[DFTB-Plus-User] Correct Way to Run MPI-enabled DFTB+ (Conda Installation)

[Bálint Aradi]

Dear Ankur,

For sure, the xTB parameters were created without considering periodic 
structures. But even with a suboptimal parametrization, I'd not expect 
problems with the lattice optimization convergence. This looks to me 
more like some noise, which makes optimization with tight tolerances 
problematic. But currently, I am somewhat puzzled, where this noise 
should come from...

Best regards,

Bálint


On 29.04.22 21:03, Ankur Gupta wrote:
> Dear Dr. Aradi,
> 
> I am already using the new geometry optimizers. I have tried all three 
> optimizers available (Rational, LBFGS, FIRE). It seems like the 
> optimization starts wandering when the max. gradient reaches around 
> 1.00E-03. So, we probably need to decrease the step size when the 
> maximum gradient is around that value. I am currently working with 
> simple periodic organic molecules, with only two molecules in a unit 
> cell, so I expect the optimization to converge within 150 steps.
> 
> I have used GFN2-xTB (in the xTB code) for molecular optimizations of 
> large molecules and found it to be quite robust for a wide variety of 
> cases. However, since the xTB methods have not been parameterized for 
> periodic systems, I was wondering if that's the reason for their poor 
> performance in optimizing periodic systems.
> 
> Best,
> Ankur
> 
> On Fri, Apr 29, 2022 at 2:36 AM Bálint Aradi <aradi at uni-bremen.de 
> <mailto:aradi at uni-bremen.de>> wrote:
> 
>     Dear Ankur,
> 
>      > I have another question related to the usage of DFTB+. I am
>     currently
>      > trying to optimize periodic organic systems (containing around
>     60-110
>      > atoms) using the GFN2-xTB method; however, it seems like the
>     geometry
>      > optimization is extremely slow in terms of gradient convergence. For
>      > example, in around 300 geometry optimization steps, the gradient
>     norm
>      > dropped from 0.72 H/a0 to only 0.0005499 H/a0. Is there a way to
>     speed
>      > up the gradient convergence so that the geometry optimization
>     finishes
>      > in fewer steps?
> 
>     Use the new geometry optimizers (Driver = GeometryOptimization {}).
>     Depending on the system, if the potential surface is very flat, it can
>     be difficult to obtain the minimum.
> 
>     Best regards,
> 
>     Bálint
> 
>     -- 
>     Dr. Bálint Aradi
>     Bremen Center for Computational Materials Science, University of Bremen
>     http://www.bccms.uni-bremen.de/cms/people/b-aradi/
>     <http://www.bccms.uni-bremen.de/cms/people/b-aradi/>
> 
>     _______________________________________________
>     DFTB-Plus-User mailing list
>     DFTB-Plus-User at mailman.zfn.uni-bremen.de
>     <mailto:DFTB-Plus-User at mailman.zfn.uni-bremen.de>
>     https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user
>     <https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user>
> 
> 
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at mailman.zfn.uni-bremen.de
> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user


-- 
Dr. Bálint Aradi
Bremen Center for Computational Materials Science, University of Bremen
http://www.bccms.uni-bremen.de/cms/people/b-aradi/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature
Type: application/pgp-signature
Size: 840 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20220502/1a8188a9/attachment.sig>