[Poll] CCC 2009 Results

[bbi5291]

These aren't my official school results - these are just the marks we sent in. If there are any issues with format or timing (we have a quad-core Linux system with 2GB of RAM, so everything is really fast) then the results may change. (I personally believe that a C++ solution that takes more than 15 seconds needs improvement, considering that it's possible to pass within 60 seconds in slow languages like Turing).

Since the math department has our school login info, and they aren't that well-informed about the CCC, we haven't gotten our official results. We might be able to get them tomorrow.

I am not nearly so optimistic as you, A. J.

I estimate that the cutoff will be around 60. This is because most people would have had trouble with problem 2, there are some issues with efficiency on problem 4 (especially if you used std::priority_queue), and problem 5 was a killer; in the ideal case everybody would code a partial solution to problem 5 but most of the stage 2 invitees from last year had minimal experience with actual contests; I suspect they garnered their skills from the USACO training webpage which doesn't reward partial solutions at all.

[Foundation]

A.J @ Mon Mar 23, 2009 6:05 pm wrote:

In my opinion (although I don't think it well help you at all Peter, as I am not too good), I believe the cutoff this year will be about 53 +/- 2.

No, A.J. Every bit helps, I was merely asking Brian only because he is very experienced and knows more about this kind of stuff than the rest of us.

bbi5291 wrote:

These aren't my official school results - these are just the marks we sent in. If there are any issues with format or timing (we have a quad-core Linux system with 2GB of RAM, so everything is really fast) then the results may change. (I personally believe that a C++ solution that takes more than 15 seconds needs improvement, considering that it's possible to pass within 60 seconds in slow languages like Turing).

Since the math department has our school login info, and they aren't that well-informed about the CCC, we haven't gotten our official results. We might be able to get them tomorrow.

I am not nearly so optimistic as you, A. J.

I estimate that the cutoff will be around 60. This is because most people would have had trouble with problem 2, there are some issues with efficiency on problem 4 (especially if you used stdpriority_queue), and problem 5 was a killer; in the ideal case everybody would code a partial solution to problem 5 but most of the stage 2 invitees from last year had minimal experience with actual contests; I suspect they garnered their skills from the USACO training webpage which doesn't reward partial solutions at all.

Ah. I understand. Our official results didn't change, so I doubt yours will. Around 60 would be a bit high in my opinion as #2 definitely required thinking if you did not embark on the right perspective to begin with, and #4 you must know Dijkstra, though I am optimistic, a lot of people won't know as most of them are very smart yet thinking of the algorithm in a contest environment is kinda difficult. #5...experience helps a lot, but thinking nevertheless is a must here.

[bbi5291]

You're too kind, Peter - now that I think of it I'm not as certain anymore. 60 might indeed be a bit high, but I find it doubtful that it would be less than 55. Someone from Woburn last year just barely managed to get into stage 2 [interestingly enough, he was initially rejected from UW, and then they reversed their decision after he got invited], and I am convinced that he would have earned at least 50 out of 60 on the first four problems, and at least coded up a partial solution to #5, getting 6 marks out of 15 (the test case weights were 2 2 2 3 3 3 and for the 2nd testcase RCK = 100 million, which will definitely run within 60 seconds in C++)...

[Analysis Mode]

[/quote] Ah. I understand. Our official results didn't change, so I doubt yours will. Around 60 would be a bit high in my opinion as #2 definitely required thinking if you did not embark on the right perspective to begin with, and #4 you must know Dijkstra, though I am optimistic, a lot of people won't know as most of them are very smart yet thinking of the algorithm in a contest environment is kinda difficult. #5...experience helps a lot, but thinking nevertheless is a must here. [/quote]

Or perhaps, you coded up a working program, except to realize after the contest that instead of running Dijkstra's once, you ran it up to N times (N=5000). yeah, instead of running dijkstra's from the destination vertex, I ran it from every pencil-selling city.

[Cyril]

I saw the "unofficial" results at school today. For some reason, our school's top three senior scores (64, 60, ~48) weren't posted. I suspect that this is because the submissions that made the honour roll have to be re-tested due to test case discrepancies, but I just want to confirm this. Is anyone else here in a similar situation?

I think the cutoff will be around 55-58. In 2007, which had problems of roughly comparable difficulty, the cutoff was 57. Add a few points for suboptimal solutions for #5, subtract a few points for memory in #4... seems reasonable.

[Foundation]

Cyril @ Mon Mar 23, 2009 11:56 pm wrote:

I saw the "unofficial" results at school today. For some reason, our school's top three senior scores (64, 60, ~48) weren't posted. I suspect that this is because the submissions that made the honour roll have to be re-tested due to test case discrepancies, but I just want to confirm this. Is anyone else here in a similar situation?

I think the cutoff will be around 55-58. In 2007, which had problems of roughly comparable difficulty, the cutoff was 57. Add a few points for suboptimal solutions for #5, subtract a few points for memory in #4... seems reasonable.

Hm. All solutions above 40 will be tested again to my knowledge. I think 2007 was more standard CCC than this year. But quite possible.

Analysis Mode wrote:

Or perhaps, you coded up a working program, except to realize after the contest that instead of running Dijkstra's once, you ran it up to N times (N=5000). yeah, instead of running dijkstra's from the destination vertex, I ran it from every pencil-selling city.

Yeah. That happens sometimes. It's okay, you'll know next time, right?

bbi5291 wrote:

You're too kind, Peter - now that I think of it I'm not as certain anymore. 60 might indeed be a bit high, but I find it doubtful that it would be less than 55. Someone from Woburn last year just barely managed to get into stage 2 [interestingly enough, he was initially rejected from UW, and then they reversed their decision after he got invited], and I am convinced that he would have earned at least 50 out of 60 on the first four problems, and at least coded up a partial solution to #5, getting 6 marks out of 15 (the test case weights were 2 2 2 3 3 3 and for the 2nd testcase RCK = 100 million, which will definitely run within 60 seconds in C++)...

That is very true. Getting 15, 15, 15, 6, 6 is not going to be uncommon if one used floyd warshall (most people know this) so 57 might be common. It'll all depend on exactly how hard #2 was and whether a lot of people had the guts to move on to 5 when their solution for 4 was not optimal.

[Analysis Mode]

yeah, i swear i won't make those mistakes next year. Bug free coding, something I have to master by next February.