GVB calculation in Gaussian'98

["Demetrio A. da Silva Filho" <dasf %! at !% ifi.unicamp.br>]

 Dear CCL'ers,

 I am trying to perform a GVB calculation in Gaussian'98. The basic idea
 is to create
 just one GVB pair taking one electron from HOMO-1 one from LUMO. All the
 orbitals below the pair
 will be considered as closed shells and will not enter in the
 calculation. The system will be forced to be singlet.

 I was able to run that calculation in GAMESS, but the results are not
 consistent, so
 I want to do the same calculation in Gaussian'98. But I could not figure
 out which keywords I must use in order
 to do  the same thing using Gaussian'98.

 Using Gamess, I just need to run a Hartree-Fock calculation to have a
 guess for the orbitals
 and then I use the keywords

 $SCF NCO=88 NSETO=2 NO(1)=1,1 $END
 

 To create the GVB pair and, freezing all orbitals below the HOMO
 and  the keywords:

 $GUESS  GUESS=moread norb=90
 NORDER=1
 $END                                    
 
 $guess iorder(88)=89 iorder(89)=88
 $end                                       
 

 To change the HOMO-1 orbital (N=88) into HOMO (N=89).

 Then, at the end, I will have just one GVB pair composed by an
 electron
 > from HOMO-1 and one from LUMO.

 Does anybody knows how can I do the same thing using
 Gaussian'98?

 Thanks in advance and have a nice day!

 Demetrio Filho