[Columbus] Using MOLCAS rasscf orbitals as input to CI

Felix Plasser F.Plasser at lboro.ac.uk
Thu Jun 16 05:26:18 EDT 2022


Hi Zihan,

For combining OpenMolcas/RASSCF with Columbus, you can see
EXAMPLES/MOLCAS/Acetylene-RASSCF-AQCC-CASPT2

Basically, you need to add the keyword "rasscf" to control.run and add the appropriate section to molcas.input.

This will work for singlepoints but not for gradients. For gradients, you will always need to run Columbus/mcscf.x to get the required information about the wavefunction Hessian. In principle, it is possible to run OpenMOLCAS/RASSCF first to get the orbitals and then Columbus/mcscf.x to get the wavefunction Hessian. But unless you have a strong reason, I would stay with the Columbus programs to keep things simpler.

-Felix

On 15/06/2022 20:28, Zihan Pengmei via Columbus wrote:


Dear developers, With regarding to the Openmolcas/columbus link, how could I use the molcas’s rasscf MO as the input to mrci module? There are some short description on the webpage but it’s very vague. I benchmarked scf and mcscf module of
Dear developers,

With regarding to the Openmolcas/columbus link, how could I use the molcas’s rasscf MO as the input to mrci module? There are some short description on the webpage but it’s very vague.  I benchmarked scf and mcscf module of columbus and openmolcas 22.04 using 1 core on stilbene with (2,2) active space, openmolcas used 8.03 minutes on one run, while columbus using seward integral used 15 minutes.

And how would this affect the SHARC interface?

Much appreciated for looking into this!
Zihan


============== timings ===================



   unik.gets.x

 * scfpq.x          walltime    0:05:07

   mcdrt.x

   mcuft.x

 * mcscf.x          walltime    0:10:05

   mcpc.x

   mcpc.x

   mcpc.x

   mo_convert.x

   mo_convert.x

   mo_convert.x



==========================================



--- Module rasscf spent 4 minutes 35 seconds ---



.########################.

.# Non-zero return code #.

.########################.



    Timing: Wall=481.58 User=429.19 System=50.89






_______________________________________________
Columbus mailing list
Columbus at lists.osc.edu<mailto:Columbus at lists.osc.edu>
https://lists.osu.edu/mailman/listinfo/columbus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osu.edu/pipermail/columbus/attachments/20220616/d50f824e/attachment-0001.html>


More information about the Columbus mailing list