Search code examples
floating-pointdoublefloating-accuracyieee-754

Is it possible to get 0 by subtracting two unequal floating point numbers?


Is it possible to get division by 0 (or infinity) in the following example?

public double calculation(double a, double b)
{
     if (a == b)
     {
         return 0;
     }
     else
     {
         return 2 / (a - b);
     }
}

In normal cases it will not, of course. But what if a and b are very close, can (a-b) result in being 0 due to precision of the calculation?

Note that this question is for Java, but I think it will apply to most programming languages.


Solution

  • In Java, a - b is never equal to 0 if a != b. This is because Java mandates IEEE 754 floating point operations which support denormalized numbers. From the spec:

    In particular, the Java programming language requires support of IEEE 754 denormalized floating-point numbers and gradual underflow, which make it easier to prove desirable properties of particular numerical algorithms. Floating-point operations do not "flush to zero" if the calculated result is a denormalized number.

    If an FPU works with denormalized numbers, subtracting unequal numbers can never produce zero (unlike multiplication), also see this question.

    For other languages, it depends. In C or C++, for example, IEEE 754 support is optional.

    That said, it is possible for the expression 2 / (a - b) to overflow, for example with a = 5e-308 and b = 4e-308.