Firefly and PC GAMESS-related discussion club

 
Learn how to ask questions correctly  
 
 

 

Alex Granovsky
gran@classic.chem.msu.su

Sorry for some delay with my response.

Please see my comments below.


On Mon Oct 22 '12 7:55am, Panwang Zhou wrote
--------------------------------------------
>Dear Alex,

>I want to calculate vertical excitation energy between the ground state and the first excited state with XMCQDPT2, first I do a state-averaged MCSCF on the optimized S0 geometries, with the following settings:

>$MCSCF CISTEP=ALDET FULLNR=.F. SOSCF=.T. MAXIT=100 $END
>$DET NCORE=72 NACT=12 NELS=12 NSTATE=6 WSTATE(1)=1,1 $END

>From this discussion club and the GAMESS-US manual I understand why the NSTATE should be larger than 2 as determinants do not correspond to any spin number and the triplet will also be obtained when the CISTEP=ALDET. The WSTATE(1)=1,1 will exclude the triplet and only do averaging on the first and second singlet.


That's correct. In addition, Firefly v. 8.0.0 includes
the possibility to filter triplets out but it's a different story

>Then I using the optimized MCSCF MOs to do the XMCQDPT2 calculation with the following settings:

>$MCSCF CISTEP=ALDET FULLNR=.F. SOSCF=.T. MAXIT=100 $END
>$DET NCORE=72 NACT=12 NELS=12 NSTATE=6 WSTATE(1)=1,1 $END
>$XMCQDPT NSTATE=10 EDSHFT=0.02 THRGEN=1D-12 INORB=2 $END
>$XMCQDPT WSTATE(1)=1,1,-0 AVECOE(1)=1,1,-0 $END

>Is this correct for calculating the vertical excitation energy between S0 and S1?

It is correct provided you put the converged MOs into the $vec group
of the input file and moread it using

 $guess
  guess=moread norb=the_total_number_of_MOS
 $end

In addition, for better precision I'd suggest to use

 $contrl wide=.t. $end

This will punch MOs with all significant digits.

Alternatively, you can set inorb to zero and perform MCSCF then
XMCQDPT2 computations using single input file.


>I set NSTATE=10 because sanya has said "The size of the effective
Hamiltonian given by NSTATE has serious effect on the transition
energy (the energy grows with NSTATE and levels off at NSTATE~10)."

That's not always true i.e. quite often nstate=2 is enough.
However, unlike with other MS-MR-PT2 theories, with XMCQDPT2
you do can improve results increasing NSTATE.


>However, I'm not clear the relation between the NSTATE, WSTATE
> and AVECOE in XMCQDPT2, could you comment on this? In the firefly
> manual I could not find the explain for WSTATE in XMCQDPT.

Both WSTATE and AVECOE arrays are used to define the state's weights
used compute state-averaged density matrices. Technically, the
averaging used to define semi-canonical orbitals and to define
orbital energies of these orbitals, can be different. If both WSTATE
and AVECOE are given, the first one defines the averaging used to
compute orbitals while the second is used for energies. If only one
of them is given in input, then the second one will be set to that
provided in the input so that WSTATE and AVECOE will be identical.
The latter option should be used almost exclusively unless one has
really good reasons why to use different averaging for different
stages.  


Normally, for S0->S1 transitions one needs to average over S0 and S1
only and with equal weights i.e. to set WSTATE(1)=1,1,-0

>Additional two small questions:
>From this dicussion club I know there are two ways to avoid CASCI integral transformation step in XMCQDPT calculation.
>(1) First do the MCSCF calculation with the following settings:
>$mcscf iforb=.t. sd=.t. $end
>$contrl wide=.t. $end

>The do the XMCQDPT with:
>$guess guess=moread norb=the_overall_number_of_mos $end
>$xmcqdpt inorb=2 iforb=0 $end

>My question is: when the iforb is true in $mcscf, can the obained MOs be used for next MCSCF calculation, where the geometry is close to the previous one.

Yes they can be used. They correspond exactly to the same overall
wavefunction but with different MOs and CI vector components.

>(2) Do the XMCQDPT wih the following settings:
>$mcscf iforb=.t. $end
>$xmcqdpt iforb(1)=-1,1,1 $end

>My question is:  what is the value should be set for inorb in this
condition?

The only valid is the default i.e. inorb=0

>
>
>Best Regards,
>Panwang Zhou

>

Kind regards,
Alex Granovsky