<br>Hi Konrad,<br><br>Thank you for pointing me towards the previous discussion. The last time I searched my archives I don't think I understood where the problem was coming from, and so I overlooked them because I searched for the wrong key terms. I apologize to the list if I sounded frustrated earlier - it is a testament to MMTK's utility that I expected the task ahead of me to be quite easy and fun, and quickly became frustrated when I encountered this hiccup. Thank you for your advice.
<br><br>Cheers,<br><br>Seth<br><br><div><span class="gmail_quote">On 9/19/06, <b class="gmail_sendername">Konrad Hinsen</b> <<a href="mailto:hinsen@cnrs-orleans.fr" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
hinsen@cnrs-orleans.fr</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On Sep 19, 2006, at 2:38, Dr. Seth Olsen wrote:<br><br>> As the eigenvectors() routine in LinearAlgebra.py is obviously<br>> going to be a heavily used algorithm (not just in MMTK), is it<br>> really possible that no one has had this problem before?
<br><br>This is a well-known problem and has been discussed on this list<br>before. It concerns most (if not all) Linux distributions that use<br>GCC 4. GCC 4 does not translate CLAPACK code correctly when used with<br>its default options, though I can't say if this is GCC's or f2c's
<br>fault (CLAPACK is LAPACK translated to C using f2c). Both Numeric and<br>MMTK contain CLAPACK code.<br><br>The solution is to install both Numeric and MMTK using<br><br> CFLAGS='-ffloat-store' python setup.py build
<br><br>in place of the simple<br><br> python setup.py build<br><br>With MMTK 2.5.14, this is included already in setup.py, but you will<br>still have to apply the fix for Numeric.<br><br>Konrad.<br>--<br>---------------------------------------------------------------------
<br>Konrad Hinsen<br>Centre de Biophysique Moléculaire, CNRS Orléans<br>Synchrotron Soleil - Division Expériences<br>Saint Aubin - BP 48<br>91192 Gif sur Yvette Cedex, France<br>Tel. +33-1 69 35 97 15<br>E-Mail: <a href="mailto:hinsen@cnrs-orleans.fr" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
hinsen@cnrs-orleans.fr</a><br>---------------------------------------------------------------------<br><br><br></blockquote></div><br><br clear="all"><br>-- <br>ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms
<br><br>Dr Seth Olsen, PhD<br>Postdoctoral Fellow, Biomolecular Modeling Group<br>Centre for Computational Molecular Science<br>Australian Institute for Bioengineering and Nanotechnology (Bldg. 75)<br>The University of Queensland
<br>Qld 4072, Brisbane, Australia<br><br>tel (617) 3346 3976<br>fax (617) 33654623<br>email: <a href="mailto:s.olsen1@uq.edu.au" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">s.olsen1@uq.edu.au</a>
<br>Web: <a href="http://www.ccms.uq.edu.au" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">www.ccms.uq.edu.au</a> <br>
<br>ccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccmsccms<br>The opinions expressed here are my own and do not reflect the positions of the University of Queensland.