Page #501 (Appendix D. Java Modifiers)

[Page 1278]

Appendix E. Special Floating-Point Values

Dividing an integer by zero is invalid and throws ArithmeticException , but dividing a floating-point value by zero does not cause an exception. Floating-point arithmetic can overflow to infinity if the result of the operation is too large for a double or a float , or underflow to zero if the result is too small for a double or a float . Java provides the special floating-point values POSITIVE_INFINITY , NEGATIVE_INFINITY , and NaN (Not a Number) to denote these results. These values are defined as special constants in the Float class and the Double class.

If a positive floating-point number is divided by zero, the result is POSITIVE_INFINITY . If a negative floating-point number is divided by zero, the result is NEGATIVE_INFINITY . If a floating-point zero is divided by zero, the result is NaN , which means that the result is undefined mathematically. The string representation of these three values are Infinity, -Infinity, and NaN. For example,

  System.out.print(1.0 / 0); // Print Infinity   System.out.print(1.0 / 0); // Print Infinity   System.out.print(0.0 / 0); // Print NaN  

These special values can also be used as operands in computations . For example, a number divided by POSITIVE_INFINITY yields a positive zero. Table E.1 summarizes various combinations of the / , * , % , + , and “ operators.

Table E.1. Special Floating-Point Values

x	y	x/y	x*y	x%y	x + y	x “ y
Finite	± 0.0	±	± 0.0	NaN	Finite	Finite
Finite	±	± 0.0	± 0.0	x	±
± 0.0	± 0.0	NaN	± 0.0	NaN	± 0.0	± 0.0
±	Finite	±	± 0.0	NaN	±	±
±	±	NaN	± 0.0	NaN	±
± 0.0	±	± 0.0	NaN	± 0.0	±	± 0.0
NaN	Any	NaN	NaN	NaN	NaN	NaN
Any	NaN	NaN	NaN	NaN	NaN	NaN

Note

If one of the operands is NaN, the result is NaN.