Hi.<div><br></div><div>There is one SVD available freely in the CImg Library (<a href="http://cimg.sourceforge.net">http://cimg.sourceforge.net</a>).</div><div>It is based on the code from the Numerical Recipies. As CImg is a single header, you don't need</div> <div>to install binaries or library files to make it working nicely.</div><div><br></div><div>David.</div><div><br></div><div><div class="gmail_quote">2011/4/1 Pascal Monasse <span dir="ltr"><<a href="mailto:monasse@imagine.enpc.fr">monasse@imagine.enpc.fr</a>></span><br> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">Hi all,<br> <br> First, I really regret to have to add a library dependency for such a small<br> part of the algorithm (SVD), but I could not find an independent and free<br> implementation, and it seems too complex to do it right by ourselves without a<br> significant effort.<br> <br> Now, two of the solutions, LAPACK and GSL, would need to be installed (they<br> are not by default on general purpose linux distributions), with difficulties<br> for Mac and Windows users. They are also too large to be included in the<br> source code.<br> <br> The advantage I see with eigen, apart from the clean interface, is that it is<br> not too large to be included in the tarball, though it is still almost 3MB. On<br> top of that, it is composed of only header files, so no build is necessary and<br> no library is created. If we isolate the footprint of eigen from our main<br> code, with a minimal interface to it, it should be easy for a user to replace<br> it with his preferred alternative while allowing easily a first test without<br> trouble.<br> <br> Best,<br> Pascal<br> <br> On Friday, April 01, 2011 10:59:05 am Pierre Moulon wrote:<br> > Hi all,<br> ><br> > My opinion (from an integration side).<br> ><br> > Lapack and GSL need to be compiled as a library and linked to the code in<br> > order to be used.<br> > They are huge and take a long time to compile. On linux they are already<br> > compiled on the machine so it's easy to use them.<br> > On windows we have to use pre-compiled libraries (so we have to add in the<br> > archive the .lib and .dll => Does we want to add 32/64 bits version.... ?)<br> ><br> > For Eigen, we use it add a header library.<br> > So when you compile your application the matrix liibrary and related<br> > fonction will be compiled on the fly. It makes the compilation more easy<br> > because the process is the same for all the platform.<br> ><br> > I think Pascal "black box" idea is a good one.<br> > => So code could be personalized for own purpose if someone want to do it.<br> > A default solution is purposed by the way of (Eigen, ...).<br> ><br> > PS : I add the reference over Eigen <a href="http://eigen.tuxfamily.org/" target="_blank">http://eigen.tuxfamily.org/</a><br> ><br> > Regards,<br> > Pierre<br> _______________________________________________<br> discuss mailing list<br> <a href="mailto:discuss@list.ipol.im">discuss@list.ipol.im</a><br> <a href="http://tools.ipol.im/mailman/listinfo/discuss" target="_blank">http://tools.ipol.im/mailman/listinfo/discuss</a><br> </blockquote></div><br><br clear="all"><br>-- <br>-----------------------------------------------------------------------------<br>David Tschumperlé<br>CNRS Researcher<br>GREYC (UMR-CNRS 6072) E-mail: <a href="mailto:David.Tschumperle@greyc.ensicaen.fr">David.Tschumperle@greyc.ensicaen.fr</a><br> 6, Bd du Marechal Juin Tel: +33 (0)2-31-45-29-24<br>F-14050 CAEN Cedex Fax: +33 (0)2-31-45-26-98<br>France <a href="http://www.greyc.ensicaen.fr/~dtschump">http://www.greyc.ensicaen.fr/~dtschump</a><br> -----------------------------------------------------------------------------<br><br> </div>