<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<tt>Dear members of the (s)Ipopt mailing list,<br>
<br>
with great interest I have read the literature on sIpopt and I
have already set up Ipopt and sIpopt and now I'm playing around
with some of my robotics optimal control problems where a few
questions about the characteristics of the approximation of the
solution for the perturbed case arose:<br>
- Are the results obtained with sIpopt (with bound checking
enabled) admissible, i.e. do they satisfy not only the variable
bounds but also general (in-)equality constraints? I found them
violating constraints. Is this normal or am I doing something
wrong?<br>
- I noticed that sIpopt does not seem to converge for larger
perturbations (with bound checking enabled) which is expected. Is
there a way to estimate a parameter interval in which sIpopt can
produce a valid result? There are approaches such as Büskens'
"sensitivity domain" method that work for a constant active set,
but is there a method that works reliably for a changing active
set?<br>
<br>
Thanks,<br>
Alexander Reiter<br>
<br>
</tt>
<div class="moz-signature">-- <br>
<span style="font-size: 10pt;"><font face="arial" color="black"><b>DI
Alexander Reiter, BSc</b><br>
University Assistant<br>
Institute of Robotics<br>
<br>
<b>JOHANNES KEPLER<br>
UNIVERSITY LINZ</b><br>
Altenberger Straße 69<br>
Science Park 1, Room 0537<br>
4040 Linz, Austria<br>
T +43 732 2468 6497<br>
<a class="moz-txt-link-abbreviated" href="mailto:alexander.reiter@jku.at">alexander.reiter@jku.at</a><br>
<a href="http://www.jku.at">www.jku.at</a><br>
<a href="http://www.robotik.jku.at">robotik.jku.at</a></font></span></div>
</body>
</html>