[Pw_forum] rVV10 calculations in espresso 5.1

Paolo Giannozzi paolo.giannozzi at uniud.it
Thu Oct 16 14:31:09 CEST 2014


The correct DFT label for rVV10 is "rVV10", not "vv10". By 
setting "vv10", all you get is the nonlocal part, with no
other XC contributions. The output is misleading:
     Exchange-correlation      = VV10 ( 0  0  0  0 3)
but the four 0's mean respectively "no exchange", "no correlation", 
"no gradient correction to exchange", "no grad. corr. to correlation".
In the svn version there is a further check to prevent this kind of 
misunderstanding, but it was added only recently.

Paolo

On Thu, 2014-10-16 at 12:03 +0100, Andrea Floris wrote:
> Dear quantum-ESPRESSO users,
> 
> 
> I am running some calculations with espresso 5.1 and the van der
> Waals 
> rVV10 functional. 
> 
> 
> First, with the program: 
> 
> 
> generate_rVV10_kernel_table.x
> 
> 
> 
> 
> I generated the universal file:
> 
> 
> rVV10_kernel_table
> 
> 
> 
> 
> 
> 
> 
> Then I tried  three systems up to now,
> 
> 1. two water molecules  (insulating)
> 2. a surface with some organic molecules on top (insulating). This is
> the system I am actually interested in.
> 3. graphite, with  some smearing
> 
> 
> In the first two cases 
> the energy seems to reach convergence but then
> suddenly the scf accuracy increases , like this
> 
> 
> estimated scf accuracy    <       0.06260677 Ry
>      estimated scf accuracy    <       0.04294545 Ry
>      estimated scf accuracy    <       0.02031181 Ry
>      estimated scf accuracy    <       0.01888999 Ry
>      estimated scf accuracy    <       0.01019674 Ry
>      estimated scf accuracy    <       0.00222739 Ry
>      estimated scf accuracy    <       0.00317661 Ry
>      estimated scf accuracy    <       0.00370097 Ry
>      estimated scf accuracy    <       0.00396386 Ry
>      estimated scf accuracy    <       0.00018021 Ry
>      estimated scf accuracy    <       1.54950129 Ry
>      estimated scf accuracy    <       1.54973018 Ry
>      estimated scf accuracy    <       1.54679045 Ry
> 
> 
> 

> ...and so on and does not converge after 100 steps. I tried to reduce
> the mixing to 0.1, or even smaller, without any effect. (see
> attachments)
> 
> 
> For the more complex surface+molecules system, the behaviour is
> similar.
> 
> 
> For the graphite with some smearing, the eigenvectors of the
> Hamiltonian fail to converge in the diagonalization process. (see
> attachment)
> 
> 
> I would appreciate some help to ascertain if: 
> 
> 

> 




> 1.  I am doing something wrong somewhere? 
> 2. there is an issue in the convergence of rVV10 with insulators?
> 3. is there a (different) issue with rVV10 and smearing? 
> 
> 
> I am aware that water calculations with rVV10 have been successfully
> performed, so I would  appreciate very much your suggestions.
> 
> 
> Thanks
> 
> 
> Best,
> Andrea
> 
> 
> -- 
>  Dr Andrea Floris
>  Research Associate
>  King's College London
>  Strand, London WC2R 2LS
>  United Kingdom
>  Phone: +44 (0) 207 848 2064
>  Fax    : +44 (0) 207 848 2420
>  Location:  Strand Building, 4th floor, Room 4.02 
>  Emails: andrea.floris at kcl.ac.uk, an.floris at gmail.com
> 
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
> 
> 

-- 
 Paolo Giannozzi, Dept. Chemistry&Physics&Environment, 
 Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
 Phone +39-0432-558216, fax +39-0432-558222 




More information about the users mailing list