Re: Machine epsilon, Octave and Matlab

[John W. Eaton]

On 30-Aug-2001, Geraint Bevan <address@hidden> wrote:

| The original poster was actually Craig Stoudt enquiring about the
| difference between Matlab and Octave under Win2000. However, you are quite
| right about the result being the same on the SGI as on the Alpha:

Sorry, I lost track.

Now, are there any experts out that who know whether this difference
is something that is a bug (in the floating point implementation, not
Octave) or allowed by the standard?  It seems odd that 1/(1+one bit)
results in exactly 1, but maybe I should not be surprised.  Is this
due to the rounding mode and possibly having different defaults for
that on different systems?

jwe



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------