Molcas Forum

delta

Member

Registered: 2019-05-21

Posts: 1

Orbitals during a PES-Scan

Dear Molcas-Users

I have recently  begun performing computations with Molcas. I am currently scanning the potential energy surface (PES) of a C_{2V}-symmetrical system on CASSCF/cc-PVTZ level of theory in Molcas 8.0.
In this scan I am slowly tearing the molecule apart while adjusting a bond-length through linear interpolation. Chemically, this corresponds to a dissociation process.

While this method of varying the coordinates of the system works as intended, I am observing no notable change in the occupation numbers of orbitals in the active space during the dissociation and a visualisation of the orbitals (*.grid obtained through &GRID_IT for each step) shows that all orbital coefficients are kept constant - it appears that an optimisation of the orbitals is not really performed even though the CASSCF-calculation at every step converges.

This is the code I used for the scan:

>>foreach L in ( 0 .. $stepnum )
         >>eval Z_1 = $z_1_0+(0.1*$L)
         >>eval Y_2 = ( ($y_2_fin - $y_2_sta ) / $stepnum ) * $L+$y_2_sta

          &GATEWAY
          coord
              lso.xyz

          basis = cc-PVTZ
          group = xy y

          &RASSCF
                  nActEl = 9 0 0
                  Spin = 2
                  Symmetry = 3
                  Inactive = 3 0 0 2
                  Ras2 = 4 2 1 3
                  ciroot = 5 5 1
                  rlxroot = 1
                  LumOrb

           &GRID_IT
              >> COPY $CurrDir/$Project.grid $CurrDir/$Project.step_$L.grid
              >> COPY $CurrDir/$Project.rasscf.molden $CurrDir/$Project.step_$L.rasscf.molden

>>enddo

The orbital-coefficients are imported via LumOrb from a previously optimised structure (the starting-point of the PES-scan) in the *.RasOrb file-format. This exact set of coefficients and occupation is the one utilised in every step. Even when the "LumOrb"-keyword for reading from initial input is omitted and an SCF-computation is performed initially before the scan, coefficients and occupation of the SCF-output are found again in every successive step.

Can you help me with this problem? How can I obtain properly optimised and occupied orbitals adjusted to the new molecular structures in a PES-scan?

With thanks and regards,
D.

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,202

Re: Orbitals during a PES-Scan

The orbitals should be optimized. Are the files for different steps identical? If not, they have been optimized. Note that the $Project.RasOrb and $Project.rasscf.molden files contain the state-average orbitals, so I wouldn't expect large occupation changes if the states included in the average remain more or less the same. Have you checked the state-specific natural orbitals ($Project.RasOrb.*, $Project.molden.*)?