Here's the main part of my code:

for i:=0 to n-1 do
  begin
   cpt[i]:=1441;
   for j:=0 to n-1 do t[i,j]:=1441;
  end;
 for i:=0 to m-1 do
  begin
   read(x,y);
   dec(x); dec(y);
   readln(t[x,y],w[x,y]);
   dec(w[x,y],3000000);
   w[x,y]:=w[x,y] div 100;
   w[y,x]:=w[x,y];
   t[y,x]:=t[x,y];
  end;
 cup[0]:=10000000; cpt[0]:=0;
 for i:=0 to n-1 do
  for j:=0 to n-1 do
   begin
    if i=j then continue;
    m:=min(cup[j],w[i,j]);
    r:=cpt[j]+t[i,j];
    if (cup[i]<m)and(r<1441) then
     begin
      cup[i]:=m;
      cpt[i]:=r;
     end;
   end;

Please, who can give me tricky tests where my code fails?
I can't find my mistake. More over, my algo is true as well.

I have wa #25 too:(

Use random tests!
Very helpfull.
After 450 I couln't without this method.

Thanks, but how I get answers for this tests?

On some random tests answer of wrong program
is obviously wrong.
For example you may use broote forse method
as alternative on small dimensions.

Edited by author 20.12.2007 22:13

I have AC:-) it was is stupid bug in dijkstra...
But that is 'Brut Force Method'? Can you send me letter about this method to mail on russian, because i don't know english good... x2m@mail.ru

Broote forse method is full search with gatanteed right
answer.

Can you tell me what did you change in your program? i can't still find my mistake. can you send me your code to my e-mail?(yashar25.92@mail.ru)
Thanks