[gpaw-users] Error estimations for different XCs

[Jens Jørgen Mortensen]

Den 10-09-2013 01:58, Wellendorff, Jess skrev:
> Dear Wei,
>
> a much more systematic approach to your quest is to use the BEEF-vdW 
> functional with the corresponding Bayesian Error Estimation Ensemble.
> This ensemble actually does exactly what you are looking for without 
> the need for you to specify which "standard functionals" define the 
> spread of DFT results. Instead, the error estimate (or estimate of the 
> spread of predictions by different density functionals) is calculated 
> simply as the square root of the variance over the BEEF-vdW ensemble 
> predictions. Search for "BEEF-vdW" on the GPAW web pages.
>
> I really recommend this approach. It gives you a much less biased and 
> more quantitative estimate of the sensitivity of your DFT result wrt 
> the density functional you have chosen. Think about the inherent bias 
> in this way: How can you be sure that you know by heart the two 
> density functionals that would provide predictions at the two ends of 
> the true distribution of DFT predictions of the quantity you are 
> interested in ??? That is very hard, if not impossible. But the BEE 
> approach offers a convenient and cheap estimate of how much your DFT 
> prediction would change had you chosen other XC functionals than the 
> BEEF-vdW one (provided the quantity of interest is posed in terms of 
> total energies).

Here is a simple example showing how to calculate the energy for another 
functional non-selfconsistentily (look at the PBE.py file):

   https://wiki.fysik.dtu.dk/gpaw/exercises/iron/iron.html

Jens Jørgen

> Best
> Jess Wellendorff
>
> On Sep 9, 2013, at 11:49 PM, Glen Jenness wrote:
>
>> Dear Wei,
>>
>> Off the top of my head you can do something like :
>>
>> calc = GPAW(....)
>>
>> potentials = ['PBE', 'LDA', 'PW91']
>>
>> for xc in potentials:
>>     calc.set(xc=xc)
>>     <calculate binding energy here>
>>
>> This will actually work with most of the ASE calculators.
>>
>> Glen
>>
>>
>>
>>
>> On Mon, Sep 9, 2013 at 5:18 PM, wei guo <guoweicyclone at gmail.com 
>> <mailto:guoweicyclone at gmail.com>> wrote:
>>
>>     Hi guys,
>>
>>     We are interested in calculate the binding energies using
>>     different XC to get a sense how the errors are related to XCs.
>>
>>     To save time, I guess one can start from the relaxed structure
>>     and in one calculation obtain all the potential energies of a few
>>     XCs.
>>
>>     Is there any function or module in GPAW or ASE for doing this
>>
>>     Thanks.
>>     -- 
>>     Best wishes
>>     Wei
>>     ----------------------------------------------------------------
>>     353 Colburn Lab
>>     Center for Catalytic Science and Technology
>>     Department of Chemical Engineering
>>     University of Delaware
>>     Newark, DE 19716-3111
>>     Tel:(302)-831-1132 <tel:%28302%29-831-1132>
>>     ----------------------------------------------------------------
>>
>>     _______________________________________________
>>     gpaw-users mailing list
>>     gpaw-users at listserv.fysik.dtu.dk
>>     <mailto:gpaw-users at listserv.fysik.dtu.dk>
>>     https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>
>>
>>
>>
>> -- 
>> Dr. Glen Jenness
>> Schmidt Group/Morgan Group
>> Department of Chemistry/Materials Science and Engineering (MSAE)
>> University of Wisconsin - Madison
>> _______________________________________________
>> gpaw-users mailing list
>> gpaw-users at listserv.fysik.dtu.dk 
>> <mailto:gpaw-users at listserv.fysik.dtu.dk>
>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
> -----------------------------------------------------------------------
> Dr. Jess Wellendorff
> Postdoctoral Researcher
> SUNCAT Center for Interface Science & Catalysis
> SLAC National Accelerator Laboratory, U.S.A.
> -----------------------------------------------------------------------
>
>
>
> _______________________________________________
> gpaw-users mailing list
> gpaw-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20130910/55fec1ec/attachment-0001.html>