Eigenvalue calculation aborts

[Muthiah Annamalai]

Hello there,

I have been using 64-bit Octave version 3.2.3 compiled on Ubuntu 9.10.

For the most part Octave solves my problem great. For large data size
( sparse matrix size > 4GB)
the eigenvalue calculation in the following code aborts. QDs is sparse
and symmetric.

QDs = [real(QDs); imag(QDs)];
G = QDs*(QDs.');
l = eig( G );
flipud(l)

I saw this post here,
http://old.nabble.com/R:-eigs-crash-on-a-particular-sparse-matrix-td26423079.html
,
but I'd like to know any new developments to overcome the problem, or
try and use like the posted
suggestion, G = G + j*eps;

My Octave configuration goes like this:

Configuration
-----------------------------------------------

uname output:     Linux fresnel 2.6.31-17-generic #54-Ubuntu SMP Thu
Dec 10 17:01:44 UTC 2009 x86_64 GNU/Linux
configure opts:   '--prefix=/usr' '--enable-shared' '--enable-64'
Fortran compiler: gfortran
FFLAGS:           -O
FLIBS:            -L/usr/X11R6/lib
-L/usr/lib/gcc/x86_64-linux-gnu/4.4.1
-L/usr/lib/gcc/x86_64-linux-gnu/4.4.1/../../../../lib -L/lib/../lib
-L/usr/lib/../lib -L/usr/lib/gcc/x86_64-linux-gnu/4.4.1/../../..
-L/usr/lib/x86_64-linux-gnu -l$
CPPFLAGS:         -I/usr/include/freetype2
INCFLAGS:         -I. -I. -I./liboctave -I./src -I./libcruft/misc
C compiler:       gcc, version 4.4.1 (Ubuntu 4.4.1-4ubuntu8)
CFLAGS:           -g -O2
CPICFLAG:         -fPIC
C++ compiler:     g++, version 4.4.1
CXXFLAGS:         -g -O2 -I/usr/include/freetype2
CXXPICFLAG:       -fPIC
LD_CXX:           g++
LDFLAGS:
LIBFLAGS:         -L.
RLD_FLAG:         -Wl,-rpath -Wl,/usr/lib/octave-3.2.3
BLAS_LIBS:        -llapack -lblas
FFTW_LIBS:        -lfftw3 -lfftw3f
LIBS:             -lreadline  -lncurses -ldl -lz -lm  -lfreetype -lz
-L/usr/X11R6/lib -lGL -lGLU
LEXLIB:
LIBGLOB:
SED:              /bin/sed
DEFS:

  -DPACKAGE_NAME="" -DPACKAGE_TARNAME="" -DPACKAGE_VERSION=""
  -DPACKAGE_STRING="" -DPACKAGE_BUGREPORT="" -DPACKAGE_URL=""


Thanks,
-Muthu