Unreasonable Norm in direct CCSD calculation

[Benoit Oswald Stanislas BRAIDA]

>Hello Molpro-Users,
>
>I am attempting a 380bf C1 CCSD/avqz SP.  I don't have enough disk 
>space for a conventional run so I have attempted to use the integral 
>direct code in Molpro2000.1.
>
>The SCF runs as expected but the CCSD stops with:
>
>
> ITER.       NORM       CORR.ENERGY   TOTAL ENERGY   ENERGY CHANGE        DEN1
>    VAR(S)    VAR(P)  DIIS     TIME
>   1      1.21231612    -0.79682143  -227.39947808     0.02337258 -0.02968085
>  0.10D-01  0.56D-02  0  0 20369.79
>   2     22.96434217    -0.82003054  -227.42268720    -0.02320911 
>-1086.37070545
>  0.22D-03  0.22D+02  0  0 40445.27
>
> UNREASONABLE NORM. CALCULATION STOPPED
>
>Does anyone have general suggestions on how this problem can be avoided.
>
>Thanks in anticipation,
>
>Tony Scott
>______________________________________________________________________________


You face this problem when in the calculation the denominator (DEN1) 
become to large,  and it is generaly due to the external pairs. You 
should try the  keyword "shift", which  by shifting the value of the 
denominator for pairs will give you a better damped calculation.
Try for instance the following input :

ccsd(t)
shift,1

If it doesn't work, try to increase the shift value. Anyway, have a 
look to the T1diagnostic at the end of your calculation, because this 
kind of error is often the sign of a deeper problem in the CCSD 
calculation (mutideterminantal case...).



 

Benoit BRAIDA

Laboratoire de chimie physique
Groupe de chimie theorique
Universite Paris-Sud, Batiment 490
91405 ORSAY CEDEX
FRANCE

Tel : 33-1-69-15-61-91
Fax : 33-1-69-15-44-47