Molcas Forum

chemphys

Member

Registered: 2020-11-10

Posts: 27

CASSCF/MCPDFT Geometry Optimization

Hi everyone,
I'm hoping to optimize geometry while running a CASSCF/MCPDFT calculation and not sure if this is the correct set-up.  The output file seems to mention geometry optimization, but then it does not save a final .geo.molden file. 

My project.rasscf.inp file:

 &RASSCF &END
NACTel
 3 0 0
Spin
 2
Inactive
 21
Ras2
 4
Ras3
 0
CIRoot
 1 1 1
RLXRoot=  1
Iterations
 200 100
Cholesky
LumOrb

&MCPDFT
  KSDFT=TPBE

End of Input

My submission script:

pymolcas $Project.seward.inp > $Project.seward.log
     ln -fs $Project.RasOrb.S1.old   INPORB
     pymolcas $Project.rasscf.inp > $Project.rasscf.S1.log
     mv *.RasOrb $Project.RasOrb.S1
     mv *.JobIph $Project.JobIph.S1
     rm INPORB

Thanks for any tips!

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: CASSCF/MCPDFT Geometry Optimization

To do a geometry optimization (with any method) you have to create a loop with SLAPAF, e.g.

&GATEWAY
...

> Do While

  &SEWARD
  ...

  &RASSCF
  ...

  &MCPDFT
  ...

  &SLAPAF
  ...

> End Do

chemphys

Member

Registered: 2020-11-10

Posts: 27

Re: CASSCF/MCPDFT Geometry Optimization

Thank you very much!  So I already have a project.seward.inp file, should I put the &SEWARD in this file instead?  If not, will this work for my project.rasscf.inp file, where seward.inp specifies the basis-set in more detail?:

&GATEWAY
Title=  geo-test
Coord=  molecule.xyz
Basis=  DEF2-TZVPPD

> Do While

  &SEWARD

  &RASSCF &END
  NACTel
   3 0 0
  Spin
   2
  Inactive
   21
  Ras2
   4
  Ras3
   0
  CIRoot
   1 1 1
  RLXRoot=  1
  Iterations
   200 100
  Cholesky
  LumOrb

  &MCPDFT
   KSDFT=TPBE

  &SLAPAF

> End Do

End of Input

And will this automatically save the steps as a .geo.molden or should I also edit my submission script?

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: CASSCF/MCPDFT Geometry Optimization

Remove the "End of input" at the end. It should work, provided that is the correct active space, etc. I suggest you try a simple SCF (Hartree-Fock or DFT) optimization first.

chemphys

Member

Registered: 2020-11-10

Posts: 27

Re: CASSCF/MCPDFT Geometry Optimization

Hi, thanks for your help!  Using the code above as project.rasscf.inp (without the "End of Input") I get the following errors in project.rasscf.S1.log:

###############################################################################
 ###############################################################################
 ###                                                                         ###
 ###                                                                         ###
 ###    WARNING: There were warnings during the execution                    ###
 ###    Please, check the output with care!                                  ###
 ###                                                                         ###
 ###                                                                         ###
 ###############################################################################
 ###############################################################################
 ###############################################################################
 ###############################################################################
 ###                                                                         ###
 ###                                                                         ###
 ###    Location: AixCheck                                                   ###
 ###    File: ORDINT                                                         ###
 ###                                                                         ###
 ###                                                                         ###
 ###    Active unit. Should have been closed!                                ###
 ###                                                                         ###
 ###                                                                         ###
 ###############################################################################
 ###############################################################################
--- Stop Module: rasscf at Wed Nov 11 13:05:15 2020 /rc=_RC_INTERNAL_ERROR_ ---
*** files: xmldump
    saved to directory /global/u2/c/claired/Han-Molcas/geo-opt
--- Module rasscf spent 2 seconds ---

>>> END DO

.########################################################.
.# Some internal inconsistency of the code was detected #.
.########################################################.

My submit script is:

pymolcas $Project.seward.inp > $Project.seward.log
     ln -fs $Project.RasOrb.S1.old   INPORB
     pymolcas $Project.rasscf.inp > $Project.rasscf.S1.log
     mv *.RasOrb $Project.RasOrb.S1
     mv *.JobIph $Project.JobIph.S1
     rm INPORB

Is the project.rasscf.inp space-sensitive?  I notice I am not consistent with indentation, could this be causing the problem?  Thanks for your patience and insight!

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: CASSCF/MCPDFT Geometry Optimization

No, the input doesn't care (mostly) about spaces.