NAG Library Chapter Introduction
Scope of the Chapter
This chapter provides routines to handle various aspects of IEEE floating-point arithmetic behaviour.
Background to the Problems
Modern systems allow you to control what happens to your program when an exceptional event such as overflow or division by zero occurs. Often, the default behaviour is for program execution to continue, while setting an appropriate flag. Sometimes the default behaviour is to halt execution and print a warning or error message.
The routines in Chapter X07
allow creation and detection of NaNs (Not a Number) and infinities, as well as alteration of the behaviour of a program when an exception occurs.
Recommendations on Choice and Use of Available Routines
Routines are provided to detect and create IEEE NaN (Not a Number) and infinity values, and to get and set the halting mode of various floating-point exceptions.
|Create a floating-point infinity|| ||x07baf|
|Create a floating-point NaN (Not a Number)|| ||x07bbf|
|Determine whether a floating-point number is finite|| ||x07aaf|
|Determine whether a floating-point number is NaN (Not a Number)|| ||x07abf|
|Get current behaviour of floating-point exceptions|| ||x07caf|
|Set behaviour of floating-point exceptions|| ||x07cbf|
Auxiliary Routines Associated with Library Routine Arguments
Routines Withdrawn or Scheduled for Withdrawal