Precision Error

[taskin]

how can i solve prescision error when comparing to long double value?

i use fabs(a - b) <= EPS where EPS is 1e-15, but its not working.
use fabs(a - b) < EPS also but its not working again.

[Cho]

This depends on the problem.
But I think 1e-15 is too small. I ususally use 1e-9 or even 1e-6.