CCL:G: Relaxed scan using GDIIS with Gaussian

 Sent to CCL by: "Gianluca  Santarossa"
 [gianluca.santarossa^chem.ethz.ch]
 Dear CCL subscribers,
 I am trying to run a relaxed scan simulation of a molecule at PM3 level using
 Gaussian03.
 In order to make the geometry converge during the optimization I need to use the
 GDIIS algorithm.
 When the GDIIS option is activated, the program is only performing the first
 geometry optimization,
 and then it exits instead of moving the dihedral.
 This is the relevant part of the input:
 $RunGauss
 #P PM3 Opt(Z-matrix,tight,GDIIS) gfinput gfprint IOP(6/7=3) pop=full
 Theta scan of KDO disaccharide
 !
 [cut] ... here comes the z-matrix ...
 dih9         0.000 S 35 10.0
 [cut] ... here are more z-matrix parameters ...
 When only the parameter Opt=Z-matrix is used, the scan is performed correctly.
 The 'Initial Parameters'  section of the output shows that the 'Scan' is on:
 [cut]
  !      cco9       109.737    estimate D2E/DX2                          !
  !      dih9         0.0      Scan                                      !
  !      cc10         1.5379   estimate D2E/DX2                          !
 [cut]
 However, adding the GDIIS options to the input - Opt(Z-matrix,GDIIS) - results
 in a geometry
 optimization that stops without the scan.
 Apparently, the program redefines the Z-matrix, as it is shown in the output,
 and could not scan
 anything:
 [cut]
                            ----------------------------
                            !    Initial Parameters    !
                            ! (Angstroms and Degrees)  !
  --------------------------
 --------------------------
  ! Name  Definition              Value          Derivative Info.
 !
  ---------------------------------------------------------------------
  ! R1    R(1,2)                  1.4317         estimate D2E/DX2
 !
  ! R2    R(1,6)                  1.4288         estimate D2E/DX2
 !
 [cut]
 May you please suggest me how to run a relaxed scan using the GDIIS optimizer?
 Regards,
 Gianluca