[molpro-user] Trouble with N3 calculation

[Bastiaan J. Braams]

Dear Molpro User,

I would appreciate advice about a calculation for the N3 complex.
Here is my simplest Molpro 2002.6 input (small basis, low-accuracy
method):

  ***,N3 TEST
  memory,84,m
  punch,mpsp.pun,new
  geomtyp=xyz
  geometry
  nosym
   3
    null
   N     -0.38895870  -0.40066306  -1.64729519
   N      0.51391218  -0.14686307   1.52149866
   N      0.06191495   0.47094736  -1.09433651
  end
  basis=vdz
  rhf;wf,charge=0,spin=1
  mp2
  ---

The particular geometry doesn't matter; for any geometry the result is
an error message at the MP2 command:

  NUMBER OF OCCUPIED ORBITALS NOT CONSISTENT WITH NUMBER OF ELECTRONS
  FOR CCSD. NELEC=  21  NCORE=  3  NCLOS=  8  NOCC= 11

  ERROR EXIT
  CURRENT STACK:      CIINP  MAIN

If rccsd or rccsd(t) is used instead of mp2 then an answer is
obtained, and I appreciate that.  But what is it with my use of mp2
that causes the error message?

I have tried to let Molpro make use of the spatial symmetry, as
follows:

  ***,N3 TEST
  memory,84,m
  punch,mpsp.pun,new
  geomtyp=xyz
  geometry
  x,y;mass
   3
    null
   N     -0.38895870  -0.40066306  -1.64729519
   N      0.51391218  -0.14686307   1.52149866
   N      0.06191495   0.47094736  -1.09433651
  end
  basis=vdz
  rhf;wf,charge=0,sym=1,spin=1
  mp2
  ---

or likewise with sym=2 on the rhf card.  In either case (sym=1 or
sym=2) there is an error already at the RHF command:

  The request for symmetry elements X,Y, cannot be honoured
 
  Prototype coordinates
                    1             2             3
          1     2.4714893    -3.7447190     1.2732297
          2    -0.7858047    -0.1876460     0.9734508
          3     0.0000000     0.0000000     0.0000000
  ? Error
  ? Symmetry problem
  ? The problem occurs in zmatrix.f:zmat_orient
 
  ERROR EXIT
  CURRENT STACK:      MAIN

Please advise!  Is it possible to use the spatial symmetry and still
get an RHF result?  Is it possible to obtain an mp2, and then mp4,
result for the N3 system?

I might explain my interest in the mp2 (or really, mp4) calculations
for this system.  I have in mind to obtain a potential energy surface
for the reaction N2 + N2, up to energies where one of the N2 may break
up or where atoms may exchange.  For the N4 complex calculations using
mp4 on an aug-cc-pvtz basis are affordable, but rccsd(t) on that same
basis is much more expensive.  So I want to do at first mp4 on the
aug-cc-pvtz basis for the N4 complex, and in that connection I want to
do also the N3 subsystem using the same method on the same basis.
This failed, and the example posted above shows that it already fails
at the mp2 level for the smaller vdz basis.

Bas Braams
--
Bastiaan J. Braams
braams at mathcs.emory.edu
Emory University, Atlanta, GA