Molcas Forum

Rae

Member

Registered: 2016-11-24

Posts: 2

[SOLVED] how to follow a root in CASSCF calculation with PCM

Dear all,

I'm currently working on excited states calculation considering the solvent effect with PCM. The 6th root of the gas phase 7-root sa-casscf calculation is a complete charge transfer state with a very large dipole moment, while the roots 2 to 5 are local excited states with small dipole moment. Now I want to calculate the solvent effect on this CT state. However, when I do equilibrium PCM calculation using RFroot=6 in RASSCF module, the root flips and I cannot get the correct result. The following is the Input file I use, and c2-s0-gas.RasOrb is the RasOrb file from the above mentioned gas phase CASSCF calculation.

&GATEWAY
 Title= c2-s0-vert
 coord =  c2-s0.xyz
 group= c1
 basis = 6-31G**
 RICD
 RF-Input
   PCM-Model
   Solvent=Acetonitrile
 End of RF-Input
&SEWARD
>>> COPY c2-s0-gas.RasOrb INPORB

&RASSCF
 LUMOrb
 Spin= 1; Nactel= 12 0 0; Inactive= 131; Ras2=10
 CiRoot= 7 7 1
 RFroot=6

I have read the related part in molcas manual. It seems the keywords CISElect and CIRFroot could be used to follow the root of insterest. However, I'm not clear how to use these keywords, especially CISElect. Could anyone give a sample input or is there any other way to solve this problem?

Thanks so much,
Rae

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: [SOLVED] how to follow a root in CASSCF calculation with PCM

With CISelect, you specify a reference CI vector (CSFs and coefficients) to select the root(s) by maximum overlap. For instance, suppose you want to select a root that looks like this:

      printout of CI-coefficients larger than  0.05 for root  2
      energy=    -224.805438
      conf/sym  11111111     Coeff  Weight
             1  22222000   0.05961 0.00355
             2  2222ud00   0.66009 0.43571
             4  2222u0d0  -0.19386 0.03758
             5  22220ud0  -0.12703 0.01614
             6  22220020  -0.09285 0.00862
            11  222u2d00  -0.08777 0.00770
            12  222ud200   0.21060 0.04435
            13  222u20d0  -0.51391 0.26411
            14  222udud0   0.20413 0.04167
            17  222ud020  -0.13781 0.01899
            51  2u222d00   0.12973 0.01683
            52  2u22d200   0.15014 0.02254
            54  2u22dud0  -0.08997 0.00809
            91  22202200  -0.12988 0.01687
            92  22202ud0   0.08262 0.00683
            94  22202020   0.08651 0.00748
            95  2220ud20  -0.05324 0.00283
           131  2u2d2200   0.07992 0.00639
           132  2u2d2ud0   0.07027 0.00494
           286  0222ud02  -0.05732 0.00329

which is mainly two configurations (2222ud00, 222u20d0) you could add this in the input:

CISelect
2 * two configurations
2 13 * number 2 (2222ud00), number 13 (222u20d0)
0.66 -0.51 * reference coefficients

but as the orbitals change the CI coefficients change too, and this may not be a good indicator for the desired root. Maybe you'll need a series of calculations toggling CIOnly, to make sure you get the right state at the end.

Rae

Member

Registered: 2016-11-24

Posts: 2

Re: [SOLVED] how to follow a root in CASSCF calculation with PCM

Thank you very much for your kindly reply.
I have tried the keywords CISElect and CIRFroot in my calculation. As you say, the orbitals change in rasscf iteraions, and iteration stops when the overlap of some root is too small. After getting this result, I re-selected the configurations and CI coefficients according to the previous output, read the previous RasOrb file and calculated it once again. And luckily I got the right result this time! Thank you so much!

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: [SOLVED] how to follow a root in CASSCF calculation with PCM

Glad to read that. Thank you for reporting back