On Sun, 21 Feb 2010 17:09:13 +0100, Alan Grimes <agri...@speakeasy.net> wrote:
> > The case where it fails is if the matrix is still good but is sparse. It > will frequently fail to find a permutation that will solve the matrix. > This causes a number of the failures we are seeing in the final > simulation. > A x = b So problems appear when A matrix fill of zeroes, and a few nonzero elements, and the right side also a few nonzero elements? Or the right side doesn't matter? If you have any matrix that might fail, write it here, so I'll try to make a testcase from it. The testFuzzySpase testcase was intented to test such problems, see here: < http://ktechlab.git.sourceforge.net/git/gitweb.cgi?p=ktechlab/ktl-zoltan_p;a=blob;f=matrixtest/matrixtester.cpp;h=f009b87ec0492e20c8d2ae9295659594c2058b1b;hb=testing-1#l360 > Strange.. It seems that this test somehow didn't made it to Julian's kde4-port branch. ------------------------------------------------------------------------------ Download Intel® Parallel Studio Eval Try the new software tools for yourself. Speed compiling, find bugs proactively, and fine-tune applications for parallel performance. See why Intel Parallel Studio got high marks during beta. http://p.sf.net/sfu/intel-sw-dev _______________________________________________ Ktechlab-devel mailing list Ktechlab-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/ktechlab-devel