Floating-point values are difficult to compare for equality because of numerical errors. This class provides methods to compare floating-point values with regard to numerical errors.

To compare floating-point values a tolerance value epsilon will be used. Values smaller than the epsilon tolerance are treated as equal to 0. Floating-point values are considered equal when the difference is less than the epsilon tolerance.

Some methods allow to use a custom epsilon. If the method has no epsilon parameter, then the default values EpsilonF (for float) or EpsilonD (for double) are used.

Note that when two large floating-point numbers need to be compared a bigger epsilon tolerance needs to be used than when comparing a floating-point number with zero. When comparing two floating-point numbers using the default epsilon tolerance EpsilonF or EpsilonD (see AreEqual(Single, Single) or AreEqual(Double, Double)) then the epsilon values are automatically scaled to match the range and resolution of the floating-point numbers. When two floating-point numbers are compared using an explicit epsilon value (see AreEqual(Single, Single, Single) or AreEqual(Double, Double, Double)) then this epsilon value is used directly.

Reference