[Pw_forum] Error:too many bands are not converged

idoldog idoldog at gmail.com
Wed Mar 18 04:52:25 CET 2009


Dear Paolo Giannozzi and other PWSCF users,

I make a mistake by mixing up the parameters mixing_beta and mixing_ndim because i didn't mention mixing_deta while lan haiping did. I take the parameter mixing_ndim for granted. I apologize to you.

However, I have solved the problem now. One can get different results by changing the parameters of nbnd, mixing_beta, K-point, ecutrho and ecutwfc, even the method of diago, these don't eradicate the problem.  By learning the source code, I find this might a problem of another kind of convergence.

All best.


------------------				 
李飞

 

========================

南开大学化学学院

天津市南开区卫津路94号

Email:lifei.chn at mail.nankai.edu.cn

QQ:155192560

MSN:lifei.chn at hotmail.com

=======================
2009-03-18

-------------------------------------------------------------
发件人:Paolo Giannozzi
发送日期:2009-03-18 01:35:59
收件人:PWSCF Forum
抄送:
主题:Re: [Pw_forum] Error:too many bands are not converged

idoldog wrote:

> i can get a successful calculation using a large mixing_beta=16 but for mixing_beta=8 it failed.

really? mixing_beta typically is between 0.1 and 0.7

> i always meet a situation that jobs still run but it didn't output informations 
> and modify the data in out directory (just like endless loops) when i use david 
> diagonalization. that might be caused by poor resources. for cg diagonalization, 
> everything is OK. 

please provide a test job

-- 
Paolo Giannozzi, Democritos and University of Udine, Italy
_______________________________________________
Pw_forum mailing list
Pw_forum at pwscf.org
http://www.democritos.it/mailman/listinfo/pw_forum


More information about the Pw_forum mailing list