problem you think you can solve instead of reading different problems and try them. Last year he sticks to problem E not moving for F small and this year he stick to problem D not moving for F small. Yes you need a bit of luck but this is the great things to learn.
This strategy works if you are, not saying tourist,but at least an IGM.
Of course it is not for me for a time now but it is good point to note for the future.
tourist is becoming Messi of CP, winning every major individual trophy possible and multiple times :)
And probably, ITMO university is Barcelona :D
I would find this analogue accurate if tourist had never won [ICPC] World Finals, like Messi. :)
True in National Level:)
tourist has never won any cup as Belarus national team :D
In the world of competitive programming, nation is more like an university. I would love to have some national team programming competition though. Something along the lines of "Battle of Giants", but on a much larger scale and with a single team per nation.
He is better. He's no real opponents like Critsiano Ronaldo for Messi. Also he's won World Cup for two times, but Messi hasn't.
tourist == mangus carlsen
tourist == magnus* carlsen
If he was a bit more dominating then he could have been compared with Ronda Rousey of UFC.
Where is wuzhengkai? Did he participate?
https://code.google.com/codejam/contest/5224486/scoreboard# Why tourist is 3rd here?
it is not the final result. as you can see, there are ?s on the scoreboard
Because they didn't update the page yet.
Current TC handle of Xhark is RRx.
All I see is blonde, brunette, redhead...
And this should be expressly forbidden in the Ts&Cs:
Don't understand the joke about last four. "Taking over the world"?
It's not beyond plausibility that for tourist it's all in a day's work.
This is the second year he's won and it has felt like such a close call. What extra skill did he posses to be able to outperform a 3 time IMO Gold winner? I thought being very good in discrete maths and being able to prove your programs with heavy use of induction and many other techniques were the key.
At any rate, congrats to tourist. It was a phenomenal match, and such a close one. Tomek was in awesome form to not have competed much recently, also. I guess it is a bit like riding a bike?
tourist is like CR7 :D Best in the world. Congrats Gennady
I think you meant CR7 is like tourist ;)
Yes :D lets just agree to the disagree ;)
Scoreboard has been updated (also with solutions!)
It looks like tourist used the strategy everyone speculated for A, B and E (His solutions for small and large are identical).
tomek's Simulated Annealing solutions to D and E are really interesting too.
-
Oh and for a good laugh, look at the filename for tourist's solution to C-small.
Where is my cookie for predicting everything exactly as it was?
Here you go!
STEP5 is another participant, that is missed in the table.
He is step5 in Codeforces.
Pretty disappointed that up until now there are exactly 0 comments/blog entries regarding to Distributed Code Jam.
My main feeling about distributed code jam: the problems were too hard. Anyone who solved D or E would have gotten first place, and no one did. I did my best on E, but didn't manage even after trying for 3 hours, ending up with >300 lines of code and an awful amount of special cases... Even B was only solved fully by two people. (To be fair, perhaps just no-one looked at D. I don't know how hard that one was.)
Hopefully next year, when people are more used to the format, and if the problems get less heavy on implementation, things will be more interesting. I also hear they are planning to run the solutions on more machines then. :-)
linguo — was the only person to use the programming language Brainfuck in order to complete a set. :O source
who won?
tourist
where is the standing?