[molpro-user] nocheck in optg

[Alexander Mitrushchenkov]

On Thursday 04 February 2010 02:50:26 pm Natalia Inostroza wrote:
> Hi users.
Natalia,

yes, indeed molpro ignores NOCHECK and many other options during
geometry optimization. This has a very sound base: if your energy
is not converged, then geometry optimization most probably will fail,
as it may contain artificial jumps etc, and in case of analytic gradients
(not implemented for CCSD(T) though) it relies on the fact that the WFs 
are fully convergent. There is no way to override this, and it would be not
a good idea anyway. So, instead you should understand the reasons of
bad convergence looking at your log file. If the reason is the too big jump
in geometry, you should play with optg step parameters, trust ratio etc etc.
If the reason is the  bad convergence of CCSD(T) itself, close to optimal
geometry, this may indicate the multi-reference character and so
CCSD(T) is not adequate. Maybe, tightening the integral thresholds,
change of the basis set, etc, could help.

Best,

Alexander

> I'm working with MOLPRO 2009.1.
> I have a problem with the option nocheck in RCCSD(T).
> 
> Input
> 
> ***,a2
> gthresh,energy=1.d-10,gradient=1.d-3,step=1.d-3
> gprint,basis,orbital,civector
> 
> MEMORY,2300.0,M
> basis=av5z
> bohr;
> geomtrry={....}
> 
> {hf,wf,34,1,0;}
> {RCCSD(T),NOCHECK;}
> OPTG;
> 
>  Energy with triples contribution (RCCSD(T) converges, BUT  when it begins
> to optimize the geometry to give me a failure
> that it's a message:
> 
> ?WARNING:NO CONVERGENCE. TRIPLES WILL DONE ANYWAY SINCE THE NOCHECK OPTION
> IS SET.
> 
> I THINK NOT CONSIDER THE OPTION NOCHECK IN GEOMETRY OPTIMIZATION.  I have
> used different routes, but the problem continues. ANYONE KNOW HOW TO SOLVE
> THIS.
> 
> thank you very much.
> 

-- 
Dr. Alexander Mitrushchenkov, IGR
Laboratoire de Modélisation et Simulation Multi Echelle
UMR 8208 CNRS
Université Paris-Est Marne-la-Vallée
5 Bd Descartes
77454 Marne la Vallée, Cedex 2, France

Phone:    +33(0)160957316
Fax:      +33(0)160957320
e-mail:   Alexander.Mitrushchenkov at univ-mlv.fr
-----------------------------------------