[Pw_forum] Structural optimization of Biotin molecule

Ari P Seitsonen Ari.P.Seitsonen at iki.fi
Mon Dec 7 16:42:13 CET 2015


Dear Rolly,

   You could try using other pseudo potentials: The HGH ones indeed require 
ecutwfc > 100 Ry, in particular the oxygen.

   I guess that the input_dft = 'LDA' was only for a test, right?

     Greetings,

        apsi

-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   Ari Paavo Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
   Ecole Normale Supérieure (ENS), Département de Chimie, Paris
   Mobile (F) : +33 789 37 24 25    (CH) : +41 79 71 90 935


On Mon, 7 Dec 2015, Rolly Ng wrote:

> 
> Dear QE developers and users,
> 
>  
> 
> I am trying to run a “relax” structural optimization of the Biotin molecule
> on QE-5.2.0, the default nstep=50 does complete but I made a few
> observations on the output file.
> 
>  
> 
> (1) “Check: negative starting charge=   -0.823869” Is this quite large since
> I have only 90 electrons? Is it expected to be 0.001*90=0.009 or less
> according to the FAQ.
> http://www.quantum-espresso.org/faq/self-consistency/#6.12
> 
> (2) It ends by the maximum number of steps reached. Is this normal to end
> BFGS geometry optimization?
> 
> “energy   old            =    -277.9634535886 Ry
> 
>   energy   new            =    -277.9645708192 Ry
> 
>           CASE: energy  _new < energy  _old
> 
>           new trust radius        =       0.1278123491 bohr
> 
>           The maximum number of steps has been reached.
> 
>           End of BFGS Geometry Optimization”
> 
> (3) I have increased the nstep=100 and 1000,  It only takes long to end the
> job but still ended by maximum number of steps reached. Which is very
> strange.
> 
> (4) I have tried to increase  ecutwfc=200.0d0,  ecutrho=800.0d0, then the
> negative starting charge becomes much smaller to -0.000171
> 
> (5) How can I speed up this task?
> 
>  
> 
> Please take a look at the input file as attached. Any comment are welcome.
> 
>  
> 
> Thanks,
> 
> Rolly
> 
> 
>


More information about the users mailing list