fine tuning the == operator to match double percision

In my code there is a parameter I am calculating. During many of the test runs, this parameter should be 0. Since the parameter is calculating through multiple additions and subtractions, it is not exactly 0 but it is less than 10^-10. Currently I am using:

double tol = pow(10,-10);
if (fabs(delta_U) < tol)){//whatever
}

Is there a more elegant way to do 开发者_StackOverflowso?

It looks fine. You could just write the tolerance directly, instead of computing it with pow.

double tol = 1e-10;

jpalecek is correct - saves computation of a constant.

I do not know the nature of the calculations but you could either possibly

• Reduce the errors by modifying the order of calculations and operators
• Possibly make the calculation use integers

Wrap it in a function so it is more clear what you are trying to do:

if (near_zero(delta_U))
{
   // ...
}

where:

inline bool near_zero(double d)
{
   return fabs(d) < 1e-10;
}

You can specify floating point in scientific notation:

if (fabs(delta_U) < 1e-10)
{
   //whatever
}