<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<p>Hello,</p>
<p>Another case with []: <br>
The <a href="http://bugzilla.scilab.org/show_bug.cgi?id=15579">bug
15579</a> reports that <b>cond([])</b> returns <b>1</b>, while
<b>cond([],2)</b> -- that computes the same 2-norm conditional
number for any other matrix than [] -- returns <b>0</b>.<br>
<br>
Actually, <b>cond([], p)</b> returns 0 for any accepted p value.<br>
<br>
At least since Scilab 4, and likely since the beginning of
Scilab, <b>cond([])</b> is set to 1 as a specific case.<br>
I am wondering why.<br>
<br>
In my opinion, the mismatch between <b>cond([])</b> and the
equivalent <b>cond([],2)</b> must be fixed.<br>
There are at least two way to do it : <br>
</p>
<ul>
<li>either keep <b>cond([])</b> to <b>1</b> and set all <b>cond([],
p)</b> to 1 instead of 0</li>
<li>or set <b>cond([])</b> to <b>0</b>.</li>
</ul>
<p>I don't see any clear reason enforcing a choice rather than the
other.<br>
Do you?<br>
<br>
Let's note that</p>
<ul>
<li>Octave returns 0 in both cond([]) and cond([],p) cases</li>
<li>Julia, Python, and R (with kappa()) yield an error</li>
</ul>
Regards<br>
Samuel<br>
<br>
<p><br>
</p>
</body>
</html>