Closed felipeZ closed 4 years ago
Thanks @felipeZ!
The only thing missed is passing/using the residues in the correction vectors:
This is how it is implemented in Champ:
correction= compute_DPR( rs, parameters, eigenvalues_sub, diag_mtx, diag_stx)
Then in the compute_DPR function:
diag = 0.0_dp
do j = 1, parameters%basis_size
diag= eye( m , m, eigenvalues( j))
correction( :, j)= rs( :, j)
do ii= 1, size( correction, 1)
correction( ii, j)= correction( ii, j)/( eigenvalues( j)* diag_stx( ii)- diag_mtx( ii))
end do
end do
fixed #34