Xref: utzoo comp.unix.ultrix:7438 comp.lang.fortran:5556
Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!uunet!interet!jim
From: jim@interet.UUCP (User)
Newsgroups: comp.unix.ultrix,comp.lang.fortran
Subject: Re: How to detect NaN's
Summary: simple test for non-ordinary numbers
Message-ID: <15@interet.UUCP>
Date: 31 May 91 13:16:46 GMT
References: <1991May30.204332.16506@litwin.com>
Organization: Interet, Maplewood, NJ
Lines: 23

In article <1991May30.204332.16506@litwin.com>, vlr@litwin.com (Vic Rice) writes:
    > How can I test a variable for the presence of a NaN or Infinity ???

We test for NAN and INF by using the following:
      if(x*0.0 .ne. 0.0) write(*,*) 'ERROR: X is not a number'
The idea is that if x is NAN, then x times zero is also NAN and not  zero.
Unfortunately,  the  above code may be optimized out by an optimizing com-
piler which "knows" that zero times x is zero, so you may have to code  it
as:

      if(notone.eq.1) then
        temp=1.0
      else
c  this branch is always taken
        temp=0.0
      endif
      if(x*temp.ne.0.0) write(*,*) 'ERROR: X is not a number'

Note that x may be NAN or maybe infinity or some other special bit pattern
depending on your floating point implementation.  But if the test succeeds,
then X is definitely not an ordinary number.

Jim    uunet!interet!jim