Molcas Forum

panwang

Member

Registered: 2020-02-19

Posts: 6

Convergence problem of CI optimization with QM/MM

Dear all,

I used the QM/MM in molcas to investiage the photochemistry of an organic molecule dissolved in water, I've successfully optimized the ground state and the first excited state geometries. However, when I try to optimize the MECI between the S0 and S1, I encountered a very difficult convergence problem. The energy oscillates back and forth between the two similar geometries, at first geometry:     
energy=   -1488.282710 (S0)
energy=   -1488.276618 (S1)

and at the second geometry:
      energy=   -1488.278701  (S1)
      energy=   -1488.273485  (S0)

The root-flip occurred and I have used the track keyword. I have tried to reduce the maxstep to 0.1 and it didnot help.

So, is there a way to deal with this situation. Thanks.

Following is my input:

> EXPORT  TINKER=$MOLCAS/tinker/bin
> COPY  /home/pwzhou/Fluorescein/QM-MM/TG/conical-opt/TG-conical-opt.xyz $WorkDir/TG-conical-opt.xyz
> COPY  /home/pwzhou/Fluorescein/QM-MM/TG/conical-opt/TG-conical-opt.key $WorkDir/TG-conical-opt.key
&Gateway
 RICD
 Tinker
 Basis=6-31G*
 Group=Nosym
 Constraints
  E = Ediff 1 2
 Values
  E = 0.0
 End of constraints
>>> EXPORT MOLCAS_MAXITER=300
>>> Do While
 &Seward
  DoAnalytical
 &Espf
  External=Tinker
  LAMorok
 &RASSCF
  LumOrb
  FileOrb=s1.RasOrb
  Spin=1
  Inactive=103
  NACTEL=12
  RAS2=12
  CIRoot=2 2 1
 &SLAPAF
  Maxstep = 0.1
  Cartesian
  rHidden = 20 angstrom
  Track
  THRS = 1.0e-4 3.0e-3
>>> EndDo
&CASPT2
 IPEAshift=0.0
 Multistate=2 1 2
 MAXIter=100

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Convergence problem of CI optimization with QM/MM

I'm not sure the Track keyword is any good for CI optimization. Try first without it. It could be that there is a root flipping with a higher state that you didn't include in the average, try with more states in CIRoot.

panwang

Member

Registered: 2020-02-19

Posts: 6

Re: Convergence problem of CI optimization with QM/MM

Thaks. Actually I first didnot use the Track keyword and the similar problem occurrs; then I added the Track keyword and the problem still occurrs.
I will try to increase more states, thanks.