Ceilidh ... Re: Once again about XMCQDPT

Firefly and PC GAMESS-related discussion club

Re: Once again about XMCQDPT

Dear Pasha,

a couple of additional comments in addition to what Sanya has already suggested.

1. Be careful to include all pairs of degenerate states into state-averaging procedure (and use equal weights for each of two states in pair). This is important to keep symmetry of MOs and to keep states
degenerate. Otherwise, results can be senseless. In some cases,
you may need to lower symmetry to a subgroup of the full symmetry group or even to C1 group. Below, I'm assuming you are working in C1 group.

> STATE 1ST ORDER 2ND ORDER
> 1 E(MCSCF)= -1727.1284797779 E(MP2)= -1731.6455874427

> 2 E(MCSCF)= -1726.9913511748 E(MP2)= -1731.6292601411
> 3 E(MCSCF)= -1726.9913511741 E(MP2)= -1731.5827124526

This is clearly non-sense as degenerate states should preserve their degeneracy in PT treatment. Check if the symmetry of MOs is not broken, and that you included each component of degenerate pair in PT treatment and in various averaging procedures with equal weights.

Check for correct syntax. Possibly, FF can do some strange things if you specify weights for states not included into PT:

>$xmcqdpt state(1)=1,1,1,1,0,0
> avecoe(1)=1,1,1,1,1,-0
>wstate(1)=1,1,1,1,1,-0 $end

Is this really the input you have used? I.e., four states in PT and five weights?

>Q1:
>If only 1-n states corresponds to single electron excitation, then we don't need to include another state for XMCQDPT. Or we should consider only state from CASSCF that corresponds to single electron excitation. Am i right?

Perhaps you are right, but could you be a bit more specific?

>Q2:
>What is a difference between
>kstate(1)=1,1,1,1,1,0
>avecoe(1)=1,1,1,0,0,0
>wstate(1)=1,1,1,0,0,0

>and

>kstate(1)=1,1,1,0,0,0
>avecoe(1)=1,1,1,0,0,0
>wstate(1)=1,1,1,0,0,0

>In both cases we have same number of states but weight "0"
>in second corresponding state in kstate array is "0".

This implies that the same averaging procedures are used to obtain semi-canonical orbitals and orbital energies, but different number of states enter PT treatment (i.e., the dimensions of the model space and the effective Hamiltonian differ).

>Q3:
>What should we look in an output file to check XMCQDPT results
>"EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN"
>"EIGENVALUES OF THE NON-SYMMETRIC EFFECTIVE HAMILTONIAN"
>anything else?

Basically, this was already answered here:

http://classic.chem.msu.su/cgi-bin/ceilidh.exe/gran/gamess/forum/?C35b4d3df7Uee-7819-1031-00.htm

All the best,
Alex Granovsky

On Fri Aug 19 '11 6:43pm, Solntsev Pasha wrote
----------------------------------------------
>Dear FF users.

>I have been trying to investigate an excited state of a molecule with the XMCQDPT. From CIS,TDDFT is known that formation of first excited state rises from HOMO->LUMO transition and 2nd excited state from HOMO-1->LUMO. The LUMO orbitals are double degenerated.

>To start my work, i decided consider only HOMO-1, HOMO, 2xLUMO orbitals in my active space. So, i had done SA-CASSCF(4,4)/6-311G(d,p) with nstate=6
>wstate(1)=1,1,1,1,1,0
>ntrack=5
>and found after MCSCF minimization (only the biggest coefficients are shown)
>1st state corresponds to my ground state
>STATE # 1 ENERGY = -1727.128479778
>
>
> CSF COEF OCCUPANCY (IGNORING CORE)

> --- ---- --------- --------- -----

> 1 0.964072 2200

>.....
>2nd state corresponds to single electron excitation (HOMO->LUMO)
>STATE # 2 ENERGY = -1726.991351175
>
>
>
> CSF COEF OCCUPANCY (IGNORING CORE)

> --- ---- --------- --------- -----

> 1 0.001215 2200

> 2 0.942511 2101

>..............................
> 8 0.261968 1210

>..............................
> 13 0.136427 1012

>.............................

>3rd state also corresponds to single electron excitation (HOMO->LUMO)
> STATE # 3 ENERGY = -1726.991351174
>
>
> CSF COEF OCCUPANCY (IGNORING CORE)

> --- ---- --------- --------- -----

>.................................
> 3 0.261719 1201

>.......................................
> 7 -0.942700 2110

>......................................
> 10 0.129259 2011

>....................................
> 18 -0.137203 1021

>...............................
>Rest of states don't have single electron excitation and corresponds to two electron excitation. The are not presented here.

>It is of interest to note, there is no state corresponding to excitation from HOMO-1->LUMO. Such states exist only as a minor component of another transition (HOMO->LUMO). Is it problem with active space?

>After that i did XMCQDPT and used
>$xmcqdpt state(1)=1,1,1,1,0,0
> avecoe(1)=1,1,1,1,1,-0
>wstate(1)=1,1,1,1,1,-0 $end

>I got following results:
> *** MC-XQDPT2 ENERGIES ***

> -----------------------------------------------------------------------

> STATE 1ST ORDER 2ND ORDER

> 1 E(MCSCF)= -1727.1284797779 E(MP2)= -1731.6455874427

> 2 E(MCSCF)= -1726.9913511748 E(MP2)= -1731.6292601411

> 3 E(MCSCF)= -1726.9913511741 E(MP2)= -1731.5827124526

> 4 E(MCSCF)= -1726.9163689912 E(MP2)= -1730.2704096821

>EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN
>
>
>
> 1 2 3 4
>
>
>
> ********************************************
>
>
>
> 1 0.990372 0.120637 0.025590 0.062892

> 2 0.107653 -0.972246 0.185035 0.094396

> 3 0.044631 -0.178233 -0.982213 0.038706

> 4 -0.074714 0.091749 0.019079 0.992792
>
>
>
> EIGENVALUES OF THE NON-SYMMETRIC EFFECTIVE HAMILTONIAN
>
>
>
> 1 -1731.63989383779100000000 + I * 0.00000000000000000000

> 2 -1731.62858390071600000000 + I * 0.00000000000000000000

> 3 -1731.58269032210300000000 + I * 0.00000000000000000000

> 4 -1730.27680165785000000000 + I * 0.00000000000000000000

>As we can see from "EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN
>" part i steel have contribution from 2nd CASSCF state to the 1st XMCQDPT. And the same situation for 2nd and 3rd states for XMCQDPT.

>Experimental value for 1st excited state is 564 nm or 17730 cm^-1 or 0.08078 a.u.

>It looks i very close to finish, but maybe i missed something.

>Also i have couple questions to clarify my understanding.

>Q1:
>If only 1-n states corresponds to single electron excitation, then we don't need to include another state for XMCQDPT. Or we should consider only state from CASSCF that corresponds to single electron excitation. Am i right?

>Q2:
>What is a difference between
>kstate(1)=1,1,1,1,1,0
>avecoe(1)=1,1,1,0,0,0
>wstate(1)=1,1,1,0,0,0

>and

>kstate(1)=1,1,1,0,0,0
>avecoe(1)=1,1,1,0,0,0
>wstate(1)=1,1,1,0,0,0

>In both cases we have same number of states but weight "0"
>in second corresponding state in kstate array is "0".

>Q3:
>What should we look in an output file to check XMCQDPT results
>"EIGENVECTORS OF THE EFFECTIVE HAMILTONIAN"
>"EIGENVALUES OF THE NON-SYMMETRIC EFFECTIVE HAMILTONIAN"
>anything else?

>I will be very appreciate for you help.
>If you need extra information about my current problem just let me know.

>Best wishes.

>Pavel.
>
>
>
>

Sat Aug 20 '11 6:18pm

This message read 990 times