[molpro-user] Excessive gradient in CI error

Hans-Joachim Werner werner at theochem.uni-stuttgart.de
Thu Jan 21 18:34:11 GMT 2010

Most likely this error is due to "root flipping", i.e. the state or  
even state symmetry changes during the optimization. Possible remedies  

(i) Try a larger p-space threshold, e.g. pspace,1 (default is 0.4).  
This will
include more CSFs or determinants into the P-space Hamiltonian which is
explicitly constructed and diagonalized.

(ii) If this does not help include more states in the state-averaged  
calculation. Such problems most likely occur near avoided crossings or  
conical intersections or in other near degeneracy situations.

(iii) Check if your active space is appropriate and that you get a  
variation of all orbitals along the PES you are computing.

Best regards
Joachim Werner

Am 21.01.2010 um 16:52 schrieb arimet at eie.gr:

> Many users of MOLPRO have the same problem.  But no one seems to  
> know how
> to explain it and fix it.  I wish some one will tell us what to do  
> about
> this error.
> A. Metropoulos
> Original Message:
> -----------------
> From: Ulises Miranda ulimirandone at gmail.com
> Date: Wed, 20 Jan 2010 22:23:22 -0600
> To: molpro-user at molpro.net
> Subject: [molpro-user] Excessive gradient in CI error
> Hello, dear Molpro users.
> I start a MCSCF/MRCI calculation for a potential energy curve but  
> when it
> starts the iterations I always get an error of "EXCESSIVE GRADIENT  
> IN CI" at
> any point, even when a previous calculation at the same point  
> converged.
> Could anybody help me solving this problem.
> Thank you in advance.
> Ulises Miranda
> --------------------------------------------------------------------
> mail2web.com – Enhanced email for the mobile individual based on  
> Microsoft®
> Exchange - http://link.mail2web.com/Personal/EnhancedEmail
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user

More information about the Molpro-user mailing list