GramSchmidt.Solve has different behavior (SparseMatrix vs DenseMatrix)


Using SparseMatrix I get coefficients equal to zero, but when I use DenseMatrix, I get nonzero values for all coefs. Program to reproduce attached. Run and check standard output for differences. For now I will just use DenseMatrix, since that result seems to make more sense.

file attachments

Closed Sep 6, 2009 at 8:05 AM by cuda


dnum wrote Aug 28, 2009 at 12:46 AM

This was using 2009.8 Beta, managed code.

cuda wrote Aug 29, 2009 at 7:46 AM

I'll have a fix in early next week. Thanks, Marcus

cuda wrote Sep 6, 2009 at 8:05 AM

The problem was how we the sparse vector and matrices were testing for zero values.

wrote Sep 6, 2009 at 8:05 AM

wrote Feb 14, 2013 at 2:00 AM

wrote May 16, 2013 at 7:16 AM