# Revision history [back]

Step 1 - Discussion

1) The subspace spanned by eigenvectors U, 2) the subspace spanned by eigenvectors V, and 3) the subspace spanned by meanX-meanY. The last of these is a single vector.

T = [U,v]

G =  transpose(U) V

H = V - U G

These zero vectors are removed to leave Hnq0 . We also compute the residue h of yÿx with respect to the eigenspace of using (6)

h = x U g (6)

Step 1 - Discussion

1) The subspace spanned by eigenvectors U, 2) the subspace spanned by eigenvectors V, and 3) the subspace spanned by meanX-meanY. The last of these is a single vector.

T = [U,v]

G =  transpose(U) V

H = V - U G

These zero vectors are removed to leave Hnq0 H . We also compute the residue h of yÿx neanY-meanX with respect to the eigenspace of using (6)

h = x U g (6)

v can now be computed by finding an orthonormal basis for [H, h], which is sufficient to ensure that ns is orthonormal. Gramm-Schmidt orthonormalization [12] may be used to do this: v =  Orthonormalize([H, h])