We have to perform some service works with Timus Online Judge. So for a couple of days solutions could be checked with delays up to several hours. Just keep patience.

If you're doing a greedy solution and you're sorting by the rightmost endpoints (B_i), you need to sort with the leftmost endpoints as second key. That is, if B_i == B_j, the shortest segment of i and j goes first in the list.

This basically means that longer segments are kept, so for inputs:

3
3 5
1 5
5 6

You would get result
2
1 5
5 6

instead of
2
3 5
5 6

which means you get a bigger covering of the set but the problem statement didn't mention this requirement.

Edited by author 24.12.2014 21:07

program Bourne;
  var
    n,i,j,x:longint;
    a:array[0..150000] of smallint;
    b:array[0..150000] of longint;
  begin

    readln(n);
    for i:=1 to n do
      begin
        readln(b[i],a[i]);
      end;

    for x:=100 downto 0 do
      begin
        for i:=1 to n do
          if a[i]=x then writeln(b[i],' ',a[i]);
      end;

  end.

nt

Задачи долго проверяются с чем это связано? И как это исправить?

Сделал небольшой замер выполнения задач на C++, Java, Scala. Взял задачу задачу 1000, так-как на нее есть решения на всех языках.
Результаты выполнения: C++ - 0.001, Java - 0.124, Scala - 0.577, Scala с использованием JDK классов - 0,421
Результаты неутешительны, так-как задачи с ограничением в 0.5 с не выполнимы. Воспользоваться стандартными классами JDK но выигрыш был ненамного. Да и не вижу смысла их использовать. Предполагаю что медленно стартует исполнение программ и это особенность JVM. Задачи хочу решать на Scala так-как есть такая возможность.
Есть ли какие то решения данной проблемы?

Don't use cin/cout or Scanner in java.

i ended up counting all possible Strings can be composed quickly.

so preprocessing gives me

552 of the 16 symbols words
27956 of less than 16 symbols words (including those who are 1)

WA1 why?can i get the test data?

Edited by author 31.08.2015 18:13

can anyone please give the #4 test case?

I understand the problem. There is one frying pan!
Edited by author 29.08.2015 13:17

Edited by author 29.08.2015 14:14

Edited by author 29.08.2015 14:15

Can someone please tell me the problem with this code. It looks okay to me.
Environment: Linux, G++ C++11, GCC version 4.9.2

#include <cmath>
#include <iostream>
#include <iomanip>
#include <vector>
#include <sstream>
#include <ios>
#include <string>

using namespace std;

int main(void)
{
        vector<long long> v;

        string strBuf;
        long long result;
        while (getline(cin, strBuf)) {
                result = atoll(strBuf.c_str());
                v.push_back(result);
        }

        vector<long long>::reverse_iterator ritr = v.rbegin();
        while (ritr != v.rend()) {
                cout << fixed  << setprecision(4) << sqrtl(*ritr) << endl;
                ++ritr;
        }

        return 0;
}

Compilation:
$ g++ ktm_0_1.cpp -std=c++11

Sample execution seems fine to me:
$ ./a.out
 1427  0

   876652098643267843
5276538 <-- After this, I enter CTRL-D manually
2297.0716
936297014.1164
0.0000
37.7757

try this one
7
1 1
2 2
4 3
5 3
7 2
9 0
5 7

answer is ccw

have no idea what this test may be

this test can help you:
9
4 1
2 4
3 4
6 4
5 4
7 5
8 5
9 7
BLOOD
5

answer is: 2 3 6

Good luck!

My algo works O(N^3) at 0.015 sec.
Solution always exist.

Idea:
We must paint first half of edge in 'Y' and second half in 'G'.

1)Delete cycles (if the edge belongs to the cycle then we delete it from the graph) and paint edges;
2)Paint a tree.


Edited by author 27.08.2015 15:36

I use bfs...but record the route in a map, and maybe the copy of the map cost too much time...
How can I get all the routes in another good way?

WA#8 confuse me.

47 times i had submitted a code with a small bug in heap... I'm sad...