forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

Re: convergence speed

From: myid520@hotmail.com
To: forum@abinit.org
Subject: Re: convergence speed
Date: Sat, 9 Oct 2004 16:23:25 +0200

Dear WANG Yuan Xu :
I think A SCF converged within 30 step is acceptable. Because some
calculations need more 100 steps to converge. But If you want your
convergence speed more rapid,you can use and optimize these variables:
diemix,iprcel,nline and so on. Besides default iscf is 5 which runs fast but
not always reliable. on the contrary, iscf=3 is safe but slow to converge.
you should keep these in mind.

Regards

Zhenhua Zeng
Hunan University, PR. China

convergence speed, WANG Yuan Xu, 10/09/2004
- <Possible follow-up(s)>
- Re: convergence speed, myid520, 10/09/2004
  - error in documentation --- PLS fix this {Re: [abinit-forum] Re: convergence speed]}, Steven Homolya, 10/10/2004
    - Re: error in documentation --- PLS fix this {Re: [abinit-forum] Re: convergence speed]}, Xavier Gonze, 10/14/2004