NEVPT2: Inconsistent reference energies for state 1

[Sebastian Reiter]

Dear Molpro users,

I would like to perform a NEVPT2 calculation on a transition metal complex with ~100 atoms.

At first, I ran into the error described here, so I tried reading the CI vectors from the CASSCF calculation as suggested:

https://groups.google.com/g/molpro-user/c/NYhqEDzdcxY/m/3pzdSsOxCAAJ

Now the calculation crashes with the following error message:

 Inconsistent reference energies for state 1
 Reference energy from start record:     -3776.933910150434
 Energy from imported reference vector:  -3776.933910144378
 Please check if orbitals are consistent!
 ? Error
 ? Inconsistent reference energies
 ? The problem occurs in cihdia_s

To me, this looks like a numerical issue that could perhaps be resolved by adjusting a threshold (?). The input certainly works for smaller molecules.

Here is the relevant part of the input, I am using Molpro 2022.3:

=============================================================
 gthresh,energy=1.d-9
 gthresh,orbital=1.d-8
 gthresh,civec=1.d-8

 ...

 {df-multi,first-order;
  config,csf;
  start,2142.2;
  ciguess,2141.2;
  closed,205.1;
  occ,214.1;
  wf,spin=2,symmetry=1,charge=1;
  state,7.1;
  natorb,2143.2,state=1.1,cirec=5143.2;
 }

 {df-nevpt2,cirec=5143.2,thrden=1.0d-8,thrvar=1.0d-8;
  orbit,2143.2,state=1.1;
  closed,205.1;
  occ,214.1;
  wf,spin=2,symmetry=1,charge=1;
  state,1.1;
 }

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

If anyone could assist me with this issue, I would be very grateful.

Kind regards
Sebastian

[Hans-Joachim Werner]

Dear Sebastian,
this is in fact due to too tight a threshold. The problem has been fixed in November 2022, and should be in the latest Molpro version.
We are surprised that you still see it with 2022.3. Could you please send the SHA printed in the output (in the header part after the input).
Best wishes
Joachim Werner

> --
> You received this message because you are subscribed to the Google Groups "molpro-user" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to molpro-user...@googlegroups.com.
> To view this discussion on the web, visit https://groups.google.com/d/msgid/molpro-user/2a3ccd62-1480-1788-ef40-e1cf7caeb506%40campus.lmu.de.

[Sebastian Reiter]

Dear Professor Werner,

thank you very much for your help.

I just realized that the Molpro version that gave this error was in fact 2022.2. Apologies for the confusion.

With 2022.3 (SHA1: e31ec9a5ea85254ab76f59d122cbdd51c71cf98b), I run into a different error message, which has already been discussed on the forum:

=============================================================
 Maximum of off-diagonal core    Fock matrix is:     0.2248993224D-08
 Maximum of off-diagonal virtual Fock matrix is:     0.5452586085D-05
 ? Error
 ? Virtual block of Fock matrix not diagonal. Please use state specifc orbitals
 ? The problem occurs in Readfm
=============================================================

Unfortunately, it did not help to read the CI vectors in the NEVPT2 block as suggested in

https://groups.google.com/g/molpro-user/c/NYhqEDzdcxY/m/3pzdSsOxCAAJ 

The relevant input here was:

=============================================================
 gthresh,energy=1.d-8
 gthresh,orbital=1.d-10
 gthresh,civec=1.d-10

 ...

 {df-multi,first-order;
  config,csf;
  start,2142.2;
  ciguess,2141.2;
  closed,205.1;
  occ,214.1;
  wf,spin=2,symmetry=1,charge=1;
  state,7.1;
  natorb,2143.2,state=1.1,cirec=5143.2;
 }

 {df-nevpt2,cirec=5143.2,thrden=1.0d-8,thrvar=1.0d-8;
  orbit,2143.2,state=1.1;
  closed,205.1;
  occ,214.1;
  wf,spin=2,symmetry=1,charge=1;
  state,1.1;
 }
=============================================================

Is there perhaps another way to deal with this problem?

Best regards
Sebastian

[Sebastian Reiter]

Dear Prof. Werner, Dear Molpro users,

I just wanted to follow up on this topic: Using the latest Molpro 2023.2
with the input posted previously has resolved the issue for me.

Thank you very much to the developers for fixing it.

Best wishes
Sebastian



Am 02.05.23 um 12:43 schrieb Hans-Joachim Werner:
> Dear Sebastian,
> Possibly this comes from the restrict options. Probably, nevpt2 assumes CAS references (but I am not sure). I will try to find out. If this is the case, it should of course be trapped by an error message. The problem is that this is a program provided long time ago by Renzo Cimiraglia (who is retired now), and nobody in the Molpro crew knows it in detail.
> Best wishes
> Joachim Werner

>> To view this discussion on the web, visit https://groups.google.com/d/msgid/molpro-user/76390c43-5494-294f-678d-c58e1b8e7714%40campus.lmu.de.
>

--
Sebastian Reiter
Ludwig-Maximilians-Universität München
Department Chemie, AK de Vivie-Riedle
Butenandtstr. 11, Haus E, Raum E0.019
D-81377 München

Phone: +49 89 2180 77619
Email: Sebastia...@cup.uni-muenchen.de