Problem with KRCI-LUCIAREL

[jvi...@gmail.com]

Dear DIRAC Experts,

I faced this error

"*** ERROR in IREAKRMC ***

Label <MJVEC   > does not exists on unit  59"

in KRCI-LUCIAREL for UN+

I have used DFPCMO file from earlier SCF computation with the same NUp.xyz file (so, program shall recognize maximum symmetry).

The crashed KRCI output is attached as inp_NUp.out.

The working SCF output as SCF__inp_NUp.out (I apologize for all the long printouts there).

What is wrong and what shall be the workaround, please?

With best wishes for 2023,

Jakub 

[andreas...@gmail.com]

Dear Jakub,

In SCF__inp_NUp.out (line 273), you specify using aug-cc-pVTZ (triple zeta) while in inp_NUp.out (line 280) it is aug-cc-pVDZ (double zeta).

Does this solve your problem?

Best regards

Andreas

[Jakub Visnak]

Dear Andreas,

Thank you very much. That was a stupid mistake, and I apologise I did not double-check stuff like that better myself.

However, now I did run both SCF and KRCI in one job, and the error is still there... (please see attached)

And the same happened when I changed double to triple zeta in the KRCI run and performed them separately. 

Best regards,

Jakub

út 17. 1. 2023 v 18:40 odesílatel andreas...@gmail.com <andreas...@gmail.com> napsal:

--
You received this message because you are subscribed to a topic in the Google Groups "dirac-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/dirac-users/2wwagaFclfw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to dirac-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/6d87b838-f57b-4fad-b590-faeeea1b711en%40googlegroups.com.

[andreas...@gmail.com]

Dear Jakub,

You're welcome!

I see that you are using the Dirac 2019 version. Can you check if the error still occurs with a more recent version?

Best regards

Andreas

[andreas...@gmail.com]

Dear Jakub,

I just tried to run your calculation with your input, but a smaller basis set.

With DIRAC 22.0 it seems to be working correctly. I have attached the output so that you may also check if it is the desired output.

If it still doesn't work, you might also try using the following keywords, although I think they might not be relevant for your issue:

Under *KRCI, insert the following lines:

.MK2REF

0

.MK2DELTA

12

Best regards

Andreas

[Jakub Visnak]

Dear Andreas,

Thank you very much. I have yet only DIRAC21 at hand. But I plan to compile DIRAC22 as well soon.

As for DIRAC21, I got a bit further into KRCI, but still with error(as attached):

"   - Beginning task  44 of  44 after   2668. seconds and   1689. CPU-seconds

 - Starting symmetrization after        2699.08 seconds
 - Finished symmetrization after        2699.08 seconds


------ Timing report (in CPU seconds) of module  integral transformation

 Time in  Initializing MS4IND file            0.006 seconds
 Time in  Computing+transform. integral    1719.378 seconds
 Time in  Symmetrizing MO integrals           0.000 seconds

------ End of timing report ------


 (RTRACTL1) Total CPU (wall) time used :   00:28:39 (   00:44:59)
            Transformation ended Tue Feb 14 06:28:48 2023
  //  LUCIAREL called
  //  CIRUN = KR-CI
  //  Running large-scale CI calculation

* REAFCK: Fock matrix read from file /scratch.ssd/visnak/job_14432265.meta-pbs.metacentrum.cz/scr
* Heading : NU+ molecule, CASCI(4,5)                         Tue Feb 14 01:15:53 2023
SCR        scr.thr.    Step1    Step2  Coulomb  Exchange    CPU-time
SOfock:LL  1.00D-12    0.00%   32.46%    0.00%    0.00%  37.75976562s
SOfock:SL  1.00D-12    0.83%   39.93%    0.00%    0.08%   5min23.879s
 >>> CPU  time used in SO Fock is  6 minutes  2 seconds
 >>> WALL time used in SO Fock is  6 minutes  2 seconds
 *** error in GROUP_CONSTRUCTOR_REL*** impossible group size:
           -4294967295

  Master node : --- SEVERE ERROR, PROGRAM WILL BE ABORTED ---

 Date and time (Linux) : Tue Feb 14 06:34:52 2023
 *** error in GROUP_CONSTRUCTOR_REL*** impossible group size <= 0.

 Traceback from this point (group_constructor_rel_+0x4a3): 10 frames.

  9   function name:     group_constructor_rel_+0x4a3
  8   function name:                krmc_luci2_+0x6bf
  7   function name:                 luci_krci_+0x914
  6   function name:                 luci_rcist_+0x5a
  5   function name:                     rcist_+0x4c9
  4   function name:                   roptst_+0x2038
  3   function name:   Unknown
  2   function name:   Unknown
  1   function name:                     MAIN__+0x788
  0   function name:                        main+0x32

DIRAC pam run in /scratch.ssd/visnak/job_14432265.meta-pbs.metacentrum.cz

 ====  below this line is the stderr stream  ====

  Master node : --- SEVERE ERROR, PROGRAM WILL BE ABORTED ---
 *** error in GROUP_CONSTRUCTOR_REL*** impossible group size <= 0.
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image              PC                Routine            Line        Source
libifcoremt.so.5   0000149737C49522  for__signal_handl     Unknown  Unknown
libpthread-2.31.s  00001497384A9140  Unknown               Unknown  Unknown
dirac.x            000000000466D1E1  Unknown               Unknown  Unknown
dirac.x            000000000436829A  quit_                     388  gptrygve.F
dirac.x            000000000425DC53  group_constructor        1689  gplupar.F
dirac.x            000000000201E6FF  krmc_luci2_               483  krmcluci.F
dirac.x            000000000201A424  luci_krci_                238  krmcluci.F
dirac.x            0000000002018EEA  luci_rcist_               902  krmcluci.F
dirac.x            0000000001EF1C99  rcist_                   3677  krmcopt.F
dirac.x            0000000001EF0138  roptst_                  3495  krmcopt.F
dirac.x            000000000204E692  krci_calc_.A             1708  krci_ctl.F
dirac.x            0000000000569B4B  dirac_.A                  497  dirac.F
dirac.x            00000000005611E8  MAIN__                    174  main.F90
dirac.x            0000000000560A42  Unknown               Unknown  Unknown
libc-2.31.so       00001497382F6D0A  __libc_start_main     Unknown  Unknown
dirac.x            000000000056094A  Unknown               Unknown  Unknown
--------------------------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
--------------------------------------------------------------------------
--------------------------------------------------------------------------
mpirun detected that one or more processes exited with non-zero status, thus causing
the job to be terminated. The first process to do so was:

  Process name: [[63008,1],0]
  Exit code:    174
--------------------------------------------------------------------------

"

The same error appears even if I add the options you have mentioned into inp.inp.

Best regards,

Jakub

po 13. 2. 2023 v 19:28 odesílatel andreas...@gmail.com <andreas...@gmail.com> napsal:

To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/e8984179-9414-4c67-a391-5730e7d5ac12n%40googlegroups.com.

[jvi...@gmail.com]

Dear Andreas,

does the 

.MK2REF

0

have correlation that my original question was on Omega=0+ state, please? 

Would it be 4 or 8 for Omega=4 state? 

As I cannot find this option in the documentation. The second one is about maximum difference between A and B bispinors allowed in KRCI determinants?

Sadly these options didn't help me. Now I am trying to "resolve" an open-shell DFT calculation for the Omega=4 state (in DZ basis for NU+ does not seem to be ground, but experiments suggest the molecule is Omega=4 in the ground state). And I got the same error when I tried to start small-scale KRCI from PBE orbitals for that state (linear symmetry).

What is the exact difference between .FOCC and .AOC, please?

I would suppose that in AOC, the minimised energy is simply an average over all configurations specified by .OPEN SHELL - thus, the state is described by the density matrix.

In FOCC there is some linear combination of Slater determinants constructed? But what is the observable which that linear combination is eigenfunction? For linear molecules, it shall be J_z, but in general, C1 geometry?

Best regards,

Jakub

Dne úterý 14. února 2023 v 17:04:23 UTC+3 uživatel jvi...@gmail.com napsal:

[andreas...@gmail.com]

Dear Jakub,

Have you tried out the Dirac 23 version as Hans Jørgen suggested yet?

Yes, the MK keywords refer to a way to clasify the configuration state functions (see e.g. http://dx.doi.org/10.1063/1.471644).

Using these keywords sometimes helped with solving some issues in previous versions of DIRAC, but they may be redundant now.

Yes, for AOC you minimise the average energy over the open shell configurations.

I must admit that I do not know the specifics about all the input options, including FOCC.

Best regards

Andreas

[andreas...@gmail.com]

Dear Jakub,

I have now tested with Dirac 23 too, and it works without any flaws.

DO note: I have created a new Davidson CI-solver, leading to much better convergence.

To use this new Davidson solver, you will have to compile DIRAC with mpi.

I used the following setup:

Python 3.6.2

Intel 2020.4

OpenMPI 4.0.3

Just for good measure, I have attached my output with U dyall.v3z and N aug-cc-pVTZ.

Best regards,

Andreas

[Jakub Visnak]

Hello,

Thank you very much!

I tried DIRAC 23, and the run, where SCF and KRCI are together, works fine now.

And it does seem (it is still running, now the AO-to-MO integral transfomation) that even for AOC (open-shell) SCF orbitals.

But, I need to have also "two-step" procedure working.

Step: 1: SCF (or PBE/KRMCSCF/MP2NO/...) orbital generation, into DFPCMO file

Step: 2: KRCI being run from DFPCMO, without the previous SCF step.

And that still ends with the same error...

Please find inp_NUp.out_SCF ....  SCF output  and  inp_NUp.out KRCI output attached.

Best regards and happy Easter holidays,

Jakub

st 1. 3. 2023 v 16:21 odesílatel andreas...@gmail.com <andreas...@gmail.com> napsal:

To view this discussion on the web visit https://groups.google.com/d/msgid/dirac-users/892491d8-406f-498a-8d6b-1dc21f8892ccn%40googlegroups.com.

[andreas...@gmail.com]

Hello again,

It is a very strange error which I unfortunately don't know much about. A few suggestions:

1) Does it fail when you get/put the DFCOEF instead of the DFPCMO file?

2) Could you try, in your step 2, to initially run an SCF with the converged orbitals from your prior SCF calculation?

Maybe 2) will work and you hopefully just need to run one SCF iteration with the already converged orbitals.

3) This is probably not the issue, but worth a shot: try to remove lines in your KRCI-only input that are not needed.

Best regards

Andreas

[andreas...@gmail.com]

Dear Jakub,

I don't know if you have noticed it but there is another thread on Dirac 23.0 and DFCOEF.

Maybe the suggestions in that thread will fix your problem.