[molpro-user] Excessive gradient in CI

Christian Bürgel christian.buergel at chemie.hu-berlin.de
Fri Sep 7 12:51:30 BST 2007


Hello,

I'm running a CASSCF calculation on linear SAuS- including the triplet ground 
state and the first delta and sigma singlet states.

Input:
***
memory,200,m
gprint,basis,orbital=18,civector,ref,pspace !,micro
geomtyp=zmat
geometry={
angstrom
au
s,1,raus
s,1,raus,2,180.0
}
raus=2.20212785 ang      ! ccsd(t) optimized value

basis,au=avtz-pp,s=avtz

{hf;wf,52,4,2;occ,8,3,3,1,6,3,3,0;open,3.6,3.7}

{casscf;tran2,lzz
wf,52,4,2;lquant,0;            ! 3 sigma - (GS)
wf,52,4,0;lquant,2;            ! 1 delta g
wf,52,1,0;state,2;lquant,2,0;  ! 1 delta g ; 1 sigma +
}

---

The MCSCF is successfully done; the NO as well as the P-space configurations and 
the CI vectors for the first two states are written.

Then the following error is given:


  CI vector for state symmetry 2
  ==============================

  22220 2 2 2 22 2b 2a      -0.6100920
  22220 2 2 2 22 2a 2b       0.6100920
  22220 b a 2 22 22 22       0.3371446
  22220 a b 2 22 22 22      -0.3371446

  TOTAL ENERGIES                      -929.97858884

  P-Space Configurations for state symmetry 3 selected with threshold      0.40

   Nr    Energy    Occupancy

    1   -89.720    22220222222022
    2   -89.720    22220222222220
    3   -89.597    22220022222222
  ....


*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#
                                                   ERROR DETECTED
                                                   **************

      EXCESSIVE GRADIENT IN CI
*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#
  ERROR EXIT
  CURRENT STACK:      MAIN
**********************************************************************************************************************************

When I restart the job it runs smoothly.

Is this behaviour known?

Thanks for the help.

*************************************
Christian Buergel
Institut für Chemie
Humboldt Universität zu Berlin
Brook-Taylor-Str. 2
12489 Berlin
Germany
+49 (30) 2093 5575
christian.buergel at xxxxxxxxxxxxxxxxxxx
*************************************





More information about the Molpro-user mailing list