From owner-chemistry -x- at -x- ccl.net Tue Mar 13 21:06:00 2012
From: "Cory Pye cpye!=!ap.smu.ca" <owner-chemistry_-_server.ccl.net>
To: CCL
Subject: CCL:G: Problem with DKH calculation
Message-Id: <-46485-120313210052-32393-F8MrUIawu1/bKScXRlizng_-_server.ccl.net>
X-Original-From: Cory Pye <cpye,+,ap.smu.ca>
Content-Type: TEXT/PLAIN; charset=US-ASCII
Date: Tue, 13 Mar 2012 22:10:00 -0300 (ADT)
MIME-Version: 1.0


Sent to CCL by: Cory Pye [cpye[*]ap.smu.ca]


On Tue, 13 Mar 2012, Ranjita Das ranjitadas.85() gmail.com wrote:

>
> Sent to CCL by: "Ranjita  Das" [ranjitadas.85.:.gmail.com]
> Hi,
>
> I am trying to do Douglas-Kroll-Hess calculation using G09. My system contains 33 atoms those are C, N, H and oly one Ti atom. I am trying to optimize the system using DKH calculation, in input I have used the command
>
> # B3LYP Gen Integral=DKH2 IOP(3/93=2)  opt freq
>
> {title}
>
>  {charge}, {multiplicity}
>
> {coordinates}
>
>
>
> {basis sets}
>
> but always i get the following error message
>
>
>  ************************************************
>  ** ERROR IN INITNF. NUMBER OF VARIABLES (  0) **
>  **   INCORRECT (SHOULD BE BETWEEN 1 AND 50)   **
>  ************************************************
>
>
Hello, this looks like the error you have when you have too many variables in
an eigenvector-following method.

My hunch is: You have specified the DKH Hamiltonian. Possibly, analytic forces
are not available for this method, so the system defaults to an energy only
optimization using the eigenvector following method. You have 33 atoms, and
therefore 93 degrees of freedom in the absence of symmetry. The limit for the
EF method is 50 variables, hence the error.

Workarounds:
(1) preoptimize without using DHK, do single points with. Another option might
be to change your IOP to 3/93=1 to use point charges instead of s-Gaussians for
nuclei, it is possible that analytic gradients might be available for that
nuclear model.
(2) If your system has symmetry, take advantage of it with a well constructed
z-matrix and use opt=z-matrix as well.
(3) change the default optimization method to something other than EF. The
problem is that the deprecated Fletcher-Powell and Murtagh-Sargent methods
also have this limitation.
(4) consider the use of constrained geometries ie setting all C-H distances to
be equal, optimize the rest. This can be done by partitioning your variables
into 2 sets so that neither set is larger than 50 variables, and keep changing
which set you optimize, until both sets of variables are converged.



   *************    !  Dr. Cory C. Pye
 *****************  !  Associate Professor
***   **    **  **  !  Theoretical and Computational Chemistry
**   *  ****        !  Department of Chemistry, Saint Mary's University
**      *  *        !  923 Robie Street, Halifax, NS B3H 3C3
**      *  *        !  cpye^^^crux.stmarys.ca   http://apwww.stmarys.ca/~cpye
***     *  *    **  !  Ph: (902)-420-5654  FAX:(902)-496-8104
 *****************  !
   *************    !  Les Hartree-Focks (Apologies to Montreal Canadien Fans)