When passing in the the error_vector to the function compute_new_vector in the class DIIS, we have to make sure the sign is correct. For the amplitudes we presume that error_vector should have the opposite sign of direction_vector, when the latter is the right hand sides divided by the d-tensor. That is, for instance the following code from the CoupledClusterDoubles-class
When passing in the the
error_vector
to the functioncompute_new_vector
in the classDIIS
, we have to make sure the sign is correct. For the amplitudes we presume thaterror_vector
should have the opposite sign ofdirection_vector
, when the latter is the right hand sides divided by thed
-tensor. That is, for instance the following code from theCoupledClusterDoubles
-classshould be corrected to
However, this should be checked properly.