when final results going to be avaliable?

"hz" in Lunch Scheduling answers?

I have different answers for Gentrification, anyone else?

Me and sankear all differ in problem C. Our outputs: http://pastie.org/9877127

Moreover, me and sankear differ only in test four, my answer is larger by 1, and your answers are larger then ours on significant number of tests.

Well, you also have to find how to account for weights. Doesn't seem trivial to me.

Is it really classical and wide-known theorem? If yes, then I'm totally fine with this task. Otherwise, while not denying that it's good for me that I learned a new theorem today, I personally find it rather bizarre to give 35 points on the basis of how well can one use Google. Which, apparently, I suck at.

UPD: I see, so it's a well-known and pretty widely-used theorem. Have absolutely no problems with the task in this case. Thanks for the feedback!

Since this comment is on top-level, I'm not 100% sure if this was meant to be a reply to my original question. It really looks like it is, so I'll go ahead and answer it, even though I don't consider myself a strong redcoder.

So, you are finding pretty bizarre that there is a possibility that strong redcoder doesn't know that theorem. In fact, the theorem itself doesn't even matter. Let's consider the possible ways of how can one learn it:

1. Was taught it in some kind of school/camp/etc.
2. Read about it or tried solving a problem which required the knowledge of it.

I believe that's pretty much is it in general terms. So let's consider what is required for these ways to occur.

For the first way, not everyone have that privilege, which requires an infrastructure and coaches at schools and universities, willing to devote their valuable time on coaching. Please, do no take that for granted. Even if everything is in place, it's not guaranteed that you will learn it from this way.

As for the second way, there are two scenarios around it. One is to invest a lot of time in training, which is usually expressed in solving problems from various online and onsite competitions on your own. I do not think it's bizarre that some people simply do not have enough free time to be willing to do this. For me personally, competitive programming is pretty much a hobby rather than the thing I seriously devote myself to ever since I changed university in late 2012. And I'm not alone at my University. It is ridiculously hard to organize anything regular here (I have tried, trust me, and still am), since no one has lots of free time available and pretty much everyone has his own schedule. And I'm not even talking about people having a full-time job — even less time is available there.

Second scenario is to participate in competitions and encounter a problem on the topic there. Which, again, with the same reasoning of limited time available, I don't find that bizarre that for some, like me, this particular contest might have been the first time they've encountered a problem on this topic, since, again, not enough time to participate in every contest available.

So, with this, the way I understand your statement is that you find bizarre that strong redcoders might not have the privilege to be coached or enough free time willing to invest in heavy training. Well, I don't. One likely needs to invest a lot of his time to become a strong redcoder, I'm not arguing with that. However one also likely needs to continuously invest even more time to continue learning new to them algorithms, data structures, approaches, types of tasks etc. And a vast majority of people after some point in life can't afford that anymore.

I did DP for <= 25 components. And random_shuffle for > 25.

There were 5 tests > 25 (28, 161, 191, 500, 500) Half of the tests were with 1 component. (these are the components after you build the graph with edge u->v if there is a path from u to v.

I suppose their tests are weak.

A case that can beat the random: a -> b <- c (b = 3, a = 1, c = 1) you need to choose b.

Have lots of these. Less likely you choose b over all of these.

I haven't found any. I've tried double, long double and eps from 1e-8 to 1e-15 and all solutions gave same answers.

I think so. Or at least their checker is wrong. I tested Gleb's accepted solution against mine on my testcase, and there were 41 differences, where each was 1e-6 differences of the two numbers in the output. So if his solution was correct I assumed mine would be also. But it failed.

I also checked yours on my test case, and it gave exactly the same result as Gleb's. But yours failed as well. So I think there is something fishy with the way they tested this problem.