I think the time limit for problem J was too low. The program of our team used complex<double>, cout and vector of STL and we got TLE. After I changed it to integer calculation, arrays instead of vectors and scanf/print, we got Accepted in 0.998 seconds.
As the time limit was very generous for problem G with 6 seconds, why not also a more reasonable time limit for problem J of, say 3 seconds. I am sure the first submission of my team would have terminated in 3 seconds.
Time Limits of World Finals Warmup Contest
Moderator: Board moderators
-
- Guru
- Posts: 724
- Joined: Wed Dec 19, 2001 2:00 am
- Location: Germany
-
- New poster
- Posts: 27
- Joined: Wed Apr 17, 2002 7:54 pm
time limits in contests
Not only that problem, but quite regularly in online contests there are poblems with a tight timelimit that does not allow a - in my opinion - reasonable factor in running time compared to a judge solution.
I understand that the resources for judging are limited and low time limits are needed so that the contest can run smoothly.
But I really think that the time limit should be set in a way that programs which have the required algorithmic complexity can run within the timelimit without making what I consider stupid optimization such as in Adrian's example above.
Not only that such optimizations lead to ugly, unmaintainable and error-prone code, I guess it's also not in the spirit of the ACM ICPC.
I understand that the resources for judging are limited and low time limits are needed so that the contest can run smoothly.
But I really think that the time limit should be set in a way that programs which have the required algorithmic complexity can run within the timelimit without making what I consider stupid optimization such as in Adrian's example above.
Not only that such optimizations lead to ugly, unmaintainable and error-prone code, I guess it's also not in the spirit of the ACM ICPC.
-
- Online Contest Administrator
- Posts: 20
- Joined: Fri Nov 30, 2001 2:00 am
Time Limit
Actually the time limit depends on judge solution.
For G there were three alternate solutions:
one took less than 0.1 second , another 3 seconds (around) and another 1 minute. So I had to put the time limit 6 seconds. This seem generous to you because your solution was good but even the problem setters judge solution took 3 seconds. So the run time was high.
For Problem J the judge solution ran around .4 second. So multiplying by 5 looked reasonable.
Not all problem setters use stl so it becomes difficult to determine what the time limit for codes using stl should be. I admit it was a mistake but those mistakes can be found only after the contest. Morever I thought problem J will be solved by many but it turned out to be a hard problem. A bad judgement error on my part.
For G there were three alternate solutions:
one took less than 0.1 second , another 3 seconds (around) and another 1 minute. So I had to put the time limit 6 seconds. This seem generous to you because your solution was good but even the problem setters judge solution took 3 seconds. So the run time was high.
For Problem J the judge solution ran around .4 second. So multiplying by 5 looked reasonable.
Not all problem setters use stl so it becomes difficult to determine what the time limit for codes using stl should be. I admit it was a mistake but those mistakes can be found only after the contest. Morever I thought problem J will be solved by many but it turned out to be a hard problem. A bad judgement error on my part.