[CP2K:6867] Optimize pseudopotential and Basis Set for correlation-exchange functional B97M-V

[hut... at chem.uzh.ch]

Hi

PBE pseudopotentials are a very good first guess and often good enough
for application using GGA functionals. Having a consistent set of
pseudos is certainly better but not always needed.
For first row atoms generating pseudos is rather trivial, beyond the first row
it can be tricky and needs care and experience.

regards

Juerg
--------------------------------------------------------------
Juerg Hutter                         Phone : ++41 44 635 4491
Institut für Chemie C                FAX   : ++41 44 635 6838
Universität Zürich                   E-mail: hut... at chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp... at googlegroups.com wrote: -----To: cp2k <cp... at googlegroups.com>
From: lar... at lbl.gov
Sent by: cp... at googlegroups.com
Date: 08/04/2015 07:28PM
Subject: Re: [CP2K:6867] Optimize pseudopotential and Basis Set for correlation-exchange functional B97M-V

Dear Juerg,
Thanks for the quick reply, this is great. I am fairly new to the DFT world so I have a quick inquiry regarding the use of GTH PPs with mGGAs. All the GTH pseudopotentials available in cp2k have been optimized for some given GGA xc functionals. Do you think that optimizing an existing PP (e.g. GTH-PBE) using the capability of cp2k to do so with the B97M-V xc functional would be worth doing? Or will it be fundamentally wrong to use a GGA-based PP for a mGGA? The energies that I am getting are very reasonable even with the GTH-PBE (non-optimized for B97M-V) so I am not sure what the "rigorous" option is. Thank you in advance!
Best,Luis  

On Tuesday, August 4, 2015 at 5:33:43 AM UTC-7, jgh wrote:Hi



I have implemented a basic interface to LIBXC 3. If you

download the latest version of CP2K from the SVN Trunk you

can compile using the option __LIBXC3 instead of __LIBXC2 (see

arch file) and linking the corresponding version of libxc.



Now you still need to enable the new functionals available.

For this you need to change the file  xc/xc_libxc.F

For XC_MGGA_XC_B97M_V you can un-comment line 2702 in that file.

For other functionals you have to add a corresponding line of

code with the correct name and id number.



Finally, the stress tensor is not implemented for meta-functionals

(and functionals the use the laplacian of the density).



I have attached an input file that worked for me, giving correct

forces in a debug run.



regards



Juerg

--------------------------------------------------------------

Juerg Hutter                         Phone : ++41 44 635 4491

Institut für Chemie C                FAX   : ++41 44 635 6838

Universität Zürich                   E-mail: hut... at chem.uzh.ch

Winterthurerstrasse 190

CH-8057 Zürich, Switzerland

---------------------------------------------------------------



-----cp... at googlegroups.com wrote: -----To: cp2k <cp... at googlegroups.com>

From: lar... at lbl.gov

Sent by: cp... at googlegroups.com

Date: 08/03/2015 11:05PM

Subject: [CP2K:6863] Optimize pseudopotential and Basis Set for correlation-exchange functional B97M-V



Dear cp2k users and developers,

I want to study water using the new meta-GGA xc functional B97M-V (developed by Mardirossian and Head-Gordon), which can be used in cp2k through libxc. I have been doing some tests on the water dimer binding energy using multiple PPs and basis sets on the GPW framework, and it seems that the energies agree fairly well with even all-electron calculations in QChem with same basis sets and xc functional.

While the energies seem reasonable, when I run a finite difference check on forces using cp2k (the debug option), B97M-V fails the test (i.e. the analytical and the numerical calculated forces don't agree) for most pseudopotentials and basis sets. In fact, it only passes the tests when the GTH basis functions (DZTV, TZV2P, QZV3P) are used together with the GTH-PBE pseudopotentials. However, I cannot get any combination of PPs or basis sets and the B97M-V that passes the finite differences check in stresses (I have attached an input file that I used as an example).

I am not sure if optimizing either the basis set or the pseudopotential for B97M-V will help with passing the finite difference check or not (if anybody suggests this route and has any guidelines I would greatly appreciate it as I have never used these capability of cp2k before). Otherwise, I am not sure what other options I have. 

Best,Luis







-- 



You received this message because you are subscribed to the Google Groups "cp2k" group.



To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+... at googlegroups.com.



To post to this group, send email to cp... at googlegroups.com.



Visit this group at http://groups.google.com/group/cp2k.



For more options, visit https://groups.google.com/d/optout.







[attachment "water.inp" removed by Jürg Hutter/at/UZH]

[attachment "water.xyz" removed by Jürg Hutter/at/UZH]





-- 

You received this message because you are subscribed to the Google Groups "cp2k" group.

To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns... at googlegroups.com.

To post to this group, send email to cp... at googlegroups.com.

Visit this group at http://groups.google.com/group/cp2k.

For more options, visit https://groups.google.com/d/optout.