Questions about printing out unoccupied molecular orbitals (LUMO, LUMO+x)

zh vale... at gmail.com
Thu Mar 4 09:48:53 UTC 2010


Thank all of your for your reply and discussion!

Now I understand the problem and know how to get the unoccupied
molecular orbitals in a reasonable way.

The problem can be summarized as follows:
i) ADDED_MOS = 0, NLUMO  >=0
The cube files of unoccupied molecular orbitals will be generated in a
normal way (i.e., what we expected). The eigenvalues and eigevectors
of unoccupied molecular orbitals from LUMO to HOMO+NLUMO will be
calculated by calling "make_lumo()".  But we should take care of
whether these unoccupied molecular orbitals are well converged or not.
ii) ADDED_MOS >0 , NLUMO =0
No cube files of unoccupied molecular orbitals will be generated. The
eigenvalues of unoccupied molecular orbitals from LUMO to HOMO
+ADDED_MOS will be calculated during the self-consistent calculation.
iii) ADDED_MOS >0, NLUMO >0
The eigenvalues and eigevectors of unoccupied molecular orbitals from
LUMO to HOMO+ADDED_MOS will be calculated during the self-consistent
calculation, while  the eigenvalues and eigevectors of unoccupied
molecular orbitals from HOMO+ADD_MOS+1 to HOMO+ADD_MOS+NLUMO will be
calculated by calling "make_lumo()". The index for the cube files of
unoccupied molecular orbitals is wrong.  In the current version of
source code, the index for the cube files of unoccupied molecular
orbitals starts from LUMO. Actually, it should start from HOMO
+ADDED_MOS+1.
iv) ADDED_MOS =0, NLUMO = -1
The cube files of unoccupied molecular orbitals will be generated in a
normal way (i.e., what we expected). The eigenvalues and eigevectors
of unoccupied molecular orbitals from LUMO to NAO-HOMO (NAO is the
number of atomic orbitals) will be calculated by calling
"make_lumo()". But we should take care of whether these unoccupied
molecular orbitals are well converged or not.
v)ADDED_MOS >0, NLUMO = -1
The eigenvalues and eigevectors of unoccupied molecular orbitals from
LUMO to HOMO+ADDED_MOS will be calculated during the self-consistent
calculation, while the eigenvalues and eigevectors of unoccupied
molecular orbitals from HOMO+ADD_MOS+1 to NAO-HOMO (NAO is the number
of atomic orbitals) will be calculated by calling "make_lumo()". The
index for the cube files of unoccupied molecular orbitals is wrong.
In the current version of source code, the index for the cube files of
unoccupied molecular orbitals starts from LUMO. Actually, it should
start from HOMO+ADDED_MOS+1.



More information about the CP2K-user mailing list