Molcas Forum

emar

Member

Registered: 2025-01-28

Posts: 2

Error running McKinley in SA-RAS

Dear all,

I would like to kindly ask you some help with a problem I'm having while running McKinley within a SA-RASSCF calculation with OpenMolcas 24.10. My goal is to generate a *.freq.molden file relative to an excited state (here root 2), which I can then feed to dynamixtools.py to generate a set of initial conditions for the dynamics. Please, find my input below:

&GATEWAY
coord
3
Angstrom
N       0.0000000000    -0.0000000000     0.0752739057
N       0.0000000000    -0.0000000000     1.1937134333
O       0.0000000000     0.0000000000    -1.1097925662
basis = sto-3g
group = nosym

&SEWARD

&SCF

&RASSCF
  symmetry = 1
  inactive = 3
  ras1 = 0
  ras2 = 10
  nactel = 15 0 0
  spin = 2
  ciroot = 3 3 1
  rlxroot = 2

&MCKINLEY
Perturbation
Hessian

The calculation runs smoothly until the SLAPAF program. Here, the calculation neglects the translation and rotations, while calculating the (4) vibrational normal modes of N2O. The 'Harmonic frequencies', 'Thermochemistry' and 'Isotopic shifts' sections are printed out successfully. When the calculation reaches the 'Principal components of the normal modes' section, after printing the data relative to the vibrational modes, the calculation terminates with the following error message:

 ###############################################################################
 ###############################################################################
 ###                                                                         ###
 ###                                                                         ###
 ###    Location: get_dArray                                                 ###
 ###                                                                         ###
 ###                                                                         ###
 ###    Data of wrong length: K                                              ###
 ###                                                                         ###
 ###                                                                         ###
 ###############################################################################
 ###############################################################################

A *freq.molden file is produced, which contains only data relative to the 4 vibrational modes. This works with dynamixtools.py only when -m=1, the script throwing an error when -m = 2,3. Since I would be interested to run dynamixtools.py with -m=2, I would be very grateful if somebody could help me solving this problem.

Thank you very much for your help!

Emanuele

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Error running McKinley in SA-RAS

The error is a bug that will be fixed soon. However, the freq.molden file is created correctly, and that's the only thing needed for dynamixtools.py. I tried it with -m 1 and -m 2 and both gave a result with no error.

emar

Member

Registered: 2025-01-28

Posts: 2

Re: Error running McKinley in SA-RAS

Dear Ignacio,

Thanks a lot for your reply! I run the calculation again and indeed it works great, just like you described. On the other hand, I tried to perform a calculation changing rlxroot=2 to rlxroot=4. In this case, I obtain an imaginary frequency for the first vibrational mode. When I try to run dynamixtools.py, the script runs well with -m 1, but fails with -m 2,3 in correspondence to the 'inertia' function. Here, the error is related to the 'data1' matrix, which results singular and cannot be inverted (in the last line before the return). Does this mean I can run McKinley only at configurations corresponding to the energy minimum? That is, should I run a geometry optimisation before using McKinley?

Thank you very much for your time and your work.

Emanuele

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,085

Re: Error running McKinley in SA-RAS

Vibrational analysis is only meaningful at a stationary point (a minimum or a saddle point). You can run McKinley at any point, but the "frequencies" are not really frequencies. Another thing is the sampling done by dynamixtools, which probably requires a minimum (no imaginary/negative frequencies).