thanks!
I got great help!

Fixed. Thank you.

Which VC? 'cuz my VC2005 eats both %lld and %I64d, both __int64 and long long.

It's a bug of FreePascal
This program outputs 4294967295 on server.

var
    x: integer;
begin
    x := 1;
    writeln(int64(-x));
end.

Try to avoid structures like int64(-x).
I hope the bug will be fixed soon.

I solved this problem in the same way and get AC
link: http://acm.timus.ru/status.aspx?space=1&pos=811782

I still can't pass #55
Now I suspect I made something wrong when solving
pos_A + period_A * x == pos_B + period_B * y using euclid's method
My method goes like this
(short hands: pos_A: n1, pos_B: n2, period_A: a, period_B: b)
So we want to solve n1 + ax == n2 + by, such that x, y >= 0 && n1 + ax is minimized
if n1 >= n2, we just solve
   by - ax == n1 - n2, such that x, y >= 0 && x is minimum
else
   ax - by == n2 - n1, such that x, y >= 0 && y is minimum
Here's my code for solving
ax - by == n, such that x, y >= 0 && y is minimum
( which equals to solve:
  by == a - n mod a (mod a) )

!! I am sorry for the readablility of the code since I don't know how to make the code look like indented. You can use my account 36057RJ to edit this msg and get indented code
################################
typedef typedef __int64 LL;
LL
Gcd(LL a, LL b, LL *x, LL *y)
{
    if ( b == 0 ) {
        *x = 1;
        *y = 0;
        return a;
    } else {
        LL xx, yy;
        LL gcd = Gcd(b, a % b, &xx, &yy);
        *x = yy;
        *y = xx - (a / b) * yy;
        return gcd;
    }
}
LL
SolveLinear(LL a, LL b, LL n)
{
    LL result;
    LL x, y;
    LL q;
    LL d = Gcd(b, a, &x, &y);
    n = (a - n % a) % a;
    if ( n % d != 0 )
        return -1;
    x *= n / d;
    q = a / d;
    result = x % q;
    if ( result < 0 )
        result += q;
    return result;
}
####################################
Is this the right thing to do ?
Thanks in advance for taking your time to
check this

Edited by author 26.04.2005 19:31