How to figure a state in a series of calculations

Fancy · 2018-12-03 04:47:13

Hi! Everyone, in my recently work, I do a series of calculations of SA-CASSCF, CASPT2 and it's Multi-CASPT2. The out file are as follows (take root6 and root7 as examples):
I figure root6 as 51339 state, root7 as 51212 state, so as their CASPT2 calculations. However, in Multi-CASPT2 calculation, the two state mix seriously. So, how to figure the two state in this situation? Beside, I wonder that the following three kind of printed orbitals in out files are the NO orbital of this state itself or the average-state NO orbital? Thank you very much.

SA-CASSCF out file:
printout of CI-coefficients larger than 0.05 for root 6
energy= -2456.624944
conf/sym 11111111111 Coeff Weight
15993 220020uu200 -0.15188 0.02307
47747 2200u02u200 0.11897 0.01415
51212 2u00202u200 -0.25593 0.06550
51238 2u00202ud0u 0.05894 0.00347
51286 2u00202u002 0.06777 0.00459
51293 2u0020u0202 -0.23865 0.05696
51339 u200202020u 0.84834 0.71968
51341 u20020ud20u -0.12181 0.01484
51387 u2002020u02 0.16429 0.02699
51389 u20020udu02 0.08823 0.00778
83060 2u00u020202 0.10086 0.01017

printout of CI-coefficients larger than 0.05 for root 7
energy= -2456.622688
conf/sym 11111111111 Coeff Weight
15993 220020uu200 0.33141 0.10983
16011 220020uud0u -0.08716 0.00760
16030 220020uu002 -0.09200 0.00846
47747 2200u02u200 -0.25428 0.06466
47773 2200u02ud0u 0.07574 0.00574
47807 2200u0uu20d 0.05237 0.00274
47821 2200u02u002 0.06174 0.00381
47827 2200u0uud02 -0.06658 0.00443
51212 2u00202u200 0.70326 0.49457
51236 2u00202du0u 0.07415 0.00550
51238 2u00202ud0u -0.19778 0.03912
51270 2u00202uu0d -0.09463 0.00895
51286 2u00202u002 -0.18472 0.03412
51292 2u0020uud02 0.05111 0.00261
51293 2u0020u0202 -0.10928 0.01194
51339 u200202020u 0.32553 0.10597
51371 u20020uu20d -0.07133 0.00509
51387 u2002020u02 0.07102 0.00504
83032 2u00u02u20d -0.08737 0.00763
83058 2u00u02ud02 0.10270 0.01055

CASPT2 out file:
With new orbitals, the CI array of state 6
--------------------------------------------------------------------------------
CI COEFFICIENTS LARGER THAN 0.50D-01
Occupation of active orbitals, and spin coupling
of open shells. (u,d: Spin up or down).
SGUGA info is (Midvert:IsyUp:UpperWalk/LowerWalk)
Conf SGUGA info Occupation Coef Weight
15993 ( 7:1: 3/ 31) 220020uu200 -0.075568 0.005710
51212 (11:1: 2/ 56) 2u00202u200 0.311799 0.097218
51286 (11:1: 76/ 56) 2u00202u002 -0.102780 0.010564
51293 (11:1: 83/ 56) 2u0020u0202 0.213004 0.045371
51339 (11:1: 30/ 57) u200202020u -0.813210 0.661310
51341 (11:1: 32/ 57) u20020ud20u 0.112903 0.012747
51387 (11:1: 78/ 57) u2002020u02 -0.294079 0.086483
51389 (11:1: 80/ 57) u20020udu02 -0.058266 0.003395
83060 (16:1: 70/ 16) 2u00u020202 -0.148834 0.022151
85408 (16:1: 18/ 40) uu00202d20u 0.073416 0.005390
With new orbitals, the CI array of state 7
--------------------------------------------------------------------------------
CI COEFFICIENTS LARGER THAN 0.50D-01
Occupation of active orbitals, and spin coupling
of open shells. (u,d: Spin up or down).
SGUGA info is (Midvert:IsyUp:UpperWalk/LowerWalk)
Conf SGUGA info Occupation Coef Weight
47775 (11:1: 30/ 21) 2200u02020u -0.051500 0.002652
51212 (11:1: 2/ 56) 2u00202u200 -0.809765 0.655719
51235 (11:1: 25/ 56) 2u00202200u -0.079981 0.006397
51236 (11:1: 26/ 56) 2u00202du0u -0.115098 0.013248
51238 (11:1: 28/ 56) 2u00202ud0u 0.147493 0.021754
51240 (11:1: 30/ 56) 2u00202020u 0.251750 0.063378
51286 (11:1: 76/ 56) 2u00202u002 0.264336 0.069873
51293 (11:1: 83/ 56) 2u0020u0202 0.101405 0.010283
51311 (11:1: 2/ 57) u200202u200 -0.090039 0.008107
51339 (11:1: 30/ 57) u200202020u -0.289402 0.083753
51371 (11:1: 62/ 57) u20020uu20d 0.060221 0.003627
51387 (11:1: 78/ 57) u2002020u02 -0.071964 0.005179
51392 (11:1: 83/ 57) u20020u0202 -0.054991 0.003024
83160 (16:1: 70/ 17) u200u020202 -0.062284 0.003879

Muti-CASPT2 out file:
The CI coefficients for the MIXED state nr. 6
--------------------------------------------------------------------------------
CI COEFFICIENTS LARGER THAN 0.50D-01
Occupation of active orbitals, and spin coupling
of open shells. (u,d: Spin up or down).
SGUGA info is (Midvert:IsyUp:UpperWalk/LowerWalk)
Conf SGUGA info Occupation Coef Weight
15993 ( 7:1: 3/ 31) 220020uu200 0.304879 0.092951
16011 ( 7:1: 21/ 31) 220020uud0u -0.081158 0.006587
16030 ( 7:1: 40/ 31) 220020uu002 -0.085732 0.007350
47747 (11:1: 2/ 21) 2200u02u200 -0.246722 0.060872
47773 (11:1: 28/ 21) 2200u02ud0u 0.069480 0.004828
47821 (11:1: 76/ 21) 2200u02u002 0.062085 0.003855
47827 (11:1: 82/ 21) 2200u0uud02 -0.055652 0.003097
51212 (11:1: 2/ 56) 2u00202u200 0.563005 0.316974
51236 (11:1: 26/ 56) 2u00202du0u 0.072875 0.005311
51238 (11:1: 28/ 56) 2u00202ud0u -0.146682 0.021516
51270 (11:1: 60/ 56) 2u00202uu0d -0.083910 0.007041
51286 (11:1: 76/ 56) 2u00202u002 -0.147623 0.021793
51293 (11:1: 83/ 56) 2u0020u0202 0.157217 0.024717
51339 (11:1: 30/ 57) u200202020u -0.569449 0.324272
51341 (11:1: 32/ 57) u20020ud20u 0.092441 0.008545
51387 (11:1: 78/ 57) u2002020u02 -0.106445 0.011331
51389 (11:1: 80/ 57) u20020udu02 -0.065247 0.004257
83032 (16:1: 42/ 16) 2u00u02u20d -0.078642 0.006185
83058 (16:1: 68/ 16) 2u00u02ud02 0.087870 0.007721
83060 (16:1: 70/ 16) 2u00u020202 -0.054344 0.002953
The CI coefficients for the MIXED state nr. 7
--------------------------------------------------------------------------------
CI COEFFICIENTS LARGER THAN 0.50D-01
Occupation of active orbitals, and spin coupling
of open shells. (u,d: Spin up or down).
SGUGA info is (Midvert:IsyUp:UpperWalk/LowerWalk)
Conf SGUGA info Occupation Coef Weight
15993 ( 7:1: 3/ 31) 220020uu200 0.256990 0.066044
16011 ( 7:1: 21/ 31) 220020uud0u -0.067760 0.004591
16030 ( 7:1: 40/ 31) 220020uu002 -0.071673 0.005137
47747 (11:1: 2/ 21) 2200u02u200 -0.201086 0.040436
47773 (11:1: 28/ 21) 2200u02ud0u 0.060275 0.003633
47821 (11:1: 76/ 21) 2200u02u002 0.050108 0.002511
51212 (11:1: 2/ 56) 2u00202u200 0.436750 0.190751
51238 (11:1: 28/ 56) 2u00202ud0u -0.129614 0.016800
51270 (11:1: 60/ 56) 2u00202uu0d -0.056078 0.003145
51286 (11:1: 76/ 56) 2u00202u002 -0.112881 0.012742
51293 (11:1: 83/ 56) 2u0020u0202 -0.212202 0.045030
51339 (11:1: 30/ 57) u200202020u 0.709527 0.503429
51341 (11:1: 32/ 57) u20020ud20u -0.082034 0.006730
51371 (11:1: 62/ 57) u20020uu20d -0.060659 0.003680
51387 (11:1: 78/ 57) u2002020u02 0.144004 0.020737
51389 (11:1: 80/ 57) u20020udu02 0.063241 0.003999
83032 (16:1: 42/ 16) 2u00u02u20d -0.060322 0.003639
83058 (16:1: 68/ 16) 2u00u02ud02 0.068356 0.004673
83060 (16:1: 70/ 16) 2u00u020202 0.085935 0.007385

Last edited by Fancy (2018-12-03 04:48:55)

Ignacio · 2018-12-03 09:29:23

States mix, that's a fact of life, sometimes you just have a linear combination of two things. It may be that other properties (e.g. dipole moment, Mulliken charges) show a clearer distinction, but sometimes the "which is which" question itself doesn't make sense.

The CI coefficients in RASSCF refer to the average orbitals ($Project.RasOrb). The "Natural orbitals and occupation numbers" in RASSCF refer to the natural orbitals of each state ($Project.RasOrb.n), which might be quite different from each other. The "Molecular orbitals" in RASSCF are the average orbitals. The CI coefficients in CASPT2 refer, I presume to the canonical RASSCF orbitals.

Fancy · 2018-12-03 15:14:52

Thank you very much for your immediately and kindly reply, thank you!
1. But If I really need to use the CASPT2 and Multi-CASPT2 energies of the above two states, "which is which" question doesn't make sense", how can I get their CASPT2 and Multi-CASPT2 energies?
2. BTW, according to your reply about the printed orbital information, I have the following understanding:
AS-CASSCF: natural orbitals of each state
CASPT2: canonical RASSCF orbitals of each state
Muti-CASPT2: new canonical RASSCF orbitals each state (mix orbitals of CASPT2)
The orbitals in three calculation are printed in different ways (natural orbital or canonical orbitals), right? If so, the 51212 state in CASSCF is still 51212 state in CASPT2?

Last edited by Fancy (2018-12-03 15:15:57)

Ignacio · 2018-12-03 15:37:25

Fancy wrote:

1. But If I really need to use the CASPT2 and Multi-CASPT2 energies of the above two states, "which is which" question doesn't make sense", how can I get their CASPT2 and Multi-CASPT2 energies?

For SS-CASPT2, the 2nd root (by order of computation) is the 2nd CASSCF, even if the CI coefficients look different, there's no mixing at this stage.

For MS-CASPT2, the effective Hamiltonian matrix (printed somewhere in the multistate section) should give you the mixing of the states and you can decide if it has a large enough weight for some of them.

The orbitals in three calculation are printed in different ways (natural orbital or canonical orbitals), right? If so, the 51212 state in CASSCF is still 51212 state in CASPT2?

Not necessarily. Changing the orbitals changes the CSFs and coefficients.

Fancy · 2018-12-04 04:08:21

Thank you very much.
1. Maybe I confused you about my question. This is just the point: "For SS-CASPT2, the 2nd root (by order of computation) is the 2nd CASSCF, even if the CI coefficients look different", why the CI coefficients look different? It looks different, but they come from a same CI with RASSCF, use the same conf number, right?
2. if the calculation of states from CASSCF to CASPT2 cannot figure by conf number cause they use different kind of orbitals. In my out work, I figure them just by a large weight, the state5 and state6 will both be conf 51339, it obviously wrong figure...

Ignacio · 2018-12-04 09:52:30

Fancy wrote:

1. Maybe I confused you about my question. This is just the point: "For SS-CASPT2, the 2nd root (by order of computation) is the 2nd CASSCF, even if the CI coefficients look different", why the CI coefficients look different? It looks different, but they come from a same CI with RASSCF, use the same conf number, right?

Because they are based on different orbitals. For CASSCF it is the "pseudonatural" orbitals (the average orbitals, those that diagonalize the average density matrix). For CASPT2, according to the output, it's "quasi-canonical orbitals: those which diagonalize the Fock matrix within inactive-inactive, active-active and virtual-virtual submatrices".

2. if the calculation of states from CASSCF to CASPT2 cannot figure by conf number cause they use different kind of orbitals. In my out work, I figure them just by a large weight, the state5 and state6 will both be conf 51339, it obviously wrong figure...

I don't understand what you mean here. But keep in mind that it's not only the weight that matters, the relative sign of the coefficients is significant too.

Fancy · 2018-12-07 03:54:08

Thank you very much!
In my second question, in another word, I figure the conf 51212 in AS-CASSCF and want to get the Multi-CASPT2 energy of this state. In the outfile of Multi-CASPT2 work, should I still need to figure the state whose main conf is 51212 as the Multi-CASPT2 energy of 51212 state? Although the part of CASSCF and Multi-CASPT2 printed different consititute, the main conf 51212 in two calculation represent the same occupation and same orbital, right?

Molcas Forum

Announcement

#1 2018-12-03 04:47:13

How to figure a state in a series of calculations

#2 2018-12-03 09:29:23

Re: How to figure a state in a series of calculations

#3 2018-12-03 15:14:52

Re: How to figure a state in a series of calculations

#4 2018-12-03 15:37:25

Re: How to figure a state in a series of calculations

#5 2018-12-04 04:08:21

Re: How to figure a state in a series of calculations

#6 2018-12-04 09:52:30

Re: How to figure a state in a series of calculations

#7 2018-12-07 03:54:08

Re: How to figure a state in a series of calculations

Board footer