[molpro-user] Molpro configs and rigid optimisation

[Andy May]

Daniel,

I can only answer part (1). The reason it is not expanded is because the
.molprorc file is read directly in Fortran which does not know about shell
variables or how to expand them. There is already an enhancement bug about
implementing this in the future, but currently you need to write the fully
qualified directories.

You can probably achieve what you want by using the MOLPRO_OPTIONS
environment variable instead.

I completely agree about '--backup 2' in .molprorc - this does not appear
to be working correctly, I will file a bug about this. It does work
directly on the command line:

molpro --backup 2 input.mol

and also via MOLPRO_OPTIONS:

export MOLPRO_OPTIONS="--backup 2"
molpro input.mol

Best wishes,

Andy

On 17 February 2016 at 10:29, Daniel Nurkowski <dln22.como at gmail.com> wrote:

> Dear Molpro users,
>
> I would like to ask you for a small advice. I have two questions:
>
> 1) Could you please tell me what is a proper syntax of $HOME/.molprorc
> file? At the moment I can not get my system variables expanded. For
> example, if in my .molprorc file I have:
>
> --no-backup
> --no-xml-output
> -d /scratch/$USER/molpro_scratch
> -W /scratch/$USER/molpro_scratch
> -I /scratch/$USER/molpro_scratch
>
> Then $USER is not properly expanded. I have tried many different ways
> to get it working, but for some reason it failed. Apart from that, if
> I set:
>
> --backup 2
>
> to have two files backed up then molpro fails with an error that "more
> than 1 input file is not allowed" . I would be very grateful for any
> advice. (I am aware that I might be missing something simple).
>
>
> 2) My second question is about an optimisation of the geometry to a
> specifc state in a state-average CASPT2 calculation. I am trying to
> optimise lowest electronic state either keeping fragments fixed or
> relaxed at some fixed distance R5. I need to use SA-CASSCF in order to
> avoid root-flipping errors. I also can not use symmetry in this
> particular case. My example is for CAS(6e,4o)  (1 TiCl3 radical
> orbital and 3p Cl orbitals) where I compute orbitals for 3 electronic
> states.
>
> So far my attempts have failed. The optimisation goes for a few steps
> and then there is a convergence problem in "multi". I was then
> wondering if there is a way to run a job with larger CAS prior to the
> main job in the optimisation scheme to help with the convergence?
> However, when I tried that then only the last multi job is taken into
> account for the optimisation routine.
>
> Here is my input file:
>
> ***, TiCl3--Cl rigid opt
> memory,300,M
> angstrom
> nosym
> geometry={
> Ti1
> Cl1,  Ti1,  r2
> Cl2,  Ti1,  r3,  Cl1,  a3
> Cl3,  Ti1,  r4,  Cl1,  a4,  Cl2,  d4
> Cl4,  Ti1,  r5,  Cl1,  a5,  Cl2,  d5
> }
> R2= 2.2030  ANGSTROM
> R3= 2.2030  ANGSTROM
> A3= 120.00  DEGREE
> R4= 2.2030  ANGSTROM
> A4= 120.00  DEGREE
> D4= 180.03  DEGREE
> A5=  90.17  DEGREE
> D5=  91.45  DEGREE
> R5=3.0
>
> basis=cc-pVDZ
> {rhf; wf,90,1,2}
> ! Running optimisations
>         {multi; closed, 42; occ, 46; wf, 90, 1, 0; state, 3};
>         {rs2, shift=0.2; state, 3}
>         e1=energy(1)
>         if (energy(2).LT.e1) then
>           e1=energy(2)
>         endif
>         if (energy(3).LT.e1) then
>           e1 = energy(3)
>         endif
>        {optg, variable=e1, maxit=100; active, A5, D5; inactive, R5}
> ---
>
> Best regards,
> Daniel
>
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20160223/6c111315/attachment-0001.html>