Firefly and PC GAMESS-related discussion club

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

Re^5: MCSCF for a large system.

Pavlo Solntsev

Dear Alex.

I am not expert in the modeling of reactions using MCSCF wavefunction, but i think it is more natural to take a look on a bond breaking point first. If you see two or three states close to each other (say, singlet and triplet or doublet and quartet), they should be included into state average. I am not sure when fractional contribution of a state should be considered, but maybe integer contribution is nice approach to go. In your case, potential states of interest are #1, #2. For a total number of 5 states, the input may look like nstate=5 wstate(1)=1,1,-0.

Just my two cents.


On Fri Sep 12 '14 8:27pm, alex wrote
>Thanks Pavlo! I am realizing the importance of adding the states but still can’t understood how to combine them appropriately. Lets consider (typical) situation of bond breaking with two states of interest: #1 ground only (COEF=0.8) and #2 a combination of 3 excitations - one is the desired biradical (COEF=0.6) and a two ‘others’ (COEF=0.2 and COEF=0.1). What weight (WSTATE values) would you use for the two states? Or at least what the strategy is better from your experience WSTATE(1)=1,1 or WSTATE(1)=0.6,0.8 or WSTATE(1)=0.9,0.8?

>On Fri Sep 12 '14 5:21am, Pavlo Solntsev wrote
>>What i usually do is to run single point with one weighted state: wstate(1)=1,-0 and nstate=5 for example. Check carefully all states, what they are. If you see some low lying excited state, they may affect your ground state. Then you should include them into average. You may increase nstate if needed. Details, for SA-CASSCF described very well in manual. Some examples also available.


[ Previous ] [ Next ] [ Index ]           Mon Sep 15 '14 3:18pm
[ Reply ] [ Edit ] [ Delete ]           This message read 499 times