Jump to content

  • Log In with Google      Sign In   
  • Create Account


#Actualmax343

Posted 04 November 2012 - 06:00 PM

Just a note, generally computing the cov matrix in order to find its eigenvectors is a grave mistake (may lead to a very serious round-off error).
A much better approach is to find the SVD directly, that is right after you QR'ed the deviations Nx3 matrix into Q Nx3 and R 3x3.

#1max343

Posted 04 November 2012 - 05:35 PM

Just a note, generally computing the cov matrix in order to find its eigenvalues/eigenvectors is a grave mistake (may lead to a very serious round-off error).
A much better approach is to find the SVD directly, that is right after you QR'ed the deviations Nx3 matrix into Q Nx3 and R 3x3.

PARTNERS