From bug-request at octave dot org Mon Feb  7 09:45:10 2005
Subject: Re: Inconsistency in complex division by zero
From: "John W. Eaton" <jwe at bevo dot che dot wisc dot edu>
To: bug at octave dot org
Date: Mon, 7 Feb 2005 10:49:41 -0500

On  7-Feb-2005, Gabriele Pannocchia <g dot pannocchia at cpclab dot ing dot unipi dot it> wrote:

| Hi
| 
| Matlab 7.0sp1 returns to me:
| >> 1i/0
| Warning: Divide by zero.
| 
| ans =
| 
|       NaN +    Infi
| 
| i.e. with a "+" for the imaginary part.
| 
| Since 1/0 returns (either in octave and matlab)
| >> 1/0
| Warning: Divide by zero.
| 
| ans =
| 
|    Inf
| 
| I think the correct answer should be :
| ans = NaN + Infi

And I would say both are wrong, because i is pure imaginary, so we
should really be getting just Infi, no?  When we have 1/0, we get Inf,
not Inf+NaNi (i.e., there is no imaginary part of 0 that is magically
tacked on to the 1 to screw things up).

Another example that demonstrates this problem is Inf*i, which you
would expect to produce Infi, not Nan+Infi.

To solve this correctly, we need a pure imaginary data type or perhaps
special cases in many of the operators in the complex class.

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
-------------------------------------------------------------