Skip to content

Starting Norm too high for relativistic CASSCF calc

Bug Report

Starting Norm too high for relativistic CASSCF calc using orca orbs and 1 electron relativistic Hamiltonian

Description

I am trying to do a CASSCF job where I read from a JSON file from an ORCA relativistic calculation. In the JSON file, I am also reading in the relativistic 1-electron Hamiltonian.

I am having a problem where the starting norms of the CASSCF calculation are high, as shown below:

$ grep "Norm(g)" qdnev1_5_5roots_new_a.out
         Norm(g)  =     532.63981028
         Norm(g)  =    2544.74536696
         Norm(g)  =    2367.64711939
         Norm(g)  =    1451.43139866
         Norm(g)  =    1318.95646295
         Norm(g)  =     759.17125030
         Norm(g)  =     674.23883339
         Norm(g)  =     576.65049496
         Norm(g)  =     455.45845418
         Norm(g)  =     350.35674067
         Norm(g)  =     278.20383125
         Norm(g)  =     222.16619070
         Norm(g)  =     171.32173850
         Norm(g)  =     124.32766901
         Norm(g)  =      81.65207875
         Norm(g)  =      40.00868134

I am providing all the files necessary in this zip folder TiF3_rel_calcs.zip

Steps to Reproduce

  1. Initial orbitals were generated using a relativistic ORCA calculation (init_orbs_rel_a.inp).
  2. QRO orbitals from this were used to obtain the JSON file.
  3. The JSON file was used in the CASSCF calculations (qdnev1_5_5roots_new_a.inp and qdnev7_8_5roots_new_a.inp).

Additional Information

  • ORCA can generate both the 1-electron Hamiltonian (H-Matrix) and the 1-electron relativistic Hamiltonian (H-MatrixRel) in its JSON file.

  • In HUMMR, we can only read the keyword H-Matrix.

  • So I generate the H-MatrixRel in ORCA calc, but manually write it out to H-Matrix so that it can be read in HUMMR.

Impressum | Datenschutzerklärung