[molpro-user] CASSCF geometry problem in optimization

[Nicholas Marshall]

Dear molpro users,

I am experiencing a strange error when I attempt to run a CASSCF 
optimization after an SCF optimization, in preparation for taking an 
MRCI single point. As the CASSCF optimization proceeds, the section of 
the output file where it should show the geometric parameters for each 
optimization step, as shown, (the bond angle is constrained)

-----------------------------------------------
  ATOMIC COORDINATES

  NR  ATOM    CHARGE       X              Y              Z

    1  NA     11.00   -2.965834605    0.164723731    0.000000000
    2  O       8.00    3.479009763    0.353425528    0.000000000
    3  N       7.00    0.894008151   -0.674075600    0.000000000

  Bond lengths in Bohr (Angstrom)

   1--3  3.949932964   2--3  2.781724628
        (2.090214659)       (1.472025386)

  Bond angles

   1--3--2  146.06232252

  NUCLEAR CHARGE:                   26
-----------------------------------------------

suddenly only shows one bond length, and none of the other parameters.

-----------------------------------------------

  ATOMIC COORDINATES

  NR  ATOM    CHARGE       X              Y              Z

    1  NA     11.00   -3.751982236    0.000476243    0.000000000
    2  O       8.00    2.797888129   -0.027934324    0.000000000
    3  N       7.00    2.962373635    0.031126793    0.000000000

  CAUTION
  VERY SHORT BOND LENGTHS PRESENT - IS THIS RIGHT?


  Bond lengths in Bohr (Angstrom)

   2--3  0.174767552
        (0.092483012)

  NUCLEAR CHARGE:                   26
-----------------------------------------------

Sometimes after optimizing the single bond length for a few iterations, 
and sometimes immediately, the displayed bond length suddenly jumps 
TREMENDOUSLY to a nonphysically short value, causing the optimization 
to fail when the energy is computed in the next iteration.

In some cases (different geometries) I get this error instead, after 
the geometry jump occurs:
-----------------------------------------------

  ?? Geometry inconsistency ??
  R= -1.765033122477053,  1.7650331224770526

  v10
    1-  3    1.641835    0.647858    0.000000
  ? Error
  ? Possible programming error in geometry update
  ? The problem occurs in zmatrix.f:zmat_gradz
-----------------------------------------------

Can anyone suggest what might be happening here?

Thanks very much,

Nicholas Marshall
-----------------------------------------------

INPUT FILE:

***,NaON MRCI excited state scan VERT @ CASSCF geom
memory,250,m

file,2,naon20.wfn,new

gthresh,energy=1.d-10,optgrad=1.d-6,optstep=1.d-5

set,cc_norm_max=1.0
nao=5.972250817
on=2.267840905
naon = 20

  geometry={;z;
            na
            o na nao
            x 2 1.0 1 90.0
            n 2  on 3 90.0 1 naon};

basis
   default=cc-pVTZ
end

  config
  rhf;wf,26,1,2
  optg;inactive,naon
  multi
      state,2
      noextra
      print,civector
   iterations;uncouple,1,to,24;end;
   maxiter,40;
   optg,numerical;inactive,naon
  mrci
   state,2