Molcas Forum

Anugraha

Member

Registered: 2025-12-17

Posts: 22

CASSCF excited-state optimization fails for ππ* state

Dear OpenMolcas developers,

I am performing a CASSCF excited-state optimization (CAS(8,8)) for the excited pi pi star state.

& GATEWAY
 Title = TUCA 1 methanol pi pi star casscf(8,8) opt nofloat -0.82729 angstrom   
 coord = t_uca_1meth_MP2_opt_cc_pvdz.xyz
 Group = z
 Basis set = aug-cc-pvdz
 RICD
 Constraints
 r1 = bond H13 N12
 r2 = bond H13 O17
Values
 1.0 r1 - 1.0 r2 = -0.82729 angstrom
End of Constraints

>>> EXPORT MOLCAS_MAXITER = 2000

>>> DO WHILE

  &SEWARD
    
  &RASSCF
   Fileorb=INPORB 
   symmetry = 1
   Levshift = 1.0
   Iterations = 200 100
   Spin = 1
    nActEl = 8 0 0
   Inactive = 37 4
   Ras2 = 0 8
   Staverage = 2
   Rlxroot = 2
   
  &SLAPAF
 
>>> END DO

The optimization of the ππ* state fails with an error :

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                              &ALASKA

             launched 8 MPI processes, running in PARALLEL mode (work-sharing enabled)
                       available to each process: 8.0 GB of memory, 1 thread
                                         master pid: 144783
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

--- Stop Module: alaska at Sun Apr  5 20:29:10 2026 /rc=_RC_INVOKED_OTHER_MODULE_ ---
*** files: xmldump
    saved to directory /localscratch/ipcanu/tuca_1meth_aug_cc_pvdz_exc_opt/tuca_1meth_pi_pistar_88_cas_opt_cal/tuca_1meth_pi_pi_cas_88_aug__0.82729
--- Module alaska spent 2 seconds ---

*** symbolic link created: INPORB -> tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.RasOrb
--- Start Module: mclr at Sun Apr  5 20:29:12 2026 ---

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                               &MCLR

             launched 8 MPI processes, running in PARALLEL mode (work-sharing enabled)
                       available to each process: 8.0 GB of memory, 1 thread
                                         master pid: 145049
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

 ** Cholesky or RI/DF not implemented with symmetry **
--- Stop Module: mclr at Sun Apr  5 20:29:16 2026 /rc=_RC_INPUT_ERROR_ ---
*** files: tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.GssOrb tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.rasscf.molden.1 tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.rasscf.molden.2
           tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.rasscf.molden tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.guessorb.molden tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.RasOrb.2
           tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.RasOrb tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.RasOrb.1 tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.SpdOrb.1
           tuca_1meth_pi_pi_cas_88_opt_aug__0.82729.SpdOrb.2 xmldump
    saved to directory /localscratch/ipcanu/tuca_1meth_aug_cc_pvdz_exc_opt/tuca_1meth_pi_pistar_88_cas_opt_cal/tuca_1meth_pi_pi_cas_88_aug__0.82729
--- Module mclr spent 3 seconds ---

.####################.
.# User input error #.
.####################.

The same calculation i have done using single processor also, still failed.

However i have done the pi sigma star optimization  and it finished successfully. the input is :

& GATEWAY
 Title = TUCA 1 methanol pi sig casscf(8,8) opt nofloat -0.82729 angstrom
 RICD  
 coord = t_uca_1meth_MP2_opt_cc_pvdz.xyz
 Group = z
 Basis set = aug-cc-pvdz
 Constraints
 r1 = bond H13 N12
 r2 = bond H13 O17
Values
 1.0 r1 - 1.0 r2 = -0.82729 angstrom
End of Constraints

>>> EXPORT MOLCAS_MAXITER = 2000

>>> DO WHILE

  &SEWARD
    
  &RASSCF
   Fileorb=INPORB 
   symmetry = 2
   Levshift = 1.0
   Iterations = 200 100
   Spin = 1
    nActEl = 8 0 0
   Inactive = 37 4
   Ras2 = 1 7
   ciroot = 1 1 1
   Rlxroot = 1

   &SLAPAF
 
>>> END DO

I would be thankful if anyone can help me to solve this issue.

Thank you
Anugraha

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,204

Re: CASSCF excited-state optimization fails for ππ* state

For the sigma-pi* you use state-specific CASSCF, which works fine with symmetry. For the pi-pi* you use state-average CASSCF, which requires MCLR for analytical gradients, but this doesn't work with RICD and symmetry. You have several options:

* Remove symmetry (recommended). Use "Group = NoSym". You'll lose some performance, but it will be possible to compute all states in a single calculation (ensuring that the active space is the same, for instance).
* Enforce numerical gradients: Add "Numerical" in GATEWAY. Gradients will be (much) slower and less accurate, but they're reasonably parallelized.
* Remove RICD. Everything will be slower, require more resources, and energies are not directly comparable with calculations without RICD.

Anugraha

Member

Registered: 2025-12-17

Posts: 22

Re: CASSCF excited-state optimization fails for ππ* state

Respected Sir,
Following my previous query, I modified the input to use numerical gradients as suggested. My current input is:

& GATEWAY
 Title = TUCA 1 methanol pi pi star state specific casscf(8,8) opt nofloat -0.82729 angstrom
 RICD
 Numerical
 coord = t_uca_1meth_MP2_opt_cc_pvdz.xyz
 Group = z
 Basis set = aug-cc-pvdz
 Constraints
 r1 = bond H13 N12
 r2 = bond H13 O17
Values
 1.0 r1 - 1.0 r2 = -0.82729 angstrom
End of Constraints

>>> EXPORT MOLCAS_MAXITER = 2000

>>> DO WHILE

  &SEWARD

  &RASSCF
   Fileorb=INPORB
   symmetry = 1
   Iterations = 200 100
   Spin = 1
    nActEl = 8 0 0
   Inactive = 37 4
   Ras2 = 0 8
   Ciroot = 2 2 1
   Rlxroot = 2

  &SLAPAF

>>> END DO

However, the calculation still terminates with the following error:

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                              &ALASKA

             launched 8 MPI processes, running in PARALLEL mode (work-sharing enabled)
                       available to each process: 8.0 GB of memory, 1 thread
                                         master pid: 117222
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

--- Stop Module: alaska at Mon Apr  6 14:10:22 2026 /rc=_RC_INVOKED_OTHER_MODULE_ ---
--- Module alaska spent 2 seconds ---

*** symbolic link created: INPORB -> tuca_1meth_pi_pi_cas_num_88_opt_aug__0.82729.RasOrb
--- Start Module: numerical_gradient at Mon Apr  6 14:10:25 2026 ---

()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

                                        &NUMERICAL_GRADIENT

             launched 8 MPI processes, running in PARALLEL mode (work-sharing enabled)
                       available to each process: 8.0 GB of memory, 1 thread
                                         master pid: 117497
()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()()

 Root to use:                     2
 Number of internal degrees                               40
 Number of constraints                                     1
 Number of displacements                                  80
 Effective number of displacements                        78
 Relative displacements                   1.0000000000000000E-002

   * Point #   1 of   78 done.
    (Perturbation    3)
--- Stop Module: numerical_gradient at Mon Apr  6 14:12:13 2026 /rc=_RC_FLOATING_EXCEPTION_ ---
--- Module numerical_gradient spent 1 minute 48 seconds ---

.############################.
.# Floating point exception #.
.############################.

    Timing: Wall=182.62 User=540.07 System=39.44

Since I would like to retain RICD, is there an alternative approach that would allow a stable optimization?”

Thank you
Anugraha

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,204

Re: CASSCF excited-state optimization fails for ππ* state

Run without symmetry, or with MOLCAS_REDUCE_PRT=NO and MOLCAS_REDUCE_NG_PRT=NO to debug the failure. Maybe the active space or number of roots is not stable upon displacements...