Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly  
We are NATO-free zone

Re^3: Minimum number of NSTATE required in MCSCF calculation


Dear Patrick,

For state-specific case, you should set WSTATE(1)=0.,1.
However, without state-averaging, you'll very likely obtain sort of "variational collapse", i.e., in spite of Wstate, the electron density and the energy of the resulting state will resemble ground state rather than excited one. So, I'd recommend you to set WSTATE(1)=0.1,1. for excited-state optimization and WSTATE(1)=1.,0.1 for ground-state optimization.

The detailed explanations about state-specific optuimizations for state-averaged CASSCF are here: and here:

On Sun Oct 14 '12 8:00pm, Patrick SK Pang wrote
>Dear Sanya,

>Now, I conduct the geometry optimization of excited states using CASSCF without state averaging. If I want to calculate the first singlet excited state, how can I set the WSTATE(1)? WSTATE(1)=0,1 or WSTATE(1)=0,0,1?
>I also set MULT=1 in $CONTRL.



>On Tue Oct 9 '12 1:14am, sanya wrote
>>If you set CISTEP=ALDET (rather than CISTEP=GUGA), the program generates determinants, which do not correspond to any spin number. Spin eigenfunctions are constructed from the determinants; some combinations correspond to S=0 (singlet), some to S=1 (triplet), etc., with the same Sz=0 (spin multipicity=1). So, ALDET procedure generates both triplets and singlets, and you require only singlets. According to the Hund's rules, for a given electron configuration, the term with maximum multiplicity has the lowest energy. That is, for the configuration ...2110... (numbers denote orbital populations), the triplet state (S=1) is always lower than the singlet state (S=0). So, the first state is singlet ...2200..., the second state is triplet ...2110..., and the third state is singlet ...2110...

>>When you set CISTEP=GUGA, the program generates configuration state functions, that is the combinations of determinants, which are spin eigenfunctions. So, with CISTEP=GUGA the first state will be singlet ...2200..., and the second state will be singlet ...2110...

>>However, at present, ALDET algoritm is more efficient and less memory-consuming for large active spaces.

>>On Mon Oct 8 '12 6:46pm, Patrick SK Pang wrote
>>>Dear all,

>>>When I am interested in the ground state of a molecule (charge=0 and spin multipicity=1), I specify ISTATE=1 and NSTATE=1, and the calculation can successfully completed.
>>>However, when I want to calculate the 1st excited state a molecule (charge=0 and spin multipicity=1), I specify ISTATE=2, NSTATE=2, WSTATE=0,1, and an error appears due to the wrong number of NSTATE. Then, I change to NSTATE=3, and the calculation can proceed.



[ Previous ] [ Next ] [ Index ]           Mon Oct 15 '12 2:13pm
[ Reply ] [ Edit ] [ Delete ]           This message read 737 times