When decomposing a matrix with elements that are very small there are
situations where this can cause a divided by zero error. A potential fix has
been committed that changes the way exceptional shifts, wilkinson step, and
other parts of the code are handled. Also more checks are in place for zeros.
The fix is current being verified.
Original issue reported on code.google.com by peter.ab...@gmail.com on 2 Jul 2010 at 5:08
Original issue reported on code.google.com by
peter.ab...@gmail.com
on 2 Jul 2010 at 5:08