SCF energy not identical to result of final energy of geometry optimization

[Carlos Diaz]

I am going to guess this is coming from the different values you used for 
the cutoffs
in energy.in you have 
# Energy cutoff (280,40,4)
@SET cutoff  280
@SET relcut   40
@SET ngrids    4

in opt.in you have 
# Energy cutoff (280,40,4)
@SET cutoff  320
@SET relcut   50
@SET ngrids    4

On Wednesday, May 28, 2014 7:24:24 AM UTC-4, Frank von Horsten wrote:
>
> Dear all,
>
> I am recalculating the SCF energy of a previously optimized geometry. 
> However, the results of the single point calculation differ quite 
> significantly from the final energy of the optimization.
>
> More specifically, the final energy of the geo opt reads 
> E=-3578.782615929515032. This is the final in the line "ENERGY| Total 
> FORCE_EVAL" line. Already here the energy is slightly different from the 
> subsequent "Reevaluating energy at the minimum" part.
> There the SCF cycle gives -3578.782584007462901. Admittedly, that is only 
> 0.084 kJ/mol but still I find this a bit strange.dir
>
> When I use the same input file just changing GEO_OPT to ENERGY_FORCE and 
> taking the final structure of the optimization I get an energy of 
> -3578.778182659758386, which is lower by 11.5 kJ/mol.
>
> Also the calculated forced differ. For the optization the last forces are
> SUM OF ATOMIC FORCES          -0.00006050    -0.00001418    
> -0.00343851     0.00343908
>
> Whereas for the single point calculatin I get
> SUM OF ATOMIC FORCES          -0.00000811    -0.00001851    
> -0.01197202     0.01197204
>
> The system is a 4x4 Cu(100) slab (64 Cu atoms) with a c(2x2) chloride 
> adlayer on top (8 Cl atoms) and eight Na atoms as counter ions in the 
> vacuum region.
>
> Does anyone have an idea where these discrapancies come from?
>
> Cheers,
>
> Frank
>
> PS: I have attached the input and output files in case someone wants to 
> take a closer look.
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140609/1fcbe588/attachment.htm>