ENG  RUSTimus Online Judge
Online Judge
Задачи
Авторы
Соревнования
О системе
Часто задаваемые вопросы
Новости сайта
Форум
Ссылки
Архив задач
Отправить на проверку
Состояние проверки
Руководство
Регистрация
Исправить данные
Рейтинг авторов
Текущее соревнование
Расписание
Прошедшие соревнования
Правила
вернуться в форум

Общий форум

Memory limits have been increased to 16 MB for most of problems (+)
Послано Problemset Maintainer 15 май 2006 01:34
These problems have ML lower than 16 MB:
1144 - 4 MB
1148 - 4 MB
1171 - 4 MB
1175 - 2 MB
1220 - 0.75 MB
1269 - 8 MB
1275 - 0.5 MB
1276 - 4 MB
1306 - 1 MB
1307 - 2 MB
1395 - 4 MB
Re: Memory limits have been increased to 16 MB for most of problems (+)
Послано Kant SU -Dmitry - DIVAN 15 май 2006 02:05
And what about problem 1048, for example?
Hm...
Послано Burunduk1 15 май 2006 02:21
While ML was 1 MB some problems were much more intresting!!!
When you discussed it in previous topic I thought that
finally list of problems with small ML will be larger...

1017 - Dull DP in O(N^3) needs O(N^2) of memory => a lot of MLEs.
1048 - The trouble is only to avoid MLE!
1145 - Also troubles with ML = 1 MB.
1148 - With 4 MB it's too easy to solve it without hashing.
1169 - Dull DP in O(N^4) needs O(N^3) of memory => a lot of MLEs.
1171 - Probably you are right, but IMHO - 1 MB.
1172 - O(N^3*Len) was MLE.
1219 - Solutions without random were very complicated.
1240 - Again dull DP couldn't pass cause MLE.
1249 - O(N^2) of memory was MLE.
1250 - Also troubles with ML = 1 MB.
1276 - IMHO - 1 MB.
...

If you want I can give you full list.

PS: Is it really necessary to change MLEs?
Re: Hm...
Послано Problemset Maintainer 15 май 2006 13:26
New MLs are really necessary.
If some problems have became much more easier now, I can change ML back. Please inform me about such problems.

1048 - this problem could be solved with long arithmetics, and now it can be solved with long arithmetics. This problem needs in Version 2 with appropriate limitations.

1017, 1169, 1145 - it is a pity, but 1 MB is too low, so let it be 16 MB, regardless of the fact that problems have became easier.

1171, 1172, 1219, 1240, 1250, 1276 - the problems didn't become easeir. Contrary, some boring memory optimizations became needless.

1249 - I agree. ML was changed to 4MB

Another suggestions?
Re: Hm...
Послано Burunduk1 15 май 2006 18:18
Contrary, some boring memory optimizations became needless.
Memory optimization is part of solution which is as
important as time optimization.
And more - program which uses less memory runs faster.

And in this case:
about 1171, 1219, 1250 (and even 1276) I agree with you
but:
in 1172 (IMHO) solution in O(N^3*Len) of memory is much
easier than others and, of course, it shouldn't pass.
in 1240 the main idea of right solution is not to use
one of parametrs of DP. And if ML is big dull DP passes
all tests. It's wrong!

And what about 1148?
Solution which fits in 4MB I've written in 5 minutes.

In other problems...
You are just right. The main idea is to solve it :)

Edited by author 15.05.2006 18:21
Re: Hm...
Послано Problemset Maintainer 15 май 2006 19:28
Burunduk1 писал(a) 15 мая 2006 18:18
in 1172 (IMHO) solution in O(N^3*Len) of memory is much
easier than others and, of course, it shouldn't pass.

To make O(N^2*Len)-memory solution from O(N^3*Len) you need not more than 5 minutes.
Burunduk1 писал(a) 15 мая 2006 18:18
in 1240 the main idea of right solution is not to use
one of parametrs of DP. And if ML is big dull DP passes
all tests. It's wrong!

My own solution use excess parameters but it fits in memory, so if we want to prevent such solutions we would need an extremely low ML (about 300-400 KB like in DOS environment). But our memory calculating system has a normal error up to 150 KB! So it's impossible.

Burunduk1 писал(a) 15 мая 2006 18:18
And what about 1148?
Solution which fits in 4MB I've written in 5 minutes.

And I can write a solution which use 500KB of memory in 10 minutes. So what?
Some MLs have been decreased
Послано Problemset Maintainer 17 май 2006 00:38
MLs for these problems have been decreased:
1119 - 4 MB
1249 - 4 MB
1414 - 8 MB

Edited by author 17.05.2006 00:40
Re: Hm...
Послано Гладких Максим 20 май 2006 23:52
1100!
Re: Hm...
Послано Problemset Maintainer 21 май 2006 11:15
You want to say that the trick is to write code like this:
http://acm.timus.ru/forum/thread.aspx?id=11191

Edited by author 21.05.2006 11:55
Re: Hm...
Послано Гладких Максим 21 май 2006 18:12
Isn't it?