[Pw_forum] questions about the scf convergence and accuracy

vega vegalew at hotmail.com
Mon Sep 29 06:24:16 CEST 2008


Dear all,

Now I'm suffering the convergence problems about my system.
My system is the surface of TiO2 and noble metal atoms. First I relaxed surface of TiO2 with a slab geometry.
Most of parameter of the calculations was set to default. The convergence of SCF and BFGS a bit slower but 
they both converged finally. When I add the noble metals on the surface, the convergence of SCF can't converge any more. 
Then I checked the output of the not converged systems. I found the self-consistency error fluctuated, some 
thing like the manual describing, 'A typical sign of such behavior is that the self-consistency error goes down, 
down, down, than all of a sudden up again, and so on'. So I think maybe I should increase the empty bands to 
the system because of the adding of noble metal atoms. Then I add about 20% more empty bands to the system.
I want to know whether the accuracy will be affected by doing this? Do you think adding more empty bands to the 
system will significantly affect the results of relax calculation for TiO2 and noble metals?

And I also found another suggestions to using mixing mode=’local-TF’ for slab systems, and Reduce mixing beta.
Do you think there will be a big difference between the relax calculation using default setting and adjusted settings
to improve the SCF convergence?

thank you for reading.

any hints will be appreciated.

vega


=================================================================================
Vega Lew (weijia liu)
PH.D Candidate in Chemical Engineering
State Key Laboratory of Materials-oriented Chemical Engineering
College of Chemistry and Chemical Engineering
Nanjing University of Technology, 210009, Nanjing, Jiangsu, China
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20080929/592e5aef/attachment.html>


More information about the users mailing list