subj, actually:(( i can't imagine the test where it doesn't work, tried to decrease eps - nothing better than TLE test #9...


const eps=10e-8;
var  A,N,P:real;
begin
    readln(N);
    A:=1;
    while abs((sqrt(sqr(2*A-1)+8*N))-(trunc(sqrt(sqr(2*A-1)+8*N))))>eps do
      A:=A+1;
    P:=(0.5-A+sqrt(sqr((2*A-1))+8*N)/2);
    writeln(A:0:0,' ',P:0:0);
end

Edited by author 29.04.2004 18:50

Edited by author 29.04.2004 19:06

I did. Setting eps=10e-5 leads to WA test8, eps=10e-6 and 10e-7 - to WA test 9:(( and then 10e-8 come to TLE:((

you search for P from the divisors of 2*n less or equal to
sqrt(2*n)
then you verify if A is an integer A=(2*n/P-P+1)/2
you output the first solution because P is maxim(if you begin with sqrt(2*n) and go down)
if you found no A integer you output N 1

              Good Luck!

thanks! that was the problem! the algorithm was almost the same, but I didn't manage the situation when nothing can be found:((

Edited by author 04.05.2004 15:04