4 4
1 2 0
2 3 0
3 4 0
4 1 0
Impossible

My solution resulted in Crash {FLT_INVALID_OPERATION} several times, before I understood, that the square of an integer number can be negative!

Oh! What a stupid mistake!-)

The crash mentioned above was caused by the following function in my source code:

function Dist( const p1, p2: Point ): extended;
begin
    Result := sqrt( sqr(p2.x-p1.x) + sqr(p2.y-p1.y) );
end;

I couldn't even imagine, that sqr() function can produce negative results!

So, If you get Crash on test #6, just use extended (^:

Before, the scanf/printf function works very fast and cin/cout has a reasonable speed. However, both of them are extremely slow nowadays, with MSVC2010 compiler. I used putchar() and getchar() to realise I/O, they used to be even faster than stdio. But they are slower than stdio now.
I have found out that the main cause is that MS's stdio.h is different from GCC's one. I don't understand header files much, but have worked out a way to make it swifter. Here it is:

First, copy the codes of "getchar" and "putchar" from your GCC's stdio.h. (If you don't have one, you may copy from below) To avoid coincide with original function, you have to rename them, eg. "getchar1" and "putchar1".
Then, for example, Prob.1196, you have to write two functions: int getInt() and void printInt(int num), whose functions are shown by their names. (Below I use ms and mp)
In the functions, you mustn't use scanf or printf. You ought to use putchar and getchar.
Below is mine crude version of them:

///////////////////////////////////////////////
inline int __cdecl getchar1 (void)
{
  return (--stdin->_cnt >= 0)
    ?  (int) (unsigned char) *stdin->_ptr++
    : _filbuf (stdin);
}
inline int __cdecl putchar1(int __c)
{
  return (--stdout->_cnt >= 0)
    ?  (int) (unsigned char) (*stdout->_ptr++ = (char)__c)
    :  _flsbuf (__c, stdout);}
int ms(){
int s=0;
char c;
while(1){
c=getchar1();
if(c>='0'&&c<='9')break;
}
s=c-'0';
while(1){
c=getchar1();
if(c<'0'||c>'9')break;
s=s*10+c-'0';
}
return s;
}
void mp(int s){
     char b[12];
     int l=0,i,m=0;
     if(s==0){
       putchar1(48);
       return;
     }
     if(s<0){
       m=1;
       s=-s;
     }
     while(s){
       b[l]=s%10;
       l++;
       s/=10;
     }
     if(m)putchar1('-');
     for(i=l-1;i>=0;i--)putchar1(b[i]+'0');
}
///////////////////////////////////////////////

How to use? For instance, A+B Prob:
////////////////////////
int main(){
int a,b;
a=ms();
b=ms();
mp(a+b);
}
/////////////////////////

Not very beautiful coding style though......

How sufficient they are? Let's see the following statistc:
Problem 1196 with almost same code:

Old Compiler with ms() and mp(): 2780220 19:02:10 5 Oct 2009 tiancaihb 1196 C++ Accepted
 0.062 1 625 KB

New Compiler with ms() and mp(): 3128057 17:26:43 11 Aug 2010 tiancaihb 1196 C++ Accepted
 0.093 1 604 KB

New Compiler with scanf() and printf(): 3128068 17:32:18 11 Aug 2010 tiancaihb 1196 C++ Accepted
 0.296 1 616 KB

New Compiler with cin and cout: 3128071 17:33:31 11 Aug 2010 tiancaihb 1196 C++ Accepted
 1.218 1 700 KB



Sorry for my bad English. Hope this will help some of you!
:)

Edited by author 11.08.2010 17:55

Edited by author 11.08.2010 17:55

1) For those who get WA 1 mb helps :
with
while (!cin.eof()) char c = cin.get();
i had WA 1
with
char t[10000];
cin.getline(t, 1000);
i had WA 9 !

2)
the problem says that the length of input is no more than 10000 characters

but this code
...
char t[10000];
cin.getline(t, 10000)
...
get WA9
this :
char t[20000];
cin.getline(t, 20000);
get AC

Strange, isn't it?

Just check all diagonals and try to find one of these four sequences:
1. white-black-empty
2. black-white-empty
3. empty-white-black
4. empty-black-white

And you'll get AC. Nothing hard...

3
q1 1 1 2
q2 1 2 1
q3 2 1 1

2
a1 1 1 1
a2 1 1 1

3
w1 6 6 6
w2 6 4 4
w3 5 5 5

great thanks!

Edited by author 11.08.2010 04:41

If you have WA after rejudge (or if you simply have WA and don't know why) try to test your solution with this expression:

NOT NOT A

GL ;-)

My solve O(N), but output too slow. How fast print answer?

whats wrong could you send some tests pls
if you want source I'll give it

First I used my own method for reading data, this function: http://paste.ubuntu.com/475799/ GOT TLE 5...

Then used this method:
double[] data = Console.ReadLine().Split().Select(s => double.Parse(s)).ToArray();
GOT AC...

Whats the 5 test? is it bug in my reading function or what?
Need help.

PS
Check last @Crash@: http://acm.timus.ru/status.aspx?space=1&num=1582&author=73720

Code:
if (data[0].IndexOf('.') == -1 || data[1].IndexOf('.')==-1 || data[2].IndexOf('.')==-1) throw new System.DivideByZeroException();

Some of numbers are NOT with floating point.

From problem statement:
В единственной строке через пробел записаны 3 числа с плавающей точкой k1, k2 и k3.

Edited by author 10.08.2010 14:02

Any trick on Test24?

AC this problem is very simple!
Try to solve the problem with O(n).

program timus;
var
  a,b,c:string;
procedure solving;
var
  i,j:longint;
  check:boolean;
  d:char;
begin
  check:=false;
  c:='';
  c:=c+b[1]+b[2]+b[3]+b[4];
  for i:=1 to 4 do
    begin
      d:=c[2]; c[2]:=c[3]; c[3]:=c[4]; c[4]:=d;
      if c=a
        then check:=true;
    end;
  c:='';
  c:=c+b[2]+b[3]+b[1]+b[4];
  for i:=1 to 4 do
    begin
      d:=c[2]; c[2]:=c[3]; c[3]:=c[4]; c[4]:=d;
      if c=a
        then check:=true;
    end;
  c:='';
  c:=c+b[3]+b[1]+b[2]+b[4];
  for i:=1 to 4 do
    begin
      d:=c[2]; c[2]:=c[3]; c[3]:=c[4]; c[4]:=d;
      if c=a
        then check:=true;
    end;
  c:='';
  c:=c+b[4]+b[2]+b[3]+b[1];
  for i:=1 to 4 do
    begin
      d:=c[2]; c[2]:=c[3]; c[3]:=c[4]; c[4]:=d;
      if c=a
        then check:=true;
    end;
  if check=true
    then writeln('equal')
    else writeln('different');
end;
begin
  readln(a);
  readln(b);
  SOLVING;
end.

program ural;
var n,a,b,z,i:byte;
    r:integer;
    h:array[1..222]of integer;
    l:array[1..49062]of integer;
    t:array[1..49062]of byte;
    p,u:array[1..222]of boolean;
    v:array[1..222]of byte;
procedure search(f,d:byte);
var r:integer;
  begin
    r:=h[d];
    p[d]:=true;
    while r<>0 do
      begin
        if(t[r]<>f)and(p[t[r]]=false)then
          begin
            if v[d]=0 then
              begin
                inc(z);
                v[d]:=t[r];
                v[t[r]]:=d;
              end;
            search(d,t[r]);
          end;
        r:=l[r];
      end;
  end;
function match(d:byte):boolean;
var r:integer;
  begin
    r:=h[d];
    p[d]:=false;
    while r<>0 do
      begin
        if(v[t[r]]>0)and(p[v[t[r]]])and(u[v[t[r]]]=false)and(match(v[t[r]]))then
          begin
            v[t[r]]:=d;
            v[d]:=t[r];
            exit(true);
          end;
        r:=l[r];
      end;
    u[d]:=true;
    exit(false);
  end;
begin
  readln(n);
  while not eof do
    begin
      readln(a,b);
      inc(r);
      l[r]:=h[a];
      h[a]:=r;
      t[r]:=b;
      inc(r);
      l[r]:=h[b];
      h[b]:=r;
      t[r]:=a;
    end;
  for i:=1 to n do
    if p[i]=false then
      search(0,i);
  for i:=1 to n do
    if v[i]=0 then
      begin
        fillchar(p,sizeof(p),true);
        if match(i) then
          inc(z);
      end;
  writeln(z*2);
  for i:=1 to n do
    if v[i]>i then
      writeln(i,' ',v[i]);
end.

input:
-3 3
2 2 1 2 2 1
output:
2

Edited by author 20.03.2011 23:02

don't have idea how to solve (( pls anybody write to debowir4eg@gmail.com how to solve this hard problem. Or it will break my brains.