Firefly and PC GAMESS-related discussion club

 
Learn how to ask questions correctly  
 
 

 

Alex Granovsky
gran@classic.chem.msu.su

It seems your nstate=4 XMCQDPT2 computations did not completely converged.

See pages 163-169 of the most recent manual, they may be helpful to you.

In the output file, there is a line

"EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN"

Search for it and you will find a table which columns are
contributions of all CASSCF/CASCI states into particular
XMCQDPT2 state.

Kind regards,
Alex Granovsky


On Fri Aug 21 '15 11:44am, Panwang Zhou wrote
---------------------------------------------
>Dear Prof. Alex,

>Thanks for you reply.

>>You can verify your results including more states into model space.
>>For instance, you can average over four states, while include ten states or so.

>I have done the test calculations including ten states into model space and the results are slightly different from that only including four states in model space.

>The state order of calculations are:
>i) MCSCF_S0_OPT:      s0, pipi*, npi*,  pipi*
>   MCSCF_PIPI*_OPT:   s0, pipi*, npi*,  pipi*
>   MCSCF_CI_OPT:      s0, pipi*, npi*,  pipi*
>   MCSCF_NPI*_OPT:    s0, npi*,  pipi*, pipi*
>ii)PT2_4_S0_OPT:      s0, pipi*, pipi*, npi*
>   PT2_4_PIPI_OPT:    s0, pipi*, pipi*, npi*
>   PT2_4_CI_OPT:      s0, pipi*, npi*,  pipi*
>   PT2_4_NPI*_OPT:    s0, npi*,  pipi*, pipi*
>iii)PT2_10_S0_OPT:    s0, pipi*, pipi*, npi*
>    PT2_10_PIPI_OPT:  s0, pipi*, pipi*, pipi*, npi*
>    PT2_10_CI_OPT:    s0, pipi*, pipi*, npi*
>    PT2_10_NPI*_OPT:  s0, npi*,  pipi*, pipi*

>It can be seen that, when the states number of model space is increased to ten, for pipi_opt and ci_opt, the npi* state now become to the third and fourth excited state, respectively. That is to say, a more pipi* appears below the npi* state. Now I'm not sure how to explain the new appeared pipi* state in the XMCQDPT2 calculations with nstates=10.
>And which pipi* state in the XMCQDPT2 results is corresponding to the first pipi* state in MCSCF results? Thanks.
>
>
>Kind Regards,

>Panwang Zhou