⇤ ← Revision 1 as of 2014-11-28 08:03:36
Size: 891
Comment:
|
Size: 1044
Comment:
|
Deletions are marked like this. | Additions are marked like this. |
Line 16: | Line 16: |
One uses U-TD-DFT solver to solve the X-TD-DFT in '''spin-orbit''' basis: | One uses U-TD-DFT solver to solve the X-TD-DFT in '''spin-orbit''' basis ['''recommended! for efficiency''']: |
Line 34: | Line 34: |
The other one uses the original spin-adapted TD-DFT solver to solve the X-TD-DFT in '''spin-tensor''' basis: | The other one uses the original spin-adapted TD-DFT solver to solve the X-TD-DFT in '''spin-tensor''' basis ['''This is just for MO-based algorithm, which requires traint, and hence not recommended for large applications''']: |
Open-shell Systems : U-TD-DFT and spin-adapted TD-DFT for spin-conserving excitations
Using a UKS reference in SCF, the input for U-TD-DFT reads:
$TDDFT IMETHOD 2 ISF 0 ... $END
Using a ROKS reference in SCF, the input for Spin-adapted TD-DFT (X-TD-DFT) can be set in two ways.
One uses U-TD-DFT solver to solve the X-TD-DFT in spin-orbit basis [recommended! for efficiency]:
$TDDFT IMETHOD 2 ISF 0 ... icorrect 1 itest 1 itrans 1 $END
The option itrans will transform the eigenvector in CV(aa),CV(bb) basis into CV(0) and CV(1) basis.
The other one uses the original spin-adapted TD-DFT solver to solve the X-TD-DFT in spin-tensor basis [This is just for MO-based algorithm, which requires traint, and hence not recommended for large applications]:
$TDDFT imethod 3 isf 0 ... itest 1 icorrect 1 $END
These two ways give exactly the same excitation energies. The only different is the representation if "itrans" is not used.